See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/315812760

Dry and lubricated wear of rail steel under

rolling contact fatigue-Wear mechanisms and
crack growth

Article · May 2017

DOI: 10.1016/j.wear.2017.03.025

CITATIONS READS

0 17

3 authors, including:

Santiago Maya Johnson Juan Felipe Santa

Universidade Federal de São Carlos Instituto Tecnológico Metropolitano
8 PUBLICATIONS 7 CITATIONS 21 PUBLICATIONS 153 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Santiago Maya Johnson on 07 April 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Wear 380-381 (2017) 240–250

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Dry and lubricated wear of rail steel under rolling contact fatigue -
Wear mechanisms and crack growth
Santiago Maya-Johnson a,b,n, Juan Felipe Santa a,c, Alejandro Toro a
a
Tribology and Surfaces Group, National University of Colombia, ZIP 050041 Medellín, Colombia
b
Brazilian Nanotechnology National Laboratory (LNNano), ZIP 13083-970 Campinas, Brazil
c
Grupo de Investigación Materiales Avanzados y Energía – MATyER, Instituto Tecnológico Metropolitano, Medellín, Colombia

art ic l e i nf o a b s t r a c t

Article history: In this work, the dry and lubricated wear behavior of rail steels (R370CrHT) under Rolling Contact Fatigue
Received 16 October 2016 (RCF) conditions was studied in laboratory. A twin-disc rolling-sliding machine was used to reproduce
Received in revised form contact conditions under high contact pressure (1.1 GPa) and high creepage (5%) with the aim of ac-
11 February 2017
celerating RCF failure. Two different types of tests were performed, namely dry and lubricated with a
Accepted 28 March 2017
friction modifier, to simulate the mechanisms that cause RCF. The results showed an increase of the wear
Available online 31 March 2017
rate in the tests where a friction modifier was added after an initial stage of 4,000 dry cycles. The damage
Keywords: of the surface proceeded by cracks formation during the dry stage followed by accelerated crack growth
Rolling contact fatigue and flaking due to the effects induced by the friction modifier entering into the original cracks. After
Rail-wheel tribology
several thousand cycles, the flakes were detached from the surface causing high wear rates. The cracks
Friction modifiers
morphology was observed under SEM. The cracks depth and angle with respect to the contact surface
Twin disc testing
Fluid pressure were reported and some correlations were made with the mass loss results.
& 2017 Elsevier B.V. All rights reserved.

1. Introduction A research in Australia found that the total annual maintenance

The wear on rail systems is a serious problem for the compe-
titiveness of transport industry. A global trend to address wear
issues is the use of friction modifiers (FMs) at the interface be-
tween wheel and rail. Bower (1988) [1], Kaneta and Murakami
(1985) [2–4], and others have studied rolling contact fatigue (RCF)
to gain understanding of the behavior of the FM when entering the
cracks and to evaluate its influence on wear and cracks growth
[5–9]. Bower [1] proposed three mechanisms of crack growth as-
sisted by fluids based on the experimental results of S. Way
(1935): (i) the cracks only propagate if a fluid lubricant is applied
to the surfaces in contact, (ii) the cracks always propagate in the
direction of motion of the load over the surface, and (iii) if there is
some relative sliding between the two contacting surfaces, cracks
only propagate in the driven surface. Although it is acknowledged
that a FM can minimize friction between the crack faces and
promote crack growth through the Mode II (shear), it is also well
known these days that no fluid is required to grow a crack; re-
garding (iii), it has been found that the cracks grow in both wheel
and rail.
n
Corresponding author at: Tribology and Surfaces Group, National University of
Colombia, ZIP 050041 Medellín, Colombia.
E-mail addresses: smayaj@gmail.com (S. Maya-Johnson),
jfsanta@gmail.com (J. Felipe Santa), aotoro@unal.edu.co (A. Toro).
cost of a 12 Million Gross Tons (MGT) railway system can be as
high as USD $ 34 per meter of rail. The costs and the wear rates are
influenced by the curvature radius of the railroad, the planning
intervals of rail grinding (Rail grinding refers to a controlled wear
operation used to eliminate cracks before they reach a critical
length at which its growth rate sharply increases) and the use of a
suitable FM at the interface, among others. The total annual cost/
meter of the rail maintenance in a non-lubricated 12 MGT trans-
port systems is USD $ 54, whereas with lubrication savings could
amount around 60% of the total value [10].
A survey released by the Engineering Research Programme in 2003
[11], where the information gathered from 35 rail networks was
analyzed, concluded that effective lubrication can reduce wear on
wheel and rail and the noise levels. The rails life can be increased by a
factor of two, the wheel life by a factor of five and in some cases
depending on the radius of curvature and grinding intervals by a
factor of four. However, today many of the steps and principles that
govern the phenomenon of RCF in the presence of a FM are not en-
tirely known. Several studies carried all over the world have helped
understand some of the principles that govern the phenomenon of
RCF and other problems related to the tribology of wheel/rail contact
such as the type of predominant defects, nondestructive techniques
for inspecting rails, among others [12–18].
In this work, the dry and lubricated wear behavior of rail steels
under RCF conditions was studied in laboratory. A twin-disc

http://dx.doi.org/10.1016/j.wear.2017.03.025
0043-1648/& 2017 Elsevier B.V. All rights reserved.
 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250 241

Table 1
Chemical composition of rail and wheel (wt%).

C Si Mn P S Cr Al V Cu Ti Ni Mo

R370CrHT 0.762 0.394 1.062 0.011 0.015 0.506 0.000 0.002 0.038 0.002 0.058 0.019
ER8 0.542 0.253 0.734 0.011 0.006 0.141 0.027 0.006 0.165 0.002 0.120 0.048

rolling-sliding machine was used to impose controlled contact

Table 2
pressure and creepage conditions that led the tribological pair to a Mechanical properties of the rail.
severe wear regime. The study was done to evaluate the effect of
the application of a FM to the interface when cracks are already Tensile Yield Strength Elongation (%) Hardness HV
formed at the surface of the rail. Strength (MPa) (HB)
(MPa)

R370CrHT 1373 767 10 387 7 17

2. Materials and methods (366.4)

Rail specimens were extracted from R370CrHT rail sections,

manufactured by the company Voestalpine Schienen GMBH-Austria, Fletcher and Beynon in a R260 rail steel [23], the number of dry cycles
and wheel specimens were extracted from a railway wheel grade must be carefully controlled since high tangential forces cause the
ER8. The chemical composition measured by optical emission cracks to grow rapidly and severe flaking may occur after 5000 cycles,
spectroscopy and relevant mechanical properties of the rail and so the nucleated cracks are lost. The number of dry cycles was set
wheel, in line with European standards EN 13674-1:2011 [19] and equal for all samples, since Tyfour et al. [8] demonstrated that this has
EN 13262:2004 [20], are shown in Table 1 and Table 2 respectively. a notable effect on the final fatigue life in dry-wet contact.
The hardness of the wheel was 291 717 HV31.25kgf. Accordingly, in this study, 4,000 dry cycles of pre-fatigue were
The samples were polished for metallographic analysis by using used based on the results of J.F. Santa [24]. Without stopping the
emery papers and (6 to 0.25 μm) diamond paste. After fine pol- test, after 4,000 dry cycles a commercial FM (Sintono Terra HLK,
ishing, the samples were etched with Picral (100 ml ethanol þ 4 g based on synthetic oils) was added to the surface of the wheel
specimen by using a brush. The amount of FM added was weighed
picric acid). A JEOL 5910LV SEM was used to evaluate the micro-
structure and the worn surfaces using Secondary Electrons (SE). to ensure that 0.02 to 0.03 g would be applied to the wheels sur-
For the longitudinal sections and EDS mapping a FEI Inspect F50 face. The application was repeated every 400 cycles. Once the FM
is applied, the coefficient of traction (COT) drops sharply to reach a
was used, located at the Brazilian Nanotechnology National La-
boratory (LNNano). To differentiate the inclusions from the cracks stable value after a few tens of cycles.
Three different numbers of cycles were selected to evaluate the
Backscattered Electrons (BSE) images were taken.
effect of the FM on RCF: 9,000, 14,000, and 24,000. For comparison
The wear tests were carried out in a Twin-Disk testing machine
reasons, dry tests with equal number of total cycles were also
installed at the Tribology and Surfaces Laboratory in the National
performed. For each condition two replicas were prepared, and
University of Colombia, Medellín. Constructive details about the
before each test the samples were washed and weighed to mea-
machine can be found elsewhere [21]. The Twin-Disk machine is
sure mass losses. The tests were performed using separate sets of
used to simulate either dry or wet contact in wheel/rail tribosys-
samples for each number of cycles to be tested. This is because
tems [22]. For each condition, two replicas were performed on
stopping the tests would induce a new running-in period during
different (new) samples to evaluate the repeatability. In this study,
the re-start due to misalignment, changes in real contact area and
the lubricated condition was provided by the addition of a FM to
roughness accommodation. W.R. Tyfour et al. show how the COT
the contact interface. In the test rig, two rolling discs are loaded
never reaches a stable value if the tests are stopped and re-started
upon each other and the relative speed, contact pressure and slip
with the same pair of samples [25].
percentage (creepage) are precisely controlled. Fig. 1 shows a
At the beginning of every test the applied load and contact area
schematic of the testing device.
were verified by carefully watching the contact path in the sam-
Cylindrical specimens were extracted from actual wheels and
ples with the applied load. After turning the motors on, the friction
rails by cutting and lathe turning. The rail specimens were ex-
torque was corrected to zero, the load was applied and the test
tracted from the rails head (Fig. 2a). Both rail and wheel specimens
started. Once the test ended, the samples were washed with de-
have the same dimensions, with a diameter of 47 mm and a rolling
greaser, submerged in acetone and cleaned in ultrasonic bath to
width of 10 mm. The experiments were designed to be aggressive
remove the residues of FM. Finally the samples were weighed to
enough to cause RCF under severe wear regime, so a slip percen-
measure the mass losses and to obtain the wear rates.
tage of 5% and a contact pressure of 1.1 GPa were chosen. The
wheel specimens were rotating at 401 71 RPM, and the rails
specimens at 380 71 RPM. As the samples rotate at different RPM
3. Results and discussion
they are subjected to different fatigue cycles; for the diagrams the
cycles were calculated according to the wheel disc revolutions (i.e.
The nomenclature used hereinafter has the first letter related to
400 RPM). The slip percentage was defined according to Fletcher
the interfacial condition (H for HLK and D for dry), followed by the
and Beynon as [23]:
slip percentage and the number of wet cycles. As an example, a
⎛R N −R N ⎞ test run with the addition of HLK with 1.1 GPa, 5% creepage and
Slip( %) = 200⎜ W W R R
⎟
⎝ RWNW + RRNR ⎠ (1) 10,000 wet cycles, is named “H5% 10 K”.

where N are the revolutions, R is the radius of the disk specimens, 3.1. Effect of the use of a FM on the RCF
and W and R indicate wheel and rail respectively.
In order to accelerate the cracks growth, a dry pre-fatigue period In Fig. 3 the COT reaches a maximum of 0.63 at the beginning of
was implemented before applying the FM. As it was shown by the test and then it decreases to values below 0.60. The running
242 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
Fig. 1. Schematic of the Twin-Disk testing machine.
Fig. 2. a) Extraction of the rails samples for tribological tests, b) cutting scheme for
subsurface analysis.
Fig. 3. Comparison of traction curves for wet and dry tests after 9000 fatigue
cycles.
track appearance indicates that the mechanism by which the rail
disc loses material prior to the steady state stage is oxidative wear,
where the disc track was observed to be covered by a thin, brown
oxide layer [8,26]. When the FM is applied the COT suddenly re-
duces from E 0.52 to E0.10 and then increases again once the FM
is removed from the contact zone. Centrifugal forces during the
test and the squeezing effect caused by contact pressure are re-
sponsible for removing the FM from the contact patch. This be-
havior can be characterized by the retentivity of the FM, which is
defined as the number of cycles required to remove the FM from
the surface and to reach the dry friction coefficient. For most of the
tests, the COT is close to the dry condition after one application, so
for the quantity of FM applied, the retentivity is around 400 cycles.
After five applications (1600 cycles) the COT was very stable.
In order to prevent influencing the COT by the retentivity, all
data used to calculate the COT were averaged starting from 6,600
cycles (Table 3). In both conditions, lubricated with FM and dry, a
low standard deviation was observed in the COT confirming that
the testing machine is reliable. For the tests with FM the average
COT is E0.07 while in dry condition is E 0.49. Field measure-
ments in a commercial railway with similar tests conditions
showed COT between 0.13 and 0.15 [27]. The variation between
laboratory tests and field measurements of friction have been
studied for years, some assumptions have been made but the
reasons are still unknown [28,29]. To ensure that the HLK is able to
guarantee the minimum COT to maintain the operation conditions
in a railway system (0.20), new tests simulating the top of rail/
wheel tread interface are necessary. The results with this FM are in
accordance with those from Gallardo-Hernandez, et al. [30], and
other tests with R260 at 1.5 GPa, dry and using a grease as FM
[31,32].
3.2. Mass losses
Fig. 4a shows the combined (rail þ wheel) mass losses after
tribological tests with 1.1 GPa and 5% creepage. The rails mass loss
after the application of the FM tended to be slightly higher than
that observed during the dry tests due to flaking and cracks
growth assisted by the fluid. During dry tests, the mass loss of the
wheel caused by abrasion was higher so the combined mass losses
during the dry tests become higher than in the case when a FM
was used. Up to 24,000 cycles, the mass loss in dry conditions
caused by abrasion is higher than the mass loss by fatigue and
flaking, even though the HLK has been added to a previously
cracked surface. The difference becomes smaller with increasing
fatigue cycles as more cracks grow and flakes are detached from
the surface. When the cracks length reached during the dry cycles
is greater, the mass losses are higher. S.R. Lewis et al. also reported
Table 3
Coefficients of traction.
Test condition Total fatigue cycles Coefficient of Traction (COT)
HLK 9,000 0.077 7 0.007 0.069 7 0.008
14,000 0.068 7 0.004
24,000 0.061 7 0.001
Dry 9,000 0.498 7 0.008 0.493 7 0.012
14,000 0.489 7 0.024
24,000 0.493 7 0.003
 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250 243

Fig. 4. (a) Wheel and rail mass loss, and (b) wear rate of rail samples for wet and dry tests.

Fig. 5. Images of the cleaned discs after lubricated with FM, and dry tests.
244 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
higher wear rates with lubrication starvation in comparison to dry
tests [32].
Fig. 4b shows the rails mass loss per cycle of fatigue (i.e. the
wear rate). The wear rates were calculated as the total mass loss
divided by the total fatigue cycles (dry and wet). It is interesting
that during dry tests the wear rate remains almost constant
(E 6 mg/cycle), which is in agreement to the results of W.R. Tyfour
et al., who showed that wear rate stabilized after a certain number
of fatigue cycles. For their tests at 1.5 GPa and 1% creepage the
wear rate stabilized after 15,000 rolling cycles [25]. Comparing the
wear rate with the crack angle (Fig. 12b) on the dry tests, between
14,000 and 24,000 cycles it seems to be a stable state with no
change on the crack angle (10 degrees) or the wear rate. Never-
theless, to clearly identify where the wear rate begins to stabilize
for these conditions, more tests with different cycles are needed.
For the tests with FM the wear rate of the rails is only constant
between 9,000 and 14,000 cycles while after 24,000 cycles the
wear rate is almost twice as high. If the rails total mass losses at
14,000 cycles is subtracted from the mass losses at 24,000 cycles
and the result is divided by the difference in cycles (10,000 cycles),
a wear rate almost four times higher than in the early stages is
found (20.185 mg/cycle). The mechanism responsible for the higher
wear rates is delamination of big cracks causing spalling (RCF).
3.3. Wear mechanisms
Fig. 5 shows the worn surfaces of rail and wheel samples after
the tribological tests. For the samples with FM the main wear
mechanism on the rail is RCF and small cracks are observed on the
surface. The cracks (head checks) are very similar to those re-
ported on the literature from field inspections [33,34] and in full-
scale test rig [35]. The cracks can be visible to the naked eye after
10,000 cycles of fatigue with FM. After 20,000 cycles with FM, head
checks were observed and spalling was also present on the surface.
It is worth noticing that no cracks were observed on the wheel
specimen and only a few abrasion marks were found.
In the dry tests, the rail samples appear to have no cracks or
any other marks related to RCF when observed at the naked eye.
Further observations with SEM showed shallow cracks in lesser
Fig. 6. Representation of: (a) the state of stress in the two-discs in rolling-sliding condition, (b) closed crack in wheel and opened crack in rail prior to contact zone, and,
(c) opened crack in wheel and closed crack with entrapped fluid in rail. Characteristic flake and crack orientation are show (not to scale) for each disc. Adapted from Garnham
and Beynon (1992) [38].
amounts, when compared with the samples with FM. The surface
is very smooth and shiny. On the contrary, wheel samples have
surfaces with significant damage due to abrasion. The largest
abrasion marks and the higher mass losses were expected in the
wheel specimen, since it has lower hardness than the rail speci-
men (between 245–275 HB).
The formation of cracks in rail samples during the tests with
FM is caused by several phenomena: (1) the fluid assisted me-
chanism [1,36], (2) the state of stress in the rolling sliding contact
[37,38], (3) the reduction of friction within the cracks faces [1,36]
and (4) squeeze film lubrication theory [39]. When a disc slides on
another disc it appears a tangential force that varies with the slip.
The nature of the behavior of the traction-creepage curve was
described by U. Olofsson and R. Lewis [17,18]. These tangential
forces cause compressive and tensile circumferential stresses in
and around the contact zone, and the stresses are oppositely lo-
cated in the wheel and rail (Fig. 6). Prior to the contact zone the
compressive stresses close the tips of the cracks in the wheel
avoiding the fluids to enter in the cracks. On the other hand, the
tensile stresses open the cracks tips in the rail and facilitate the
entrance of fluids (Fig. 6b); right after the contact zone the stresses
are reversed and the crack in the rail closes pressurizing the fluid
inside (Fig. 6c).
In Fig. 7 and Fig. 8 the images (a, c, e) correspond to a tangential
view of the surface and (b, d, f) to the normal view; the zoom
inside the images is 200X and 1000X respectively. Fig. 7 shows the
evolution of the worn surfaces lubricated with HLK. From 9,000 to
14,000 cycles, it is observed an increase in the number of cracks by
RCF on the rail samples. From 14,000 to 24,000 a detachment of
the deformed material occurs by the coalescence of the cracks
below the surface. This detachment explains the increase in the
mass loss per cycle of the rail samples between 14,000 and 24,000
wet cycles (Fig. 4b).
In the dry tests less cracks were observed, Fig. 8. The cracks on
the dry tests (Fig. 8c and e) appear shallower than in the wet tests
(Fig. 7c and e). This is due to a higher tangential force on the dry
tests producing more deformation nearer the surface. The fact that
there are less cracks on the dry samples than on the wet samples is
due two factors: 1) adding a fluid to a system with pre-existing
 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
Fig. 7. SE images of the worn surfaces of the rail samples after tribological test with 1.1 GPa and 5% creepage. 4,000 dry cycles followed by: a) and b) 5,000, c) and d) 10,000,
and e) and f) 20,000 wet cycles with HLK.
cracks promotes the crack growth via the fluid pressurization
phenomenon, and 2) the shallower cracks on the dry tests are
easily detached, which prevents its subsequent growth, so not
many cracks are observed.
Comparing the samples tested under dry and lubricated with
FM conditions after 9,000 cycles (Fig. 7a and Fig. 8a), the differ-
ences between the tangential views are minor. Thus, applying a
FM after 5,000 fatigue cycles does not have much influence on the
crack growth. Nevertheless, as will be seen on the longitudinal
cuts, the cracks depth and angle remains higher on samples with
FM.
3.4. Crack growth
Fig. 10 and Fig. 11 show an overview of the longitudinal section
of the rail samples, the scan angle was around 35 degrees. The
number of cracks observed on the figures is qualitative, since the
cracks distribution in samples tested with FM is uniform over the
entire circumference. On the other hand, only a few cracks were
observed in the dry samples and the selected scanning area had
the highest density of cracks. Also, the samples with FM had very
245
small cracks that are difficult to detect or are detached during
polishing. Thereby, it is a fact that the samples with FM have more
cracks and are homogeneously distributed.
Deformation of the inclusions in the direction of strain was ob-
served in all samples; the inclusions appear as gray patches and are
easily identifiable on the full resolution images (available online).
Further EDS analysis revealed that the inclusions are mainly sulfides
of the MnS type (Fig. 9). When the material becomes more strained
the inclusions deform and align with the cracks and assist their
growth. This behavior was previously reported with pro-eutectoid
ferrite and ductile inclusions in non-fully pearlitic rail steels [40,41].
Previous studies have shown that the presence of inclusions
and their orientation affect fatigue crack growth [42–44]. Fully
pearlitic steels with fewer inclusions have shown lower fatigue
crack growth rate, even without a strained microstructure [44].
When the microstructure is strained a higher content of ductile
inclusions, deformed and aligned in the direction of the crack tip,
can act as stress concentrators increasing the crack growth.
With more fatigue cycles an increase of the crack depth is ob-
served for both conditions, dry and with FM. On the samples with
FM an increase of the crack angle is also observed, but for the dry
246 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
Fig. 8. SE images of the worn surfaces of the rail samples after tribological test with 1.1 GPa and 5% creepage. a) and b) 9,000, c) and d) 14,000, and e) and f) 24,000 dry
cycles.
samples between 14,000 and 24,000 fatigue cycles seems to sta-
bilize around 10° (Fig 12b).
N. Larijani et al. evaluated the effect of anisotropy in mechanical
properties of the surface layer of pearlitic steels induced by large
deformations, using a numerical model [45]. They showed that for
isotropy or low degrees of anisotropy, the crack grows towards the
surface while for high degree of anisotropy the crack grows
downwards. On the highest levels of anisotropy the crack grows on
the same angle up to 400–600 μm before an abrupt change to a
vertical direction, this coincides with the results observed for dry
tests which have higher anisotropy and they are in early stages of
growth. In all our tests 4,000 dry cycles of pre-fatigue were used,
so all start with the same anisotropy, but the wet samples (with
low anisotropy, low strain) have higher crack angle, which does
not coincide with the results of N. Larijani et al. This shows that not
only the anisotropy is important to predict crack propagation but
also the magnitude of the forces and the pressurization effects
have to be considered. Hardwick et al. measured subsurface de-
formation on R260 at three slip levels (0.5, 5 and 20%), under dry,
water and grease lubricated conditions [31]. They showed that
deformation is higher under dry condition than lubricating with
water, and absent with grease, i.e., there is no anisotropy by
deformation when grease is used.
Fletcher and Beynon carried out tests in R220 with 1% creepage
and 1.5 GPa. They performed three types of tests: unlubricated dry
cycles, 250–500 dry cycles plus lubrication using molybdenum
disulphide suspended in oil, and 250–500 dry cycles plus lu-
brication using water; obtaining crack depths and crack angles of
11 μm and 10°, 300–360 μm and 10–20°, and 340–550 μm and 20–
30° respectively [46]. As in our results, an initial unlubricated
period followed by lubrication produce large cracks with a higher
angle at the crack mouth. Different factors govern the crack
growth such as reduction of cracks face friction, pressurization
effects, contact forces, squeeze film theory and anisotropy of the
properties due to deformation, among others. The water has a
higher COT than others FMs, such as molybdenum disulphide or
HLK, so the tangential forces and the anisotropy will be higher
when water is used as a FM, but less than unlubricated. In that
case it seems that pressurization effects and the anisotropy play a
more important role on the crack growth than the crack face
friction and the contact forces, since cracks grow faster with water
than with others FMs.
Ringsberg made twin disc tests on R260 with 1% creepage and
1.5 GPa in dry condition [47], for 5,000–10,000 cycles the crack
 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250 247

Fig. 9. EDS mapping of a longitudinal section of a rail sample; inset the EDS analysis of the MnS inclusion.

Fig. 10. BSE images of longitudinal section of the rail samples, 4,000 dry cycles followed by: a) 5,000, b) 10,000, and c) 20,000 wet cycles with HLK.
248 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250

Fig. 11. BSE images of longitudinal section of the rail, a) 9,000, b) 14,000, and c) 24,000 dry cycles.

Fig. 12. a) Crack depth, and b) crack angle versus the fatigue cycles.

angle was 13–20°, which is high compared to 5° obtained at 9,000
cycles on R370CrHT. Although the tests conditions were different,
the fact that the cracks are shallower in R370CrHT than those
found by the author, means higher Rolling Contact Fatigue re-
sistance. Moreover, shallow cracks can cause lower mass losses
since the flakes detached from the surface will be smaller. How-
ever, more tests at the same conditions comparing both materials
are needed for further analysis. Eadie et al. carried out tests on a
full scale wheel–rail rig under dry conditions and using a com-
s
mercial FM (KELTRACK HiRail), after 100,000 passes the average
crack angle for the dry condition and applying FM every 500
wheel passes were 39° and 48° respectively [48]. In all cases,
cracks appear to grow faster when a fluid is added to preexisting
cracks.
4. Conclusions
Dry and lubricated tests were carried out to evaluate the per-
formance of a friction modifier (FM) on preexisting cracks and its
effects on the wear rate and the wear mechanisms of an R370CrHT
fully pearlitic rail steel.
 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
 The FM, Sintono Terra HLK, was able to maintain the friction at
the expected levels of twin-disk tests, i.e., COT o 0.1.
 With 5% of creepage, the addition of a FM in presence of pre-
existing cracks increases the mass loss in the rail in comparison
to the dry tests. However, combining the mass loss of the wheel
and the rail up to 24,000 fatigue cycles showed that there is still
greater mass loss in the dry tests.
 The rail wear rate (mass loss per cycle) in dry tests was almost
constant for all fatigue cycles in these experiments, while for
the tests with FM the wear rate increased between 14,000 and
24,000 cycles due to flaking.
 For the evaluated conditions, no cracks on the surface of the
wheel samples were observed.
 In the evaluated conditions, the wheel loses mass faster than
the rail and the magnitude of the adhesion and abrasion wear
mechanisms are significantly higher than the growth of cracks
by fatigue.
 The depth and angle of the cracks increase with the fatigue
cycles for both dry and lubricated conditions. For the dry sam-
ples, between 14,000 and 24,000 fatigue cycles the crack angle
seems to stabilize around 10°.
 The ductile inclusions follow the plastic deformation of the
microstructure, become aligned with the crack tips, acting as
stress concentrators and increasing the crack growth.
Acknowledgements
The authors wish to acknowledge the support of the Medellín
City Hall with his program “Enlazamundos 2012”, co-supported by
Colciencias and the Metro system of Medellín-Colombia; contract
No 111850227421. Also thank to the LME at the Brazilian Nano-
technology National Laboratory (LNNano). Research supported by
LNNano - Brazilian Nanotechnology National Laboratory, CNPEM/
MCTI.
References
[1] A.F. Bower, The influence of crack face friction and trapped fluid on surface
initiated rolling contact fatigue cracks, J. Tribol. 110 (1988) 704–711, http://dx.
doi.org/10.1115/1.3261717.
[2] M. Kaneta, H. Yatsuzuka, Y. Murakami, Mechanism of crack growth in lu-
bricated rolling/sliding contact, Tribol. Trans. 28 (1985) 407–414, http://dx.doi.
org/10.1080/05698198508981637.
[3] Y. Murakami, M. Kaneta, H. Yatsuzuka, Analysis of surface crack propagation in
lubricated rolling contact, Tribol. Trans. 28 (1985) 60–68, http://dx.doi.org/
10.1080/05698198508981595.
[4] M. Kaneta, Y. Murakami, Effects of oil hydraulic pressure on surface crack
growth in rolling/sliding contact, Tribol. Int. 20 (1987) 210–217, http://dx.doi.
org/10.1016/0301-679X(87)90076-4.
[5] V. Mota, L.A. Ferreira, Influence of grease composition on rolling contact wear:
experimental study, Tribol. Int. 42 (2009) 569–574, http://dx.doi.org/10.1016/j.
triboint.2008.04.002.
[6] J. Sundh, U. Olofsson, K. Sundvall, Seizure and wear rate testing of wheel–rail
contacts under lubricated conditions using pin-on-disc methodology, Wear
265 (2008) 1425–1430, http://dx.doi.org/10.1016/j.wear.2008.03.025.
[7] S. Bogdański, P. Lewicki, M. Szymaniak, Experimental and theoretical in-
vestigation of the phenomenon of filling the RCF crack with liquid, Wear 258
(2005) 1280–1287, http://dx.doi.org/10.1016/j.wear.2004.03.038.
[8] W.R. Tyfour, J.H. Beynon, A. Kapoor, Deterioration of rolling contact fatigue life
of pearlitic rail steel due to dry-wet rolling-sliding line contact, Wear 197
(1996) 255–265, http://dx.doi.org/10.1016/0043-1648(96)06978-5.
[9] D.I. Fletcher, J.H. Beynon, The influence of lubricant type on rolling contact
fatigue of pearlitic rail steel, in: M.P.D. Dowson (Ed.), Tribol. Ser., Elsevier,
1999,
pp. 299–310 〈http://www.sciencedirect.com/science/article/pii/
S0167892299800510〉 (accessed 08.10.13).
[10] V. Reddy, G. Chattopadhyay, P.-O. Larsson-Kråik, D.J. Hargreaves, Modelling
and analysis of rail maintenance cost, Int. J. Prod. Econ. 105 (2007) 475–482,
http://dx.doi.org/10.1016/j.ijpe.2006.03.008.
[11] L. Roger, Survey of wheel/rail lubrication practices〈http://www.rssb.co.uk/RE
SEARCH/Lists/DispForm_Custom.aspx?ID ¼ 29〉 (accessed 08.10.13), 2003.
249
[12] D.F. Cannon, K.-O. Edel, S.L. Grassie, K. Sawley, Rail defects: an overview, Fa-
tigue Fract. Eng. Mater. Struct. 26 (2003) 865–886, http://dx.doi.org/10.1046/
j.1460-2695.2003.00693.x.
[13] A. Ekberg, E. Kabo, Fatigue of railway wheels and rails under rolling contact
and thermal loading—an overview, Wear 258 (2005) 1288–1300, http://dx.
doi.org/10.1016/j.wear.2004.03.039.
[14] U. Zerbst, K. Mädler, H. Hintze, Fracture mechanics in railway applications––an
overview, Eng. Fract. Mech. 72 (2005) 163–194, http://dx.doi.org/10.1016/j.
engfracmech.2003.11.010.
[15] INNOTRACK, D4.4.1 – Rail Inspection Technologies〈http://www.innotrack.net/
IMG/pdf/d441.pdf〉, 2008.
[16] K. Sawley, J. Kristan, Development of bainitic rail steels with potential re-
sistance to rolling contact fatigue, Fatigue Fract. Eng. Mater. Struct. 26 (2003)
1019–1029, http://dx.doi.org/10.1046/j.1460-2695.2003.00671.x.
[17] R. Lewis, U. Olofsson, Wheel-rail Interface Handbook, 1st ed, CRC Press;
Woodhead Pub, Boca Raton; Oxford, 2009.
[18] S. Iwnicki, Handbook of Railway Vehicle Dynamics, 1 edition, CRC Press, Boca
Raton, 2006.
[19] European standard, EN 13674-1, - Railway applications - Track - Rail - Part 1:
Vignole railway rails 46 kg/m and above, Httpwwwen-Stand. (n.d.). 〈http://
www.en-standard.eu/csn-en-13674-1-railway-applications-track-rail-part-1-
vignole-railway-rails-46-kg-m-and-above/〉 (accessed 30.08.13), 2011.
[20] European standard, EN 13262:2004 þ A1:2008 - Railway applications -
Wheelsets and bogies - Wheels - Product requirements, Httpwwwen-Stand.
(n.d.). 〈http://www.en-standard.eu/〉 (accessed 04.11.13).
[21] S. Maya-Johnson, Estudio del Comportamiento de Concentradores de Esfuerzo
Bajo Condiciones de Fatiga en Aceros para Rieles, (Master's Thesis), National
University of Colombia, 2014 〈http://www.bdigital.unal.edu.co/12071/〉.
[22] D.I. Fletcher, J.H. Beynon, Development of a machine for closely controlled
rolling contact fatigue and wear testing, J. Test. Eval. 28 (2000) 267, http://dx.
doi.org/10.1520/JTE12104J.
[23] D.I. Fletcher, J.H. Beynon, Equilibrium of crack growth and wear rates during
unlubricated rolling-sliding contact of pearlitic rail steel, Proc. Inst. Mech. Eng.
F J. Rail Rapid Transit. 214 (2000) 93–105, http://dx.doi.org/10.1243/
0954409001531360.
[24] S.M.J. Felipe, Development of a lubrication system for wear and friction con-
trol in wheel/rail interfaces, (Doctoral Thesis), National University of Colombia,
2013.
[25] W.R. Tyfour, J.H. Beynon, A. Kapoor, The steady state wear behaviour of
pearlitic rail steel under dry rolling-sliding contact conditions, Wear 180
(1995) 79–89, http://dx.doi.org/10.1016/0043-1648(94)06533-0.
[26] C. Hardwick, R. Lewis, U. Olofsson, Low adhesion due to oxide formation in the
presence of salt, Proc. Inst. Mech. Eng. F J. Rail Rapid Transit. 228 (2014)
887–897, http://dx.doi.org/10.1177/0954409713495666.
[27] Y.A. Areiza, S.I. Garcés, J.F. Santa, G. Vargas, A. Toro, Field measurement of
coefficient of friction in rails using a hand-pushed tribometer, Tribol. Int. 82
(2015) 274–279, http://dx.doi.org/10.1016/j.triboint.2014.08.009.
[28] H. Harrison, T. McCanney, J. Cotter, Recent developments in coefficient of
friction measurements at the rail/wheel interface, Wear 253 (2002) 114–123,
http://dx.doi.org/10.1016/S0043-1648(02)00090-X.
[29] J. Lundberg, M. Rantatalo, C. Wanhainen, J. Casselgren, Measurements of
friction coefficients between rails lubricated with a friction modifier and the
wheels of an IORE locomotive during real working conditions, Wear 324–325
(2015) 109–117, http://dx.doi.org/10.1016/j.wear.2014.12.002.
[30] E.A. Gallardo-Hernandez, R. Lewis, Twin disc assessment of wheel/rail adhe-
sion, Wear 265 (2008) 1309–1316, http://dx.doi.org/10.1016/j.
wear.2008.03.020.
[31] C. Hardwick, R. Lewis, D.T. Eadie, Wheel and rail wear—Understanding the
effects of water and grease, Wear 314 (2014) 198–204, http://dx.doi.org/
10.1016/j.wear.2013.11.020.
[32] S.R. Lewis, R. Lewis, G. Evans, L.E. Buckley-Johnstone, Assessment of railway
curve lubricant performance using a twin-disc tester, Wear 314 (2014)
205–212, http://dx.doi.org/10.1016/j.wear.2013.11.033.
[33] U. Olofsson, R. Nilsson, Surface cracks and wear of rail: a full-scale test on a
commuter train track, Proc. Inst. Mech. Eng. F J. Rail Rapid Transit. 216 (2002)
249–264, http://dx.doi.org/10.1243/095440902321029208.
[34] U. Zerbst, R. Lundén, K.-O. Edel, R.A. Smith, Introduction to the damage tol-
erance behaviour of railway rails – a review, Eng. Fract. Mech. 76 (2009)
2563–2601, http://dx.doi.org/10.1016/j.engfracmech.2009.09.003.
[35] R. Stock, D.T. Eadie, D. Elvidge, K. Oldknow, Influencing rolling contact fatigue
through top of rail friction modifier application – A full scale wheel–rail test
rig study, Wear 271 (2011) 134–142, http://dx.doi.org/10.1016/j.
wear.2010.10.006.
[36] D.I. Fletcher, P. Hyde, A. Kapoor, Modelling and full-scale trials to investigate
fluid pressurisation of rolling contact fatigue cracks, Wear 265 (2008)
1317–1324, http://dx.doi.org/10.1016/j.wear.2008.02.025.
[37] J.H. Beynon, J.E. Garnham, K.J. Sawley, Rolling contact fatigue of three pearlitic
rail steels, Wear 192 (1996) 94–111, http://dx.doi.org/10.1016/0043-1648(95)
06776-0.
[38] J.E. Garnham, J.H. Beynon, Dry rolling-sliding wear of bainitic and pearlitic
steels, Wear 157 (1992) 81–109, http://dx.doi.org/10.1016/0043-1648(92)
90189-F.
[39] S. Bogdañski, A rolling contact fatigue crack driven by squeeze fluid film, Fa-
tigue Fract. Eng. Mater. Struct. 25 (2002) 1061–1071, http://dx.doi.org/10.1046/
j.1460-2695.2000.00563.x.
[40] J.E. Garnham, C.L. Davis, The role of deformed rail microstructure on rolling
250 S. Maya-Johnson et al. / Wear 380-381 (2017) 240–250
contact fatigue initiation, Wear 265 (2008) 1363–1372, http://dx.doi.org/
10.1016/j.wear.2008.02.042.
[41] J.E. Garnham, C.L. Davis, Very early stage rolling contact fatigue crack growth
in pearlitic rail steels, Wear 271 (2011) 100–112, http://dx.doi.org/10.1016/j.
wear.2010.10.004.
[42] A.D. Wilson, Fatigue crack propagation in A533B steels, J. Press. Vessel Tech-
nol. 99 (1977) 459–469, http://dx.doi.org/10.1115/1.3454560.
[43] S. Ganesh Sundara Raman, Y. Kitsunai, Fatigue crack growth behaviour of
A533B steel in pressurized water at 288 and 28 °C, Eng. Fail. Anal. 11 (2004)
293–303, http://dx.doi.org/10.1016/j.engfailanal.2003.08.009.
[44] S. Maya-Johnson, A.J. Ramirez, A. Toro, Fatigue crack growth rate of two
pearlitic rail steels, Eng. Fract. Mech. 138 (2015) 63–72, http://dx.doi.org/
10.1016/j.engfracmech.2015.03.023.
[45] N. Larijani, J. Brouzoulis, M. Schilke, M. Ekh, The effect of anisotropy on crack
View publication stats
propagation in pearlitic rail steel, Wear 314 (2014) 57–68, http://dx.doi.org/
10.1016/j.wear.2013.11.034.
[46] D.I. Fletcher, J.H. Beynon, The effect of intermittent lubrication on the fatigue
life of pearlitic rail steel in rolling-sliding contact, Proc. Inst. Mech. Eng. F J.
Rail Rapid Transit. 214 (2000) 145–158, http://dx.doi.org/10.1243/
0954409001531270.
[47] J.W. Ringsberg, Life prediction of rolling contact fatigue crack initiation, Int.
J. Fatigue. 23 (2001) 575–586, http://dx.doi.org/10.1016/S0142-1123(01)
00024-X.
[48] D.T. Eadie, D. Elvidge, K. Oldknow, R. Stock, P. Pointner, J. Kalousek, P. Klauser,
The effects of top of rail friction modifier on wear and rolling contact fatigue:
full-scale rail–wheel test rig evaluation, analysis and modelling, Wear 265
(2008) 1222–1230, http://dx.doi.org/10.1016/j.wear.2008.02.029.