Proceedings of the Institution of Mechanical

Engineers, Part J: Journal of Engineering

Tribology
[Link]

The Mechanism of Rolling Contact Fatigue: An Update

A. V. Olver
Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology 2005 219: 313
DOI: 10.1243/135065005X9808

The online version of this article can be found at:

[Link]

Published by:

[Link]

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology
can be found at:

Email Alerts: [Link]

Subscriptions: [Link]

Reprints: [Link]

Permissions: [Link]

Citations: [Link]

>> Version of Record - May 1, 2005

What is This?

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 SPECIAL ISSUE PAPER 313

The mechanism of rolling contact fatigue: an update

A V Olver
Department of Mechanical Engineering, Imperial College London, London SW7 2AZ, UK.
email: [Link]@[Link]

The manuscript was received on 5 July 2004 and was accepted after revision for publication on 26 November 2004.

DOI: 10.1243/135065005X9808

Abstract: A review of some of the recent work on the mechanism of rolling contact fatigue
(RCF) is presented. Topics covered include the appearance and classification of RCF and the
processes of strain localization, texture development, microstructural change, crack formation,
crack shape and propagation, and through-fracture. It is concluded that a significant barrier to
progress is the poor current understanding of the processes of running-in and of the interactions
between plastic deformation, wear, lubricant chemistry and damage accumulation.

Keywords: rolling contact fatigue, pitting, micropitting, fatigue, gears, rolling bearings,
railway track, roughness, steel, bronze

1 INTRODUCTION has been recognized for more than a century but

recent years have seen a revival of interest because
In July 1969, at the symposium in Troy, New York, of the recognition that there was an increasing
organized by NASA, W E Littmann of the Timken problem in ‘high speed passenger, mixed and heavy
Roller Bearing Company presented a paper on ‘The haul’ (i.e. nearly all!) railways [5] and because of a
Mechanism of Contact Fatigue’. Littmann’s number of incidents attributed to the phenomenon
copiously illustrated review [1] was one of the first [6, 7]. In rail contact fatigue, corresponding
to distinguish clearly between six different manifes- categories used for describing RCF are:
tations of rolling contact fatigue:
(a) subsurface initiated ‘shelling’;
(a) subsurface originated ‘spalling’; (b) surface-initiated ‘squats’ or ‘head checks’;
(b) subcase fatigue (in surface hardened com- (c) origins at inner rail corner ‘gauge corner cracking’;
ponents) ‘case crushing’; (d) ‘Flaking’, near surface deformation and wear; and
(c) surface originated ‘pitting’; (e) complete fracture of rail.
(d) origin at local concentrations of pressure (roller Listed in this fashion, both the excellent correspon-
ends, etc.); dence with each of Littmann’s categories (except
(e) ‘Peeling’ (now usually called ‘micropitting’); and for the material-specific ‘case crushing’) and the
(f) section fracture. complete absence of any common terminology is
very obvious. Equally, none of the italicized nomen-
Littmann provided extensive illustrations of the clature above would be familiar in the wider world of
cracking and damage associated with all these structural fatigue, despite the fact that from the ear-
modes and was at least partly responsible for the pre- liest days, contact and structural fatigue were
sent accepted classification of rolling contact fatigue thought to be essentially the same phenomenon.
(RCF) damage. Understandably, his perspective was Rolling contact fatigue is not even mentioned in the
that of the roller bearing, manufactured in hard excellent and highly-regarded book by Ellyin [8], a
materials and of modest size. Companion papers remarkable, but perhaps not unexpected omission.
referred to gears [2, 3]. In these circumstances, we may reasonably ask the
However, the earliest systematic study of RCF was following questions.
by Stewart Way [4] who was at least as interested in
rail – wheel contacts as in gears, although he studied 1. Is rolling contact fatigue in rail – wheel contacts
both [1]. Rolling contact fatigue in railway tracks the same phenomenon as rolling contact fatigue

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
314 A V Olver

in rolling bearings, gears, cams or tappets – and if surface originated RCF in a bearing raceway. Many
so what are the distinguishing features and similar examples may be found in reference [13]. In
mechanisms? the presence of hoop stress, cracking may extend
2. Is rolling contact fatigue the same phenomenon beyond the surface. Figure 2 shows a roller in
as structural fatigue, familiar from the work of, which there is both surface RCF damage (‘pitting’)
for example, Wöhler [9], Smith et al. [10], Forsyth and a radial crack extending to a depth of 3.2 mm,
and Stubbington [11], and Paris and Erdogan [12]? attributed [14] to radial residual stress. Figure 3
3. And if the answer to either of the above is shows a raceway fracture with minimal pitting, also
yes, is there scope for transferring technologies, a consequence of hoop stress, in this case due to
analysis methods, investigative methods and/or excessive interference fit on the shaft. Figures 4 [1]
solutions between the three respective fields? and 5 [15] show subsurface microsections of failed
bearing raceways illustrating crack morphology and
The present review will attempt to address these
microstructural transformation at crack surfaces
questions with particular attention to the underlying
(‘butterflies’).
mechanisms of contact fatigue.
Figures 6 and 7 show RCF damage in rail wheels
and tracks respectively [16, 17]. Here, superficial
2 APPEARANCE AND NOMENCLATURE OF RCF cracking is accompanied by extensive plastic flow.
The RCF in the rail (Fig. 7) is of the surface initiated
Some illustrations are given here of typical manifes- ‘head check’ type.
tations of RCF in engineering components, together In gears, the appearance of rolling fatigue depends
with some of the classifications used in their on the material and the gear type. Figure 8 shows
description. typical RCF (pitting) on a case-hardened steel
Figures 1 to 5 are examples of RCF in rolling helical pinion, again of the surface originated type;
bearings. Figure 1 shows the typical appearance of see also [18]. Figure 9 [13] shows pitting on a

Fig. 1 Rolling contact fatigue of inner raceway of spherical roller bearing (photo by Neale Consulting Engineers, used
with permission)

Fig. 2 Spherical roller showing (left) rolling contact fatigue and (right) radial crack [14] reproduced by permission of
Taylor & Francis, Inc., [Link]

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 315

Fig. 3 Fractured raceway. Radial fatigue crack had origin at the bearing surface [14] reproduced by permission of
Taylor & Francis, Inc., [Link]

bronze worm wheel. Here there is substantial sliding

in a direction perpendicular to the contact motion
leading to appreciable wear accompanying the
cracking. Figure 10 shows the development of RCF
cracks in a carburized steel spiral bevel pinion. The
crack started near the first point of engagement in
the tooth dedendum and propagated to the tooth
crest while pitting was only apparent over a small
part of the tooth surface. The triangular shape of
the pit is typical of surface originated RCF failures
in hard steels. Figure 11 shows an example of ‘case
crushing’ in which the origin is located near the
case – core interface of a carburized gear [19].

Fig. 4 Section of surface originated rolling contact

fatigue failure in bearing inner raceway [1].
Note inclined primary crack system and
formation of ‘pit’ by secondary cracking.
Reproduced with permission of NASA from
[1]; permission by NASA does not constitute
an official endorsement, either expressed or
implied, by the National Aeronautics and Fig. 6 RCF on wheel of railway vehicle [16]. Reprinted
Space Administration from [16] with permission from Elsevier

Fig. 5 Metallurgical transformation accompanying crack formation at subsurface inclusion [15]

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
316 A V Olver

Fig. 7 RCF of railway track (scale length 30 mm approx.) [17]. Reprinted from [17] with permission from Elsevier

Fig. 8 Pitting due to RCF of helical gear tooth

Fig. 10 Pitting of spiral bevel pinion showing (left)

appearance of V-shaped surface pit and
(upper right) section of subsurface crack
system extending to tooth crest

Another example of RCF that has received much

attention in recent years is ‘micropitting’ [20 –26].
This has the appearance shown in Figs 12 and 13
and consists of numerous small pits often
accompanied by extensive cracking. It is commonest
in hard steels that have been ground and in situations
where the lubricant film is thin compared to the
height of the roughness of the surfaces. It is usually
thought to be a form of RCF associated with the
asperity (roughness) stress field and can lead to con-
siderable damage, for example to gear tooth profiles.

3 SPECIAL FEATURES OF ROLLING CONTACT

FATIGUE

3.1 Introduction
Fig. 9 Pitting of bronze worm wheel due to RCF In this section we consider some of the possible
(photo by Neale Consulting Engineers, used reasons that might cause RCF to differ from classical
with permission) structural fatigue and the consequent effects on

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 317

Fig. 11 Case crushing: carburized gear tooth with

subsurface fatigue cracking [19]. Appearance
(above left) and cross-section, showing
location of subsurface cracks. Reproduced
with permission of NASA from [1];
permission by NASA does not constitute an
official endorsement, either expressed or Fig. 12 Helicopter sungear showing worn profile and
implied, by the National Aeronautics and scanning electron micrograph of pitted
Space Administration surface

its behaviour and the means used for making life 3.2 Stress regime
predictions. The principal differences include
the stress regime, the prevalence of plasticity and If the contacting bodies are smooth, loaded below
differences in crack development. the elastic limit and not subjected to any tractive
forces, then the stress field near the contact is just

Fig. 13 Micropitted surface of disc specimen and transverse section at same scale showing
inclined cracking

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
318 A V Olver

Fig. 14 ‘Non-proportional’ evolution of stresses due to

Hertz contact. There is a complete reversal of
shear stress whereas the direct stresses rise
from zero to a single compressive maximum. Fig. 15 Instantaneous distribution of von Mises stress
Line contact, depth: z ¼ a/2, t ¼ time, during the passage of a contact of bodies with
u1 ¼ speed in x-direction over surface rough surfaces. Near the surface, the stresses
are very high because of the roughness. The
that due to the Hertz pressure distribution. This is stresses evolve in a complex manner as the
probably also true with a reasonable degree of contacting surfaces move past each other
accuracy when the contacting components are not (after reference [28])
perfectly smooth but are lubricated by an oil film
whose thickness is large compared to their rough- caused an elevated and complex stress history in
ness. Figure 14 [27] shows the evolution of stresses which there were many cycles during each passage
in such a contact during a load passage for two- of the load (Fig. 15). The exact stress history experi-
dimensional (plane strain) conditions, at a depth enced by a lubricated contact where the roughness
where the range of orthogonal shear stress is at a exceeds the film thickness is still not fully accessible
maximum. It is apparent (a) that the stresses are because of uncertainties concerning (particularly)
local, limited to a depth comparable to the contact lubricant fluid rheology and boundary lubrication.
width; (b) that direct stresses are all compressive; Nevertheless, the broad picture is one in which the
and (c) that the stresses are non-proportional – stresses are local, compressive, non-proportional
that the stress components do not rise and fall with and randomly fluctuating in magnitude and
time in proportion to one another. For example, direction.
the shear stress txz undergoes a complete reversal
while sx and sz exhibit a compressive pulse, wider 3.3 Plasticity and residual stresses
and less intense for sx than for sz. In these respects,
even the simplest RCF experiment differs from The first and perhaps the most widely investigated
classical fatigue tests such as rotating bend or consequence of the peculiarities of the RCF stress
push –pull tests. Indeed it is rare (but not unknown) regime is that plastic deformation often precedes
to find that local compressive stresses give rise to crack formation (which is perhaps inhibited by the
failure in structural fatigue. compressive stresses) on the scale either of the
The situation is even more complex if the rough- Hertz contact width or at least on that of the rough-
ness is appreciable and if there are tractive forces ness. This is particularly so for softer or tougher
or sliding. Kim and Olver [28] showed the elastic materials, such as the soft steels used in railway
stress history when rough surfaces slide together. tracks, and has led to a body of work in which the
The roughness, especially that of the counterface1, development and accumulation of plastic strain has
been studied. In particular, studies have been made
of the elastic and plastic shakedown loads, below
1
All RCF results from the contact of (at least) two surfaces. In this which plastic cyclic and ratchetting strains, respect-
paper, the term ‘counterface’ is used to describe the surface ively, cannot take place [27, 29]. These have some-
opposite to the one being evaluated or considered – for example times even been presumed to be identical to the
where there is RCF of a railway track, the train wheel(s) would be fatigue limit loads [30]. This situation is reminiscent
the counterface. of low cycle structural fatigue (LCF).

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 319

In contrast, in harder materials, RCF is less com- the compressive stresses cause extensive crack clo-
monly accompanied by widespread plasticity and sure. This usually results in the faces of any cracks
crack initiation occurs at stress and strain concen- being not only pressed but also positively rubbed
trations such as surface asperities [23, 28, 31], dents together. This has a number of consequences.
[32] or subsurface inclusions. In the field of rolling First it means that crack face friction and interlock-
bearings, a substantial effort has been expended in ing is an inevitable part of RCF development. This is
the development of empirical models to predict fati- likely in turn to mean that crack development is
gue lives [33, 34] based upon predictions of the influenced mechanically by any lubrication of the
elastic contact stress field. This recalls similar beha- crack face. Secondly, it means that a degree of
viour – and indeed similar approaches to fatigue post-failure damage is also inevitable. Often this is
prediction – in the elastic regime of structural fatigue very severe, resulting in fracture surfaces that are
[8], termed high cycle fatigue (HCF). intensely deformed and damaged by fretting. Orig-
However, even in the hardest steels some plastic inal features relating to the manner of formation of
deformation seems to take place in RCF. Voskamp the crack surfaces are therefore largely obscured
[35] has shown that residual stresses and crystallo- and a valuable technique (fractography) for investi-
graphic texture (preferred orientations of grains) gation of the underlying mechanism is largely
both develop in 1%C – Cr bearing steels at sufficiently impossible. For example, in structural fatigue,
high loads. He noted that the changes were not cycle-by-cycle crack extension results in ‘striation’
accompanied by significant grainflow, suggesting of the crack surface, allowing local crack growth
perhaps that the plastic strains were predominantly rates to be retrospectively determined. Striation
cyclic and that the experiment was above the elastic cannot usually be observed in RCF because of the
shakedown limit but below the ratchetting limit so damage, although some coarser features typical of
that plastic strain did not accumulate [29]. The fatigue, such as ‘beach marks’ (crack arrests) are
mechanism of the process is still the subject of con- often found. A third related effect is that RCF often
jecture and its complete understanding has perhaps gives rise to characteristic debris, such as the spheri-
been inhibited by the very complex microstructure of cal debris particles investigated in the remarkable
hardened bearing steels, which contain at least four experiments by Smith [36], who showed they arise
phases: retained austenite, primary carbide (cemen- from relative tangential displacement of the crack
tite), tempered martensite and coherent (temper) surfaces.
carbides. In addition, there are always non-metallic Finally, RCF changes the topography of the con-
inclusions present with a variety of compositions tacting surfaces. This, in turn affects the develop-
and morphologies. ment of contact pressure and hence the stresses
driving the fatigue mechanism. The effect of even
small changes in topography on the pressure is very
marked indeed and in the presence of developing
3.4 Post-failure damage
surface pitting it is likely that pressures greatly in
A further class of differences between RCF and many excess of the Hertz pressure will be found. This con-
manifestations of fatigue follows from the fact that stitutes a positive feedback mechanism not found in

Table 1 Factors relating to classical and rolling contact fatigue

RCF in hard steels (e.g. rolling RCF in soft steels (e.g. railway
Classical fatigue bearings) track)

Stress regime May be uniaxial, proportional, Always multiaxial, non- Always multiaxial, non-
usually tensile proportional, random history, proportional, random history,
compressive, ‘elastic’, low compressive, ‘plastic’, high m
friction
Origin Surface, associated with PSBs, Surface asperities, subsurface Surface, subsurface cementite –
notches, subsurface inclusions inclusions ferrite boundaries
and pores
Stages Cyclic strain localization, crack Local plasticity, inclined crack Local plasticity, inclined crack
formation, short crack growth, formation, inclined (short) formation, inclined crack
long crack growth, fracture crack growth, pit formation, growth, long crack growth,
fracture fracture
Factors affecting Stress– strain history, ratchetting, Slide–roll ratio, hoop stress, Ratchetting, wear, axial stress,
fatigue life environment (corrosion, roughness, environment environment (corrosion,
hydrogen, water) (lubricant, hydrogen, debris) debris water)

PSB, persistent slipbands.

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
320 A V Olver

structural fatigue and one that is probably central to of possible types of strain localization and that the
the propagation of RCF damage. There has been little precise mechanism varies, even within the
analytical study of this effect although it has been constraints of classical fatigue of simple metals.
recognized from time to time [37, 38].
Table 1 shows some factors relating to classical
4.2 Strain localization in RCF
and rolling contact fatigue and the broad correspon-
dences and differences between them. In RCF, fatigue origins are either at the surface in
regions affected by asperity or debris stresses or sub-
surface at stress raisers such as inclusions. In rolling
4 INITIATION MECHANISMS bearings, developments in secondary steelmaking
(perhaps augmented by moves to lower viscosity
lubricants) have resulted in a preponderance of sur-
4.1 Classical fatigue
face origins [41 –43]. As explained in section 3.2 this
The mechanism of fatigue crack initiation has been leads in the first instance to local cyclic (and perhaps
studied for sufficiently long that many of the relevant accumulating) plastic deformation in the vicinity of
mechanisms are now to be found in text books [7, 39] the surface asperity, debris dent or inclusion. There
and may be considered ‘classical’. A brief summary is a scarcity of evidence as to what exactly happens
of initiation in metals is as follows. next.
In response to the applied stresses (even if these Many authors have commented on the appearance
are well below the static elastic limit) there is local of microstructural changes occurring in the locality
plastic deformation leading to a range of dislocation of crack origins, especially in bearing steels. For
structures in the metal. These may include cell example, Zwierlein et al. and more recently Furu-
structures, veins, walls, ladders and more extensive mura et al. have produced fatigue plots showing
‘persistent slip bands’. None appears to be critical the incidence of transformed phases in a variety of
to the initiation process but all involve some degree bearing steels. These include ‘butterflies’ and dark
of strain localization in which continuing plastic and white etching regions. The present author
strain accumulates in or near a local region that found transformed structure close to the deformed
will eventually become a crack. There are always surface in micropitting [44] and similar transform-
accompanying changes in residual stress and local ations have been found in fretting experiments
mechanical properties and there may be defor- [45]. Again, it appears that these are not necessary
mation twins, phase transformations, void for- precursors to crack initiation and, as bearing steel
mation, surface intrusions and extrusions. is a non-equilibrium structure subjected to high
Eventually, a zone within the intensely strained local density of dissipated energy, it is perhaps not
region – or at the boundary between this and a surprising that degradation occurs.
less-strained region – fractures, leading to an Another remarkable example of microstructural
embryonic crack. The mechanism of this initial frac- change accompanying the pre-crack stage of RCF is
ture can vary from vacancy accumulation – leading provided by recent experiments on bronze [46]. In
to voids and subsequent coalescence (i.e. ductile a CuNiSi bronze progressive cyclic softening of the
fracture) – to cleavage (brittle fracture) but appears material and increasingly rapid strain accumulation
to consist broadly of local static failure following (ratchetting) was found and is shown in Fig. 16.
from earlier redistribution of strains rather than The result was an anomalously low RCF life despite
any single distinct process. the high initial strength of the material. Re-solution
The role of persistent slip bands (PSB) and of of the coherent (Ni2Si) precipitate in a CuNiSi
associated intrusions and extrusions is particularly bronze [shown in Figs 17(a) and (b)], together with
significant. These phenomena were originally the rapid evolution of low-angle grain boundaries
observed in coarse-grained metals (aluminium and and the formation of deformation twins [Fig. 17(c)],
copper) [39]. Indeed, most of the detailed studies all preceded crack initiation and provide a likely
have been carried out on single crystals of pure explanation for the changes in post-yield mechanical
copper [11, 39, 40] and yet they have come to be properties.
regarded as fundamental to the fatigue process. Such studies on non-ferrous materials have been
However, it is clear that neither are essential precon- somewhat uncommon in RCF where steel has been
ditions for fatigue; Ellyin [8] relates that crack the main focus of attention. However, steel,
initiation precedes PSB formation at high tensile especially hard steel, has a very complex microstruc-
mean stresses, and intrusions and extrusions can, ture and is relatively hard to study experimentally.
of course, only occur on a free surface. Neither has Among the significant findings is that of the evol-
been observed in RCF of steels. This again rather ution of a strong texture in the ferritic phase [35].
confirms the view that cracking results from a variety The ferrite becomes orientated in one of two ways,

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 321

Fig. 16 Subsurface RCF damage and cyclic stress –strain curves for two bronze alloys. The
phosphor bronze (left) shows isotropic hardening response to a cycle of tensile and
compressive stress and little damage in the RCF test. The CuNiSi shows kinematic
hardening, leading to ratchetting and extensive microstructural damage together with
a short initiation life

Fig. 17 Microstructural change during RCF of a precipitation-hardened CuNiSi bronze: (a) Bright
field and (b) dark field TEM images of preciptate before RCF loading; (c) bright field
image of subsurface region affected by RCF damage with dislocation structures and
deformation bands. RCF loading leads to re-solution of the Ni2Si precipitate. This has
a significant effect upon the mechanical properties and fatigue life (Fig. 16)

both occurring within the same sample subject to becomes normal to the loaded surface. At the same
repeated rolling. In the first, f100gk110l,2 a face diag- time there are changes in residual stresses, both on
onal of the BCC unit cell is orientated along the roll- the scale of the microstructre and on that of the over-
ing track with the cube face parallel to the surface, all contact stress field and there is also a marked
whereas in the second, f111gk112l, the cube diagonal decline in the proportion of retained austenite. Vos-
kamp [35] suggests that crack orientations may be
2
fhklg is the plane parallel to the surface and khkll is the direction influenced by this texture and that its mechanism
parallel to that of rolling. is likely to include dynamic recrystallization. At the

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
322 A V Olver

highest loads, widespread transformation occurs extend to greater depth because the stresses are
[15, 42] Further study of this complex and specific much lower outside the zone affected by the asperi-
behaviour using advanced electron microscopy and ties but within the surface zone can become very
related techniques seems to be a promising severe.
approach. Another possibility is provided by recent Recent studies have confirmed that micropitting is
advances in microstructural modelling [47, 48]. In indeed extremely sensitive to surface roughness.
principle this can provide a realistic picture of the Benyajati et al., using a disc machine, tested an oil
way in which local features such as dislocations additive (a ZDDP) that suppressed wear [25]. When
influence the development of mechanical properties. the additive was not present, mild wear allowed the
However, the complexity of the structure of steel surface roughness to decline and micropitting did
usually means that many simplifying assumptions not occur. When the anti-wear additive was used,
need to be made. the initial roughness persisted and severe micro-
pitting occurred. Even a short period of running
without the additive was sufficient to allow run-
4.3 Crack formation ning-in (removal of the crests of the roughness) to
occur and for micropitting to be suppressed. In
As mentioned earlier, cracks form initially either at
every case the severity of micropitting correlated
the surface or at subsurface stress or strain concen-
with the level of roughness on the counter-disc. It
trations such as inclusions. This may include trans-
seems likely that the initiation stage of micropitting
formed or otherwise evolved ‘soft’ phases or
is identical to that of other forms of surface-origin
regions, in which greater plastic strain is carried
RCF, the difference being in the subsequent propa-
than in the structure as a whole. Such behaviour is
gation behaviour of the cracks. Surface reinforce-
widely recognized in classical fatigue and has been
ment by means of a shallow very hard layer is an
a feature of more recent studies on RCF [46].
effective remedy [51 – 53] Again, the picture is one
However, it is surface initiation that has the great-
in which the detailed evolution of roughness during
est technological significance and which has
the life of the component is the controlling factor.
attracted the most interest. Here, the driving force
is usually provided by the roughness, particularly
that of the counterface, which leads to rapidly
applied cyclic stresses and strains in the affected sur- 5 PROPAGATION
face. Recent research has led to much better under-
standing of the evolution of contact forces in 5.1 Classical fatigue
bodies with rough or damaged surfaces (see, for
The mechanism of propagation of fatigue cracks is
example the review in reference [49]). Consequently,
one of the main themes in fatigue research in the
it is now clear that roughness is the main cause of
last fifty years. In early studies it was realized that,
crack initiation in RCF of hard steel components. In
especially in LCF, the cracks form at a very early
railway tracks on the other hand, a further highly sig-
stage. (Initially they are very small and cannot be
nificant factor is the tangential loading [29, 30, 50],
seen without the benefit of microscopical or metallo-
which also leads to higher near-surface stresses but
graphic methods.) In consequence, the mechanism
which, of course, is much less important in lubri-
and rate of crack propagation is of great importance
cated contacts because the lubricant controls the
to the overall life. The key finding is that, as the crack
surface tangential traction (friction).
extends, the controlling processes evolve from those
governed by plasticity at pre-existing stress/strain
concentrations to those governed by plasticity at
4.4 Running-in effect: micropitting
the tip of the crack. Knowledge and manipulation
As mentioned in section 2, one particular form of of the laws governing crack behaviour (fracture
surface-initiated rolling fatigue is the phenomenon mechanics) has made a significant contribution in
of micropitting, which has received much attention classical fatigue and it is therefore worth examining
in recent years. what can be learnt about RCF by deployment of the
We distinguish between pitting (or ‘macropitting’ same knowledge.
or spalling) in which the affected zone extends to The most straightforward situation is one in which
roughly the depth of the dimensions of the area of the stresses are elastic except very near the tip of the
contact and micropitting in which the affected crack (linear elastic fracture mechanics, LEFM).
depth is much less. Micropitting is particularly Under these circumstances, the behaviour of the
prevalent in ground steel surfaces and can be attrib- crack is described by the strength of the (elastic)
uted to the stress field associated with the roughness singularity of stress at the tip, known as the stress
of the contacting surfaces. The damage does not intensity factor (SIF, K). In the next section we

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 323

consider some of the attempts in recent years to evolves by ductile fracture and that accumulating
apply LEFM to RCF and the light such attempts plastic displacements control the crack direction
shed on the understanding of propagation of RCF. [38, 56]. In addition we may add the suggestion
of Voskamp [35] that crystallographic texture is
partly responsible.
This situation is rather surprising given the volume
5.2 Initial crack morphology
of research in RCF and perhaps underlines the need
A starting point for any LEFM treatment is the initial for more fundamental approaches to the initiation
crack size and shape. In RCF, initial cracks are nearly mechanism. Nevertheless, irrespective of how it has
always inclined to the surface. In subsurface origi- arisen, the observed crack geometry does provide
nated fatigue, the angle of inclination seems to a starting point for the study of subsequent
depend on the direction of rolling but cracks orig- propagation.
inating at the surface have a clear and widely recog-
nized dependence upon the direction of tangential
(tractive or friction) force. This is illustrated in
5.3 Stable crack growth
Fig. 18, using the example of involute spur gears.
The crack extends initially from a surface origin. If Micrographs showing a sequence of development of
the direction of the tractive force (acting on the sur- RCF from a surface origin are presented in Fig. 19.
face) is from right to left (Fig. 18) then the crack Small initial cracks of about 10 mm in size were
increases in depth toward the right and vice versa, associated with a debris dent after a long life. These
usually at about 208 to the surface. It is well known rapidly developed into a single dominant crack
that this causes a reversal of crack direction at inclined to the surface as shown in Fig. 18 and inter-
the pitch line in gears where the sliding direction secting with it to form a shallow V-shape [Fig. 19(b)].
suddenly changes. There is no single accepted A short period of stable propagation ensued with the
explanation for this dependence. crack gradually extending. This terminated with the
Early attempts at an explanation were based sudden formation of a large pit [Fig. 19(d)]. Similar
upon the elastic stress distribution but these were observations were reported by Littmann and it
not successful because the direction of the applied seems that the sequence is entirely typical of RCF
shear stress is such that it tends to close an embryo- in hard steels. An additional observation clear from
nic crack in the observed direction. Another Littmann’s work and from numerous other sources
explanation is based upon the residual stresses is that the pit does not result from the return of the
[54, 55] arising from plastic (asperity) indentation crack to the surface but rather the crack continues
(which tend to open the crack in the observed orien- to extend (often eventually becoming parallel to the
tation) and still another postulates that the crack surface). The pit results from a secondary crack (or

Fig. 18 Effect of tractive forces and contact motion on direction of surface-initiated cracks in
RCF. Top: microcracks initiate in a direction controlled by the tractive forces from
sliding; in spur gears this results in a reversal at the pitch line and a difference
between driving (left) and driven gears. Below: propagation is favoured when the
motion of the contact is from mouth to tip of the crack, perhaps because of fluid
entrapment (right, after Bower [58])

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
324 A V Olver

Fig. 19 Sequence of micrographs showing appearance of surface-initiated RCF in hardened test

roller. (a) The test was stopped after 127 hours (70  106 contact cycles) when magnetic
perturbation indicated the presence of surface cracks, around 20 mm long, near a debris
dent (dotted outline). The test was restarted and the propagation of these cracks was
followed. (b) 13 minutes later and (c) 1734 minutes later show stable propagation of a
V-shaped crack and a secondary surface-breaking crack (arrow). (d) Shows a large pit
formed 1812 minutes after the initial stoppage

cracks) connecting the original fracture to the sur- On the more positive side, the fluid entrapment
face. It is by no means clear if the secondary crack mechanism (c) does turn out to be consistent with
has a fatigue character; certainly it forms rapidly observations in two important ways:
and Wang and Hadfield [57] have suggested that it
may be brittle, at least in ceramics. (i) it explains why the direction of motion in the
Interestingly, the direction of stable fatigue propa- contact has a decisive effect on propagation of
gation of the inclined crack seems to be controlled by the crack (unless the contact moves toward
the direction of the motion of the contact zone across the crack tip, pressurization cannot occur);
the affected surface rather than by that of the tractive (ii) it explains why the angle of the crack must be
force, as with initiation. This has been the focus shallow (around 208) for appreciable crack
of a number of LEFM studies, notably by Bower, propagation (cracks nearly perpendicular to
Matsuda, Murakami and co-workers [58 –60]. the surface do not generate the necessary
Bower proposed and modelled three possible moving zone of closure).
mechanisms driving crack propagation: It may be surmised that Mode II propagation is
severely curtailed in practice by friction or interlock-
(a) shear (Mode II or III) crack propagation driven ing of the crack faces and that the presence of a fluid
by the contact stress field; assists crack propagation by reducing friction, by
(b) fluid pressurization, in which fluid pressure gen- separating the crack surfaces under compressive
erated by the entrainment of fluid (e.g. lubricant) loading or by creating internal pressure. There is
into the high-pressure contact flowed into the much evidence for this; for example, Way [4] found
crack and caused it to open (Mode I); that RCF pits only occurred when a fluid was present.
(c) fluid entrapment (Fig. 18), in which the closure More recently experiments involving optical inter-
of the mouth sealed fluid into the crack and con- ferometry have shown that the presence of an RCF
tact motion generated fluid pressure at the crack crack influences the local film thickness and pressure
tip (Mode I). in a way broadly consistent with the fluid entrap-
ment conjecture. For example, Fig. 20 [26] shows
All these proposed mechanisms have some unsatis- the presence of a reduced oil film and of fluid
factory features. The main difficulty with (a) is that expulsion into the cavitated region.
it generates a much higher range of SIF for a crack This does not mean to say that the models of fluid
perpendicular to the loaded surface than for the entrapment are at all satisfactory. Deficiencies
observed inclined crack, particularly if any friction include the following.
is allowed between the crack surfaces. It is conse-
quently hard to see why inclined crack propagation . The crack geometries have been very much simpli-
is favoured. On the other hand, mechanisms (b) fied. Either two-dimensional or penny-shaped
and (c) involve a fluid, which in the practically three-dimensional models have been used rather
important case of rail fatigue would only be intermit- than the very specific ‘shallow-V’ geometry
tently present in the form of rainwater. In addition, shown in Figs 19 and 20.
(b) results in a very high value of DK1 (at least if the . The model fluids used have been very rudimen-
full value of the contact pressure acts over tary, usually being inviscid and incompressible,
the whole crack surfaces) and is inconsistent with whereas real lubricants are piezoviscous and
observations of non-propagating cracks. compressible as well as having shear elasticity.

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 325

Fig. 20 Effect of RCF damage on fluid film lubrication. Sequence of interference images showing
region of reduced film thickness and fluid expelled from crack in cavitated region [26]
with permission from Elsevier

5.4 Endurance limits and the threshold of

propagation
Nevertheless, the fracture mechanics approach to
RCF has considerable merit. One of the significant
findings is that the value of DK passes through a
maximum as the crack extends because of the lim-
ited scale of the contact stress field. This means
that the behaviour is similar to that of the fatigue of
notched components [60]. (In fact, the simple RCF
situation is like a sharp notch since the ‘remote
stress’, distant from the area of contact is low!) If
the maximum value of DK is less than the effective
threshold for propagation, DKth, then crack arrest
occurs, leading to a fatigue limit dependent upon Fig. 21 Fatigue limits in RCF calculated from the
the size of the stress concentration (or notch), maximum stress intensity range, using the
which for the Hertz case is simply proportional to fluid entrapment analysis of Matsuda et al.
the semi-width, a. Fatigue limits are plotted in this [60] and measured Mode I thresholds for rail
way in Fig. 21, using the results of analyses from Mat- and bearing steels. The shaded areas show
suda et al. [60], for typical rolling bearing and rail typical operating conditions for the two
steels. It is immediately apparent that the fatigue materials. Despite the simplicity of the
limit for rail steel is relatively high compared to its analysis, the results show the correct order of
elastic limit, a fact that resonates with the large num- fatigue limit for the two applications and
bers of non-propagating cracks found in many rail- explain why rail steels can be used under
way tracks. It also suggests that head-hardened conditions in which the elastic limit is
rails, which have shown a lower value of DKth in lab- significantly exceeded
oratory tests [62] may lead to a lower fatigue limit, in
contrast to the expectations based on small-scale example, hoop stresses in bearing races, residual
disc tests [63] or on the simplistic interpretation of stresses from heat treatment or axial stresses in
the ‘shakedown limit’, which increases with the rails. These stresses have the ability to cause the
yield stress. The importance of both the material RCF crack to ‘turn down’ [64] and cause raceway
ductility and the size of the contact patch in addition (or through-rail) fracture. For example, an analysis
to the classical consideration of stresses and material by the present author and co-workers [17] showed
hardness or strength is evident from this work. how this approach explained the occurrence of
Finally, it is relatively straightforward to extend some roller and raceway fractures and the differ-
these considerations to the practically important ences in behaviour between a small test specimen
problem in which there are not only contact stresses and field experience with larger rolling bearings.
present but there is also a remote stress due to, for Mild wear at the surface, which has the ability to

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
326 A V Olver

shorten cracks as time progresses, has also been suc- and by damage at loads below those that cause
cessfully incorporated into an LEFM model and static failure. It is therefore tempting to conclude
forms an important part of current considerations that the various forms of RCF are essentially similar
of rail life [65]. and that therefore similar strategies and technologies
may be used for combating RCF in different types
of components. This would be a mistake.
6 DISCUSSION An example may suffice to illustrate this. As has
been shown, the normal and, especially, the tangen-
6.1 Introduction tial surface tractions in rail –wheel contacts are fre-
quently high, are difficult to predict (they are a
Rolling contact fatigue is a phenomenon of consider-
function of tractive and braking forces, of rail and
able technological importance, limiting the life of
wheel geometric errors, of train speed and of
many disparate types of engineering components
environment and can only really be calculated
and occurring in a wide variety of materials, includ-
using whole-train dynamic models) and even
ing non-ferrous metals [66], ceramics [67, 68]
harder to measure. They have a significant influence
and even polymers [69] as well as steel. For this
on plastic displacements and thence on both wear
reason, the predominant interest has been in the
and fatigue life. The whole process, from train
development of design protocols for avoiding
characteristics and operating conditions [30, 50]
damage and predicting the life of contacting engin-
through to normal and tangential stresses and strains
eering components. This has led to a number of
(for example, references [29, 71– 73]) and finally (via
well-known and widely used life models, ranging
a fatigue initiation and propagation model
from the entirely empirical to the relatively mechan-
[29, 37, 38, 56, 74, 75] and, perhaps laboratory and
istic. (A review of bearing life modelling is given
field experiments [76, 77]) to accurate, verifiable pre-
in reference [70]). However, the emphasis in the
dictions of rail (or wheel) life [65], is, at every turn,
present paper has been on the underlying mechan-
both complex and very application-specific. One
isms of RCF and on the various approaches to its
might perhaps expect the fatigue model to be more
elucidation.
general than the other parts of the system but even
However, advances in experimental techniques
here the materials (soft, corrodible steel) and the
such as XRD, electron and atomic force microscopy
conditions (frequently exceeds elastic limit, limited
and optical interferometry have now been widely
or absent lubrication, large contact) differ substan-
exploited in investigations of RCF and these, together
tially from most other applications. There are, of
with an unprecedented effort of modelling, of lubri-
course, some common aspects with rolling bearings
cation, of continuum solid mechanics and of crack-
(notably non-propagating cracks, ‘turning down’
ing and fracture have contributed to an emerging
and leading to through-fracture [14, 65]) but there
picture of RCF that is rather different from the classi-
is little to be found in the way of directly transferable
cal view. Rather than a distinct and specific phenom-
technology between the applications.
enon, the preserve of a few specialist applications,
RCF is now increasingly seen as a special case of
the wider phenomenon of fatigue of materials. How-
ever, as the boundaries between RCF and other 6.3 What remains to be done?
forms of fatigue become less distinct it is probably
A number of topics that require further research are
true that the differences between individual types
suggested by the limitations of current knowledge
of RCF are becoming more widely recognized. In
identified in the present review. They include the
the remainder of this discussion, some of these
micromechanics of crack initiation (sections 4.2
connections and distinctions will be explored and
and 4.3), research on crack shape, contact size,
a personal view of some possible future directions
propagation mechanisms and the role of fluid in
for research into the mechanism of RCF will be
the crack (sections 4.3, 5.2 and 5.3) and miscella-
presented.
neous factors affecting the fatigue limit such as
materials and treatments (sections 4.4 and 5.4).
6.2 Relationship between RCF and classical However, if a single factor in the mechanism of
RCF had to be identified that most exemplified the
fatigue
limitations of current knowledge it would certainly
There seems little doubt that both different types of be the phase of the process known colloquially as
RCF and, indeed, different types of fatigue in general ‘running-in’. This may be defined as changes
are sufficiently similar to be regarded as a single (usually beneficial changes) in the shape (profile,
phenomenon. All are, for example, characterized by roughness) and state (strength, residual stress, film
cyclic strains, by crack formation and extension formation) of the contacting surfaces occurring

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 327

during an initial period before damage is apparent. many practically important manifestations of RCF
In general, running-in is a consequence of plastic have application-specific causes and remedies. This
deformation (especially, perhaps, on the softer of is perhaps particularly so in the RCF of railway track.
two contacting surfaces) and/or wear. Recent findings tend to emphasize that a signi-
Small changes in the shape of contacting surfaces ficant barrier to the development of robust and
have a very large effect on the pressure generated verifiable RCF life models (and perhaps to Tribology
in subsequent passes. For example, in an elastic con- in general!) is the poor current understanding of the
tact the pressure depends on the slope of the surface processes of running-in and of the interactions
profile, a one degree slope corresponding to around between plastic deformation, wear, lubricant chem-
1 GPa for steel-on-steel. This means that on the istry and damage accumulation. This arises because
scale that surface cracks first appear (a few the stresses driving RCF are very strongly dependent
micrometers) the local pressure may vary between upon the shape (profile, roughness) of the contacting
1 GPa and zero with the removal of as little as surfaces and has no parallel in classical fatigue.
50 nm of material. This provides an extremely
potent feedback mechanism, which has no counter-
part in classical fatigue.
ACKNOWLEDGEMENTS
The feedback is not always beneficial, since profile
changes under one set of operating conditions can
The author is particularly grateful to his colleagues
influence others. For example, changes due to wear
and students at Imperial College, London, especially
of wheel profile on railway vehicles encountering
Dr S Medina, Mr C Benyajati, Dr E P Busso, Prof H A
track defects, can be a strong factor driving RCF in
Spikes, Prof E Ioannides and Prof R A Smith, both for
rails at other locations [30]. Such factors undermine
a number of useful discussions and for responding
one of the assumptions underlying simple shake-
helpfully to his many requests for help and
down theory, which relies on steady operating
information.
conditions.
Mostly, however, running-in is beneficial, whether
due to wear or to plasticity. Both processes are
strongly influenced by lubricants and other surface
REFERENCES
films and contaminants. Suppressing wear by the
use of lubricant (anti-wear) additives has been
1 Littmann, W. E. The mechanism of contact fatigue. In
shown to lead to micropitting [25]. The run-in topo- Interdisciplinary approach to the lubrication of concen-
graphy determines all the subsequent (fatigue) beha- trated contacts, (Ed. P. M. Ku) Proceedings of a Sym-
viour, differs greatly from the ‘as-manufactured’ posium, Troy, NY, July 1969, NASA Special Report,
topography, and is strongly dependent on the lubri- SP-237, 1970, pp. 309 – 378 (NASA, Washington DC).
cant chemistry, rather than merely its mechanical 2 Bamberger, E. N. Effect of materials – metallurgy view-
properties. Equally, wear can mitigate fatigue point. In Interdisciplinary approach to the lubrication of
through its influence on crack length [65]. It would concentrated contacts, (Ed. P. M. Ku) Proceedings of a
appear to be necessary to quantify these changes Symposium, Troy, NY, July 1969, NASA Special
with some degree of confidence if reliable prediction Report, SP-237, 1970, pp. 409 – 438 (NASA, Washington
DC).
of fatigue lives is to be achieved This is not at present
3 Coleman, W. Gear design considerations. In Interdisci-
possible and represents a significant challenge to plinary approach to the lubrication of concentrated
Tribology in the next 35 years. contacts, (Ed. P. M. Ku) Proceedings of a Symposium,
Troy, NY, July 1969, NASA Special Report, SP-237,
1970, pp. 551 – 589 (NASA, Washington DC).
7 CONCLUSIONS 4 Way, S. Pitting due to rolling contact. J. Appl. Mech.,
1935, 57, A49 –A58.
A review of some of the developments in understand- 5 Cannon, D. F. and Pradier, H. Rail rolling contact fati-
ing of the mechanism of rolling contact fatigue in the gue research by the European Rail Research Institute.
past 35 years has been presented. Rolling contact Wear, 1996, 191, 1 –13.
fatigue in hard and soft steels have been compared 6 Smith, R. A. Rolling contact fatigue of rails: What
remains to be done? In 5th World Congress on Railway
and contrasted with classical high and low cycle
Research (WCRR), 25 – 29 November 2001, available at
fatigue.
[Link] accessed 17 October 2003.
The processes of strain localization, texture devel- 7 Clayton, P. and Su, X. Surface initiated fatigue of baini-
opment, crack formation, crack shape and propa- tic steels under water-lubricated rolling-sliding contact.
gation, and through-fracture have been examined. Wear, 1996, 200, 63 –73.
It is concluded that despite the value in recogniz- 8 Ellyin, F. Fatigue damage, crack growth and life
ing the underlying unity of the fatigue phenomenon, prediction, 1997 (Chapman and Hall, London).

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
328 A V Olver

9 Wöhler, A. Versuche über die Festigkeit der Eisenbahn- 29 Johnson, K. L. and Bower, A. F. Plastic flow and shake-
wagenachsen. Zeitschrift fur Bauwesen, 1860, 10; down of the rail surface in repeated wheel – rail contact.
English summary, Engineering, 1867, 4, 160– 161. Wear, 1991, 144, 1– 18.
10 Smith, K. N., Watson, P., and Topper, T. H. A stress – 30 Evans, J. Great Western Study, A detailed review of two
strain function for the fatigue of metals. J. Mater., sites. Proceedings of Seminar on Vehicle Track Inter-
1970, 5, 767 – 778. action, Identifying and Implementing Solutions, Paper
11 Forsyth, P. E. and Stubbington, A. The influence of No. 2, Instn. Mech. Engrs, London, 2004.
substructure on the slip observed in pure aluminium 31 Nakajima, A. Effect of asperity interacting frequency on
and some aluminium alloys when subjected to fatigue surface durability. International Symposium on Gear-
stresses. J. Inst. Met., 1955, 84, 173 – 175. ing and Power Transmissions, Tokyo, 1981, paper b26.
12 Paris, P. C. and Erdogan, F. A critical analysis of crack 32 Nélias, D. and Ville, F. Detrimental effects of debris
propagation laws. Trans Am. Soc. Mech. Engrs, J. Basic dents on rolling contact fatigue. Trans Am. Soc. Mech.
Engrg, 1963, 85, 528 – 534. Engrs, J. Tribol., 2000, 122, 55 – 64.
13 Neale, M. J. Neale Consulting Engineers Ltd, Fleet, UK. 33 Lundberg, G. and Palmgren, A. Dynamic capacity of
Available at [Link], accessed June 2004. roller bearings. Mechanical Engineering Series, Royal
14 Kim, T. H., Olver, A. V., and Pearson, P. K. Fatigue and Swedish Academy of Engineering Sciences. Acta Poly-
fracture mechanisms in large rolling element bearings. technica, 1947, 1(3), 7.
Tribol. Trans., 2001, 44, 583 – 590. 34 Ioannides, E. and Harris, T. A. A new fatigue life model
15 Zwierlein, O. and Schlicht, H. Rolling contact fatigue for rolling bearings. Trans Am. Soc. Mech. Engrs,
mechanisms, accelerated testing and field perform- J. Tribology, 1985, 107, 367 –378.
ance. In Rolling contact fatigue testing of bearing steels 35 Voskamp, A. Material response to rolling contact load-
(Ed. J. J. C. Hoo), ASTM Special Publication STP 771, ing. Trans Am. Soc. Mech. Engrs, J. Tribology, 1985, 107,
1982, pp. 358 – 391 (ASTM, Philadelphia, PA). 356 – 366.
16 Ekberg, A. and Sotkovszki, P. Anisotropy and rolling 36 Smith, R. A. Interfaces of wear and fatigue. In Proceed-
contact fatigue of railway wheels. Int. J. Fatigue, 2001, ings of Conference on Fundamentals of Tribology (Eds
23, 29 – 43. N. P. Suh and N. Saka) 1980, pp. 605 – 616 (MIT Press,
17 Kalousek, J. Keynote address: light to heavy, snail to Cambridge, MA).
rocket. Wear, 2002, 253, 1 – 8. 37 Ringsberg, J. W., Bjarnhed, H., Johansson, A., and
18 Errichello, R. How to analyze gear failures. Practical Josefson, B. L. Rolling contact fatigue of rails, finite
Failure Analysis, 2002, 2, 8 – 16. element modelling of residual stresses, strains and
19 Pederson, R. and Rice, S. L. Case crushing in carbur- crack initiation. Proc. Intstn Mech. Engrs, Part F:
ized and hardened gears. SAE Trans, 1961, 69, 370 – 380. J. Railway Engineering, 2000, 214, 7 – 18.
20 Berthe, D., Flamand, L., Foucher, D., and Godet, M. 38 Franklin, F. J., Widiyarta, I., and Kapoor, A. Computer
Micro-pitting in Herzian contacts. Trans Am. Soc. simulation of wear and rolling contact fatigue. Wear,
Mech. Engrs, J. Lubrication Tech., 1980, 102, 478. 2001, 251, 949 – 955.
21 Shotter, B. A. Micro-pitting, its characteristics and 39 Suresh, S. Fatigue of materials, 1998, ISBN 0 521 57046
implications. In Performance and testing of gear oils 8 (Cambridge University Press, Cambridge, UK).
and transmission fluids, 1981 (Institute of Petroleum, 40 Zhang, Z. F., Wang, Z. G., and Sun, Z. M. Evolution and
London). microstructural characteristics of deformation bands in
22 Zhou, R. S., Cheng, H. S., and Mura, T. Micropitting in fatigued copper single crystals. Acta Materiala, 2001,
rolling and sliding contact under mixed lubrication. 49, 2875 – 2886.
Trans Am. Soc. Mech. Engrs, J. Tribology, 1989, 111, 41 Tallian, T. E. Failure atlas for hertz contact machine
605 – 613. elements, 1992 (ASME Press, New York).
23 Olver, A. V., Spikes, H. A., and Macpherson, P. B. Wear 42 Zaretsky, E. V. (ed) Life factors for rolling bearings,
in rolling contacts. Wear, 1986, 112, 121 –144. second edition, 1999, STLE Publication SP-34 (Society
24 Webster, M. N. and Norbart, C. C. J. An experimental of Tribologists and Lubrication Engineers, Park Ridge,
investigation of micropitting using a roller disk IL).
machine. Tribol. Trans, 1995, 38, 883 – 895. 43 Fumura, K., Murakami, Y., and Abe, T. Development of
25 Benyajati, C., Hamer, C. J., and Olver, A. V. An exper- long life bearing steel for full film and for poor and
imental study of micropitting using a new miniature contaminated lubrication motion and control, 1, 1996,
test rig. In Proceedings of 30th Leeds – Lyon symposium pp. 30 –37 (NSK, Japan).
on Tribology, Lyon, September, 2003 to be published, 44 Olver, A. V., Spikes, H. A., and Macpherson, P. B. Wear
Elsevier, 2004. in rolling contacts. Wear, 1986, 112, 121 –144.
26 Olver, A. V., Tiew, L. K., Medina, S., and Choo, J. W. 45 Sauger, E., Fouvry, S., Ponsonnet, L., Kapsa, P.,
Direct observations of a micropit in an elastohydro- Martin, J. M., and Vincent, L. Tribologically
dynamic contact. Wear, 2004, 256, 168– 175. transformed structure in fretting. Wear, 2000, 245,
27 Kapoor, A., Morales Espejel, G. E., and Olver, A. V. A 39 – 52.
shakedown analysis of simple spur gears. Tribol. 46 Medina, S., Olver, A. V., and Shollock, B. The develop-
Trans, 2002, 45, 103– 109. ment of rolling contact damage in two bronze alloys. In
28 Kim, T. H. and Olver, A. V. Stress history in rolling slid- Proceedings of 30th Leeds –Lyon Symposium on Tribo-
ing contact of rough surfaces. Tribol. Int., 1999, 31(12), logy, Lyon, September 2003, to be published, Elsevier
727 – 736. 2004.

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
 Mechanism of rolling contact fatigue 329

47 Deshpande, V. S., Needleman, A., and Van der 63 Benyon, J. H., Garnham, J. E., and Sawley, K. J. Rolling
Giessen, E. Scaling of discrete dislocation predictions contact fatigue of three pearlitic rail steels. Wear, 1996,
for near-threshold fatigue crack growth. Acta Mat., 192, 94 – 111.
2003, 51, 4637 – 4651; see also Acta Mat., 2002, 50, 64 Health and Safety Executive (UK) Hatfield Derailment
831 – 846. Investigation, Interim Recommendations of the Investi-
48 Busso, E. P., Lei, Y., O’Dowd, N. P., and Webster, G. gation Board, August 2002, available at [Link].
Mechanistic prediction of fracture processes in ferritic uk/[Link]fi[Link], accessed 18 February 2004.
steel, welds within the transition temperature regime. 65 Kapoor, A., Schmid, F., and Fletcher, D. Managing the
J. Engrg Mater. Technol., 1998, 120, 328 – 337. critical wheel – rail interface. Railway Gazette, January
49 Spikes, H. A. and Olver, A. V. Basics of mixed lubrica- 2002, 25 –28
tion. Lubrication Sci., 2003, 16(1), 3 – 28. 66 Ishibashi, A. and Hoyashita, S. Appreciable increase in
50 Dembosky, M. and Stow, J. C2C – The impact of a new pitting limit of steel and bronze. Tribol. Int. 1982, 15,
fleet on a commuter railway. Proceedings of Seminar on 349 – 356.
Vehicle Track Interaction, Identifying and Implement- 67 Henshall, J. L., Guillou, M. O., and Hooper, R. M. Point
ing Solutions, Paper No. 2, 2004 (Instn. Mech. Engrs, contact surface fatigue in zirconia ceramics. Tribology
London). Int., 1995, 28, 363 – 376.
51 Wilkinson, C. M. R. and Olver, A. V. The durability of 68 Hadfield, M., Stolarski, T. A., Cundhill, R. T., and
gear and disc specimens, part 1, the effect of some Horton, S. Failure modes of pre-cracked ceramic
novel materials and surface treatments. Tribol. Trans, elements under rolling contact. Wear, 1993, 169, 69 – 75.
1999, 42(3), 503 – 510. 69 Al-Sabti, S. L. and Stolarski, T. A. Surface fatigue of
52 Wilkinson, C. M. R. and Olver, A. V. The durability of brittle polymers in rolling line contact. Tribol. Int.,
gear and disc specimens, part 2, post failure examin- 1998, 31, 695 – 699.
ation and gear-disc correlation. Tribol. Trans, 1999, 70 Ioannides, E., Bergling, G., and Gabelli, A. An analyti-
42(3), 610– 618. cal formulation for the life of rolling bearings, Acta
53 Olver, A.V., Cole, S. J., and Sayles, R. S. Contact stres- Polytechnica Scandinavica, Mechanical Engineering
ses in nitrided steels. Proceedings of the 19th Leeds – Series, No. 137, ISBN 951-666-503-9, 1999 (Finnish
Lyon Symposium on Tribology, Thin Films in Tribology, Academy of Technology, Espoo).
Leeds, September 1992, Elsevier, 1993. 71 Jiang, Y. and Sehitoglu, H. A model for rolling contact
54 Olver, A. V. Micropitting and asperity deformation. In failure. Wear, 1999, 224, 38 – 49.
Developments in numerical and experimental methods 72 McDowell, D. L. An approximate algorithm for elastic –
applied to tribology (Eds D. Dowson, C. M. Taylor, M. plastic two dimensional rolling/sliding contact. Wear,
Godet and D. Berthe), 1984 (Butterworths, London). 1997, 211, 237 – 246.
Presented at the 10th Leeds –Lyon Symposium on 73 Jiang, Y. and Sehitoglu, H. Rolling contact analysis
Tribology, Lyon, France, September 1983. with the application of a new plasticity model. Wear,
55 Olver, A. V., Spikes, H. A., Bower, A. F., and Johnson, 1996, 191, 35 – 44.
K. L. The residual stress distribution in a plastically 74 Bogdański, S. and Brown, M. W. Modelling the three-
deformed model asperity. Wear, 1986, 107, 151 –174. dimensional behaviour of shallow rolling contact
56 Ringsberg, J. W., Loo-Morey, M., Josefson, B. L., fatigue cracks in rails. Wear, 2002, 253, 17 – 25.
Kapoor, A., and Benyon, J. H. Prediction of crack 75 Dubourg, M. C. and Lamacq, V. A predictive rolling
initiation for rolling contact fatigue. Int. J. Fatigue, contact fatigue crack growth model: onset of branch-
2000, 22, 205 – 215. ing, direction and growth – role of dry and lubricated
57 Wang, Y. and Hadfield, M. A mechanism for nucleating conditions on crack patterns. Trans Am. Soc. Mech.
secondary fractures near a pre-existing flaw subject to Engrs, J. Tribol., 2002, 124, 680 – 688.
contact loading. Wear, 2003, 254, 597– 605. 76 Grohmann, H. D., Hempelmann, K., and Groß-
58 Bower, A. F. The influence of crack face friction and Thebing, A. A new type of RCF: experimental investi-
trapped fluid on surface initiated rolling contact fatigue gations and theoretical modelling. Wear, 2002, 253, 67–74.
cracks. Trans Am. Soc. Mech. Engrs, J. Trib, 1988, 110, 77 Gao, N., Dwyer Joyce, R. S., and Grieve, D. G. Disc
704 – 711. machine testing to assess the life of surface-damaged
59 Murakami, Y., Kaneta, M., and Yatsuzuka, H. Analysis railway track. Proc. Instn Mech. Engrs, Part F:
of surface crack propagation in lubricated rolling J. Railway Engineering, 2001, 215, 261 – 275.
contact. ASLE Trans, 1985, 28, 209 – 216.
60 Matsuda, K., Olver, A. V., and Ioannides, E. Propa-
gation of rolling contact fatigue cracks by the fluid
entrapment mechanism. Presentation to the World APPENDIX
Tribology Congress, London, 1997 (abstract only
published). Notation
61 Smith, R. A. and Miller, K. J. Prediction of fatigue
regimes in notched components. J. Mech. Engrg Sci., a Hertz semi-width
1978, 20, 201 – 206. c length of fatigue crack
62 Hellier, A. K. and Merati, A. A. The mode I fatigue K stress intensity factor
threshold for head-hardened rail steel. Int. J. Fatigue, N number of cycles
1998, 20, 247 – 249. p0 Hertz (maximum) pressure

J04104 # IMechE 2005 Proc. IMechE Vol. 219 Part J: J. Engineering Tribology

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014
330 A V Olver

t time DKth threshold range of stress intensity factor

s direct stress
v1 speed of damaged surface with respect to t shear stress
the contact

Proc. IMechE Vol. 219 Part J: J. Engineering Tribology J04104 # IMechE 2005

Downloaded from [Link] at NATIONAL CHUNG HSING UNIV on March 25, 2014