Wear 267 (2009) 20832091

Contents lists available at ScienceDirect

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

Particle motion and modes of wear in the dry sandrubber wheel abrasion test
S.M. Nahvi, P.H. Shipway , D.G. McCartney
Division of Materials, Mechanics and Structures, Faculty of Engineering, The University of Nottingham, University Park, Nottingham NG7 2RD, UK

a r t i c l e

i n f o

Article history:
Received 29 September 2008
Received in revised form 16 June 2009
Accepted 4 August 2009
Available online 13 August 2009
Keywords:
Abrasion
Wear
DSRW
G65
G105
Steels

a b s t r a c t
In the dry sandrubber wheel test, the particles are free to move between the wheel and testpiece. The
particles may either roll as they pass through the contact, or they may temporarily embed in the rubber
wheel and groove the sample; the motion of the particles will govern the modes of deformation of the
sample and thus the dominant mechanisms of wear. In experiments concerning the abrasion of a range
of steels with widely varying hardnesses, it has been shown that the motion of the particles through the
contact depends not only upon the details of the testing conditions (for example particle feedrate, particle
size and shape, applied load) but also upon the testpiece material properties themselves, such as hardness.
Such a dependence upon testpiece material properties is a cause for concern for those who use the test,
and indicates that observations of the mechanisms of wear are an essential part of this test methodology.
Particle rolling through the contact is favoured by low applied loads and low testpiece hardness whereas
particle sliding through the contact is favoured by high applied loads and high testpiece hardness. The
wear coefcients in situations where particle sliding (grooving) occurs are not signicantly higher than
those for situations where particle rolling occurs and it is argued that this is associated with the way
in which grooving particles orient themselves with respect to the testpiece as they pass through the
contact. This is in contrast to xed-particle grooving abrasion (such as might be observed in tests using
abrasive papers), and as such, it is argued that what is commonly termed two-body abrasion should be
subcategorised into xed-particle grooving abrasion and free-particle grooving abrasion. The paper
then proceeds to provide an analysis of the motion of particles in the dry sand-rubber wheel abrasion
test, and seeks to understand the mechanics controlling their motion, and thus the dependence of particle
motion upon external factors, focussing on the effects of testpiece hardness and applied load. The effect of
hardness on particle rotation is well predicted by the model, but the effect of the applied load on particle
motion observed experimentally is opposite to that which is predicted by the model. The shortcomings of
the model are discussed, and the model has been qualitatively modied to account for this discrepancy.
The modications centre around the signicant changes in rubber wheelparticle contact geometry as
the applied load is changed; such changes are difcult to model analytically due to the large strains
associated with such a contact, and it is suggested that nite element modelling may be required to fully
understand and model the complex contacts occurring in this simple and widely employed test method.
2009 Elsevier B.V. All rights reserved.

1. Introduction
Due to its importance, much work has been conducted in the
area of abrasive wear and a correspondingly large number of test
methods have been used to evaluate abrasion resistance of materials. Often, test apparatus are designed with a specic service
application in mind [1]. Test methods can be broadly divided into
those where the abrading medium is loose as it passes over the testpiece (commonly termed three-body abrasion) and those where
the abrading medium is xed in orientation as it passes over the
testpiece (commonly termed two-body abrasion) [2]. The most

Corresponding author. Tel.: +44 115 951 3760; fax: +44 115 951 3764.
E-mail address: philip.shipway@nottingham.ac.uk (P.H. Shipway).
0043-1648/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2009.08.013

commonly employed test setup for three-body abrasion is that

encompassed in the ASTM standard G65 [3], commonly known as
the dry sandrubber wheel (DSRW) test. In this test, a stream of particles is fed through the loaded contact between a test specimen
and a rotating rubber wheel; the abrasive particles pass through
the contact region once only. Since the rubber wheel rotates and
the abrasive passes through the gap, it is clear that no permanent
embedment of the particles onto the steel surfaces can take place.
Instead a particle can traverse through the contact either by rolling
or by becoming embedded into the rubber wheel and being dragged
through and forming a groove in the metal sample as it does so.
The rst major study using a rubber wheel abrasion tester was
that conducted by Haworth [4]. The apparatus employed in that
work was akin to that now used in the wet sandrubber wheel test
[5] in that abrasive was contained in a bin in which the rotating

2084

S.M. Nahvi et al. / Wear 267 (2009) 20832091

wheel was immersed. Abrasive particles were carried into the contact zone between the specimen and the wheel by the motion of
the wheel, aided by transverse grooves in the rubber. The test could
be operated with both slurry and dry abrasive. Subsequent developments resulted in the two distinct methodologies for testing in
slurry or dry conditions [6]. The DSRW test has been employed to
examine the abrasion behaviour of a very wide range of materials.
In many programmes, the test is used simply to provide a quantitative ranking of the abrasion resistance of different materials. For
example, the behaviour of a series of steels with a wide range of
hardness has been tested, and whilst good correlation was found
between wear rates and hardness, the operative mechanisms of
wear were never examined [7,8].
The operative mechanisms of wear depend largely upon the
material properties (e.g. hardness, ductility, toughness) along with
the manner in which the particles move through the contact
between the wheel and specimen. The particles may embed into the
moving rubber wheel and slide across the sample material through
the contact region (generally termed two-body abrasion) or pass
through the contact region by rolling between the rubber wheel
and the sample (generally termed three-body abrasion). Due to
the imprecise denitions of two-body and three-body abrasion [9],
the terms sliding (or grooving) abrasive wear and rolling abrasive
wear have been suggested as more precise terminology [10]. There
is, however, a signicant difference in the mode of wear between
grooving abrasion with free particles (as potentially observed in the
DSRW test) and grooving abrasion where the orientation of the particles is xed (as commonly observed in tests which use abrasive
papers). Fixed-particle grooving is typically observed to generate
wear rates a factor of ten greater than comparable tests with free
abrasives [2]. This is due to the fact that the attack angle of a number of the particles on the abrasive paper will be high, promoting
cutting as an efcient mode of material removal [11,12]. However,
it is also known that if the attack angle of such grooving particles is
low, they will tend not to cut the surface (with the associated high
rates of material removal) but will deform the surface by ploughing
and wedge formation with much lower rates of material removal
[12]. As such, it is clear that there is no intrinsic link between observations of grooving abrasion and high rates of wear. Indeed, in work
looking at abrasion of steels with silica in both wet and dry conditions, Wirojanupatump and Shipway showed that in dry conditions,
the particles rolled through the contact whereas in wet conditions,
they slid (grooved) through the contact; here, the grooving contact was associated with wear coefcients an order of magnitude
less than that associated with the rolling abrasion and this was
attributed to the lubricating effect of the water which allowed the
particles to slide across the surface with a low attack angle, resulting in benign grooving [13,14]. It has been shown elsewhere [15]
that for particles with idealised shapes (prismatic parallelograms),
particles which are grooving through a contact must orient themselves in such a way that the turning moment on the particle is zero;
depending upon the shape of the particle, this may result in a low
attack angle, and thus to low rates of wear. Thus, in wear testing
terminology, grooving abrasion should be further subdivided into
xed-particle grooving and free-particle grooving, since these will
typically result in very different processes of material removal and
to very different rates of wear.
Whilst it has been recognised that the manner in which the
particles move through the contact affects the mode of wear and
thus ultimately the rate of material removal, the motion of particles
through the contact zone has itself been shown to depend upon a
number of parameters associated with the system, amongst them
particle shape, applied load and the hardnesses of the test surface
and counterbody [10,15,16]. It has been observed that during abrasion testing utilising the DSRW test, the mechanism of abrasion
changed from rolling at low loads to grooving at higher loads [17].

In addition, examination of the morphology of wear scars following

abrasion with loose abrasive particles has shown that softer samples tend to exhibit rolling motion whereas as hardness increases,
particle grooving across the surface dominates [16].
However, despite the recognition of the critical role of the particle motion in controlling the mode and thus (potentially) the rate
of abrasive wear, surprisingly little work has examined the basic
mechanics of the particle motion. The motion of particles in a contact which is held open either by a bed of particles or by uid forces
associated with the motion has been considered, [1822] but those
situations are different from the one being considered for the DSRW
test in which the load is borne by the same particles which are
they themselves being studied in terms of their motion. In a few
papers, models to describe the motion of individual particles have
been formulated and presented, which examine the moments upon
such particles during a test [15,23]. The model of Fang et al. [23]
considers the motion of a particle in terms of the turning moment
acting upon the particle; the simplicity of this model makes it useful
for studying the effects of various external parameters on particle
motion.
In this work, the motion of particles in the DSRW test with silica
abrasive against a range of steels was considered, both as a function
of applied load, but also as a function of the hardness of the steels.
The changes in behaviour are rationalised in terms of the mechanics
governing particle motion.
2. Experimental methods
2.1. Abrasives
A silica sand (David Ball, Bar Hill, UK) with a nominal particle
size ranging from 180 to 250 m was used. An SEM micrograph of
the rounded silica sand is presented in Fig. 1a which illustrates the
morphology of the abrasive particles. A back-lit optical micrograph
is presented in Fig. 1b which allowed image analysis of the projection of the particle shape to be conducted; the particle aspect ratio
(ARp ) was dened as the ratio of the maximum Feret diameter to
the minimum Feret diameter. The average value of ARp for the particles was 1.51 with the maximum and minimum values being 1.70
and 1.28 respectively. The particle size distribution was measured
by laser granulometry (Malvern Instruments) with the result being
shown in Fig. 1c. The particle diameter, dp , below which 50% of the
volume lay was 251 m, and this was taken as the nominal particle
diameter of the particles in any further analysis. The Vickers hardness, Hv , of the silica was measured as 956 kgf mm2 with a 200 gf
indentation load (Table 2).
The fracture strength of particles in each of the sieved size fractions was measured by compression testing between hard steel
platens. For each particle size range, ten individual particles were
chosen and their strengths measured individually with an Instron
Universal Tester tted with a 50 N load cell. The machine was
run in displacement control; when particles fractured, the load
dropped. The load at rst fracture and the maximum load borne by
each particle were measured and the averages calculated. For two
sieved fractions from the abrasives used (namely 180212 m and
212250 m), the average fracture loads of the abrasive particles
were 6.5 N and 7.2 N respectively.
2.2. Test materials
Five steels were employed as materials for wear testing. These
were Armco Iron, a low carbon steel, a mild steel, Hardox 400 and a
ground at stock (GFS). These have been designated C02, C04, C12,
C18 and C99 respectively (utilising the C to indicate the carbon level
and the two numbers to represent the rst two decimal places in

S.M. Nahvi et al. / Wear 267 (2009) 20832091

2085

Table 1
Chemical analysis of the ve steels examined.
Element

C02

C04

C12

C18

C99

Fe
C
Si
Mn
Cr
Mo
Ni
Cu

Bal.
0.020
<0.005
0.046
0.012
0.006
0.020
0.012

Bal.
0.047
<0.005
0.17
0.020
0.007
0.020
0.007

Bal.
0.12
0.22
0.77
0.067
0.021
0.113
0.373

Bal.
0.18
0.32
1.20
0.227
<0.005
0.009
0.006

Bal.
0.99
0.27
1.09
0.485
0.160
0.524
0.162

Table 2
Hardness of the ve steels examined, along with that of the silica abrasive.
Sample

Hv (kgf mm2 )

Indentation load

C02
C04
C12
C18
C99
Silica abrasive

80
117
242
473
830
956

10 kgf
10 kgf
30 kgf
30 kgf
30 kgf
200 gf

2.3. Abrasive wear testing

The DSRW technique using a rig based on the relevant ASTM
standard [3] was employed for free-particle abrasive wear testing.
This modied design is based on that reported by Stevenson and
Hutchings [24]; a schematic diagram of the abrasion rig is shown
in Fig. 2. The polyurethane rubber tyre (with an international rubber hardness of 61 at room temperature) has an external diameter
of 227 mm, a breadth of 12.7 mm and a thickness of 10 mm. It is
rotated at 195 revolutions per minute, equivalent to a sliding speed,
v, of 2.32 m s1 . The abrasive particles are metered onto a chute, by
which the abrasive is introduced onto the rubber wheel just before

Fig. 1. (a) SEM micrograph showing morphology of silica abrasive particles, (b) optical micrograph of particles for assessment of particle shape, and (c) size distribution
of silica abrasive measured with laser granulometry.

the weight percentage of carbon in the alloys). The compositions

of these steels were determined by atomic emission spectroscopy
(using a Foundry Master, Worldwide Analytical Systems AG, Germany) and are shown in Table 1. The GFS was austenitized at 810 C
before quenching into oil; 0.5 mm was then ground from its surface to remove any decarburized material. All the other materials
were used in the as-received state. The Vickers hardnesses, Hv , of
the ve steels (the average of ve individual measurements taken
under a 30 kgf indentation load) are presented in Table 2. The materials were cut into testpieces (59 mm 25 mm 12 mm) and the
surfaces to be exposed to wear (one of the largest faces on each
sample) were ground with successively ner silicon carbide abrasive papers and nally with a fabric pad loaded with 6 m diamond
abrasive.

Fig. 2. Schematic diagram of the DSRW test apparatus.

2086

S.M. Nahvi et al. / Wear 267 (2009) 20832091

Table 3
Particle parameters associated with wear as a function of applied load; applied load,
Papp ; measured wheel-specimen contact length, Lc ; ow fraction, f; number of particles in contact zone, N; average load per particle in the contact zone, Pp ; particle
packing fraction in contact zone, fpack .
Papp (N)

Lc (mm)

Pp (mN)

fpack

19.6
49.1
68.7
98.1
127.5

20
25
27
30
32

0.8
0.71
0.71
0.67
0.45

745
826
893
936
670

26.3
59.4
76.9
104.8
190.2

0.097
0.086
0.086
0.081
0.054

the contact region between the test specimen and the wheel. The
abr , was 2.37 g s1 . The abraaverage metered abrasive feedrate, m
sive is dragged through the contact zone which is loaded by a dead
weight on the sample. The mass loss of the sample was measured
before and after every test at 100, 200, 400, 600, 800 and 1000 revolutions. The wear rate is taken as the gradient of the steady-state
part of the plot of mass loss versus abrasion distance.
Not all of the abrasive fed towards the contact actually passed
through the contact zone, with some falling off the wheel to the
sides. To allow separate collection of abrasive particles which had
passed through the contact zone from that which had not, a pair
of stationary brushes were placed along the vertical radius on the
wheel up to the level of the test specimen (one each side of the
wheel); the brushes separated the two abrasive collection routes.
Thus, abrasive particles that passed through the contact zone would
be collected separately from those that fell away from the wheel
before being passed through the contact. As such, the actual feedrate through the contact was calculated. The ratio of the actual to
the metered feedrate of particles was termed the ow fraction, f;
this was measured as a function of applied load, Papp , and the results
shown in Table 3.
A number of samples were instrumented with thermocouples
which were inserted into holes drilled to within 1 mm of the contact surface in the test specimens, and the temperature measured
during the wear test.
2.4. Characterisation worn surfaces
The worn surfaces were examined after a small sliding distance
and at full sliding distances for all loads examined. The length of the
wear scar was measured, which was primarily a function of applied
load (indicating that the primary inuence on wear scar length
was the load-dependent elastic deformation of the rubber tyre). For
each load, an average contact length, Lc , was determined, as shown
in Table 3. The Youngs modulus of the rubber can be estimated
from the rubber hardness of 61 IRHD to be around 3.5 MPa [25] and
the Poissons ratio estimated to be 0.5 (typical for a rubber). Using
simple contact mechanics for a line loaded contact, contacts lengths
of 13 mm (loaded under 19.6 N) and 36 mm (loaded under 127 N)
are estimated. These are reasonably correlated with the experimental values in Table 3, but show a larger range of values than those
observed experimentally.
A Philip XL30 scanning electron microscope (SEM) utilising
secondary electron (SE) signals was employed to examine the morphology of the worn surfaces. SEM micrographs of the central zones
of the wear scars are presented to allow a direct comparison to be
made between equivalent regions on each scar.
3. Results
3.1. Abrasion of steels
Fig. 3 shows data concerning the wear rate of the ve grades
of steel as a function of applied load. Fig. 4 shows the plan view

Fig. 3. Abrasion rates of the ve steels as a function of applied load in the DSRW
test.

SEM images of the central zone of the wear scars on the ve grades
of steel following abrasion at the lowest and highest loads utilised
(19.6 N and 127 N respectively). In all cases the sliding direction
of the wheel across the sample has been in the vertical direction.
Some of the surfaces (e.g. Fig. 4a) show evidence typical of particle rolling, with signicant indentation of the surface and little
directionality. In contrast, Fig. 4j shows evidence typical of particle
sliding (grooving) across the surface of the sample. It can be seen
that particle rolling is promoted by low applied loads and low sample hardness, whilst particle sliding (grooving) is promoted by high
applied loads and high sample hardness.
The steady-state sample temperatures during abrasion of the
C99 steel are shown in Table 4 as a function of the applied load,
Papp .
3.2. Particle loading during abrasion testing
An estimate of the number of particles, N, in the contact zone
between the rubber wheel and the specimen at any point in time
is given by
N=

abr f Lc
m
v mp

(1)

where mp is the mass of a particle in that size fraction, all other

terms having been dened previously. Both Lc and f are functions
of the applied load, as shown in Table 3.
The mass of a particle is given by its volume and density. Assuming the particles to be spheres, the particle mass, mp , is given by
mp =

1
p dp3
6

(2)

where p is the density of the particle. In this case, the particles

are assumed to be silica spheres with a diameter of 251 m and a
density of 2650 kg m3 .
To estimate the fraction of the total load borne by each individual size fraction is difcult. Avery [6] suggests that due to the low
elastic modulus of the rubber wheel, it can be assumed that for an
abrasive feedstock with a relatively narrow size fraction, the load
per particle can be assumed to be the same for all particles within
Table 4
Measured steady-state temperatures of C99
steel as a function of applied load.
Papp (N)

Tm ( C)

19.6
68.7
127.5

36
87
108

S.M. Nahvi et al. / Wear 267 (2009) 20832091

2087

Fig. 4. SEM micrographs of the central regions of the wear scars on the ve grades of steel for both the minimum and the maximum applied loads.

the contact. Thus the load per particle, Pp , is given by

Pp =

Papp
N

(3)

where Papp is the total applied load. The loads per particle for the
various test conditions are shown in Table 3.

In addition to the load per particle, Stevenson and Hutchings

showed data for the packing fraction of the particles [24]. The packing fraction, fpack , was dened as
fpack =

abr f
m

v dp b p

(4)

2088

S.M. Nahvi et al. / Wear 267 (2009) 20832091

where b is the breadth of the wheel. The packing fraction under the
various test conditions are also shown in Table 3.
4. Discussion
4.1. Wear rates and mechanisms
The wear rates of the steels as a function of applied load are
shown in Fig. 3. For the three steels with the lower hardnesses
(C02, C04 and C12), it can be seen that the wear rates increase
linearly with load at the lower loads, with the rate of increase
increasing with load as higher loads are applied. Similar behaviour
has been reported in the literature [26]. Also, it was observed that
in general, the wear rate of the steels decreased with increasing
hardness; however, this statement is not universally valid since at
number of the loads examined, the C04 steel (with a hardness of
117 kgf mm2 ) exhibited a higher wear rate than the C02 steel (with
a hardness of 80 kgf mm2 ), indicating that other factors (such as
other materials properties or the mode of material removal) were
inuencing the wear rate.
Fig. 4 shows clearly that the motion of particles through the
contact zone depends upon both the material type and the applied
load. Particle rolling is favoured in testing of sample materials of
low hardness and by low applied loads, whereas particle sliding
(grooving) is favoured by samples of high hardness and by high
applied loads. For the two hardest steels (C18 and C99), particle
sliding (grooving) operates across the range of applied loads applied
(see Fig. 4). However, for the steels of lower hardness (C02, C04 and
C12), rolling is favoured at the lower loads and sliding (grooving)
at the higher loads. This change in mechanism may be the cause of
the non-linear dependence of wear rate with applied load for these
softer steels as observed in Fig. 3.
4.2. Particle motion during abrasion
A model of particle motion in abrasion testing with loose abrasives has been proposed by Fang et al. [23]. In that work, the
forces acting on the particle were indicated as in Fig. 5. It was proposed that the particle would move through the contact by sliding
(grooving) if the clockwise moment was less than the anticlockwise
moment, i.e.
Fp h < Pp e

(5)

where Fp is the lateral force on the particle and the dimensions e and
h are as dened in Fig. 5. If the inequality in Eq. (5) is not satised,
the particle would roll through the contact. Thus, to understand the
nature of particle motion through the contact, all the terms in Eq.
(5) must be considered.

Fig. 6. Schematic diagram showing the various geometrical features of a hard rigid
sphere being loaded against a plastic counterbody.

4.2.1. Consideration of the effect of sample hardness

For a given applied load on a particle, it has been shown that
a high sample hardness favours particle sliding (grooving) whilst
a low sample hardness favours particle rolling. By examination of
Eq. (5), this implies that the ratio of Fp to Pp changes with sample hardness, assuming that for a given particle type and particle
geometry the ratio of e to h is independent of sample hardness.
This latter assumption can be shown to be reasonable since the
largest metalparticle contact area in the cases considered in this
work (given by the highest load per particle of 190 mN on the metal
with the lowest hardness of 80 kgf mm2 ) is given by the ratio of
the load to the hardness which is of the order of 240 m2 . If this
area is assumed to be semicircular in plan view, it gives an estimate
of the contact dimension of around 12 m. Thus any changes over
dimensions of this order of magnitude will tend to be insignicant
compared to the dimensions of the particle size itself, these being
of the order of 250 m.
To allow a simple model of particle motion to be developed
requires that the shape of the particle in contact with the metal (and
also in contact with the rubber wheel) be considered. Torrance [27]
addressed the variation of attack angle with depth of penetration
of an abrasive particle in machining (where sharp particles such
as alumina are employed). In this case, he employed a hyperbolic
prole for the abrasive tip shape since he argued that modelling of
particles as spheres provided attack angles which were too low. In
contrast, other workers have modelled abrasives as sphere-ended
cones or pyramids, allowing the radius of the tip to be decoupled from the abrasive particle size [28,29]. Pintaude et al. [30]
conducted two-body abrasion tests against abrasive papers. They
observed that friction was higher as the metal being tested became
softer, indicating that higher penetration of the abrasive particle
into the surface leads to an increase in ploughing friction. However, they indicated that these observations cannot be explained if
a model which employs a geometry which is independent of indentation depth is employed. Similarly, work by Goddard and Wilman
[31] has shown that for indenters for which the shape is independent of depth of indentation, there can be no change in the ratio of Fp
to Pp as the hardness changes. However, if the indenter has a spherical shape, the ratio of Fp to Pp does change with depth of indentation
(and thus with sample hardness under a given indentation load).
In light of the above, if a grooving spherical contact is assumed,
it can be argued that the load on each particle is borne by contact
over half of a spherical cap (see Fig. 6). Assuming that the pressure
exerted by the steel surface on the particle is given by its hardness,
then the following may be shown [31]:
Pp =

 r2
Hv
2

and

Fig. 5. Model of forces on particle in contact.

Fg
2
g =
=
Pp




R2
r2


1

sin

r
R



R2
r2


1

(6)

S.M. Nahvi et al. / Wear 267 (2009) 20832091

where g is known as the grooving coefcient of friction, Fg is the

force required to overcome the plastic resistance to grooving and
the dimensions r and R are as dened in Fig. 6. It must be noted that
the radius of the spherical cap contact, R, in Fig. 6 is not the same
as the particle radius.
For a grooving contact, the force Fp indicated in Fig. 5 has two
main origins; (i) that associated with overcoming the plastic resistance to grooving (Fg ) and (ii) that associated with normal adhesive
friction, Fa , and as such the total lateral force, Fp , is given by
Fp = Fg + Fa = g Pp + a Pp

(7)

which thus yields

p = g + a

(8)

where p is the observed coefcient of friction in grooving. Whilst

no values for a have been measured in this work, values of p
of around 0.4 have been quoted by Stevenson and Hutchings for
similar tests [24] (although it is not reported as to whether these
values were from rolling or grooving motion of particles, nor what
the hardness of the sample steel was). Although the loads in the
work of Stevenson and Hutchings were not dissimilar to those used
in the current work, the packing fraction (fpack ) was much higher in
that work (see Table 3) and thus the loads per particle in the work
were all less than 25 mN, which is less than the lowest load per
particle utilised in the current work. Knowing that the contribution
of g reduces as the contact radius r becomes smaller (associated
with a reduction in applied load), it is thus assumed that the value
of p of around 0.4 measured by Stevenson and Hutchings can be
reasonably used as the value for a in this work.
Fig. 7 shows the calculated values of observed coefcient of
friction (p ) for the lowest and highest values of load per particle employed in the experiments over the range of hardnesses of
the steels examined (see Tables 2 and 3). The data are presented
for two values of R (see Fig. 6), namely 126 m and 12.6 m. The
former value represents the case when the radius of the spherical
cap in contact with the metallic surface is the same as that of the
particle itself, whilst the latter case represents a case where the
radius of the contacting asperity is much smaller than the particle radius. It can be seen that in both cases, for a given load per
particle, the value of p decreases with increasing hardness of the
metal sample. (The trends in p with applied load per particle will
be considered separately in Section 4.2.2.) According to Eq. (5), this
decrease in the ratio of Fp to Pp with increasing sample hardness will
tend to promote particle sliding (grooving) as the sample hardness
increases, as is observed in the experiments. As such, the change in
particle motion associated with sample hardness under conditions
of constant applied load can simply be attributed to changes in the
grooving friction coefcient associated with particle indentation
depth (and thus the overall observed friction coefcient).
Consideration of Eq. (5) indicates, therefore, that the observed
coefcient of friction must be greater than the ratio (e/h) to promote particle rolling. An estimate for the maximum value of (e/h)
for the particles can be given by the particle aspect ratio (ARp ),
namely 1.51 for this set of particles. Consideration of simple geometry indicates that the maximum apparent coefcient of friction
(p ) will be realised when the grooving coefcient of friction (g )
is at a maximum; from Eq. (6), it can be seen that this is achieved
when the depth of indentation of the spherical indenter is equal
to its radius, whereupon the observed coefcient of friction (p )
will be (1 + a ). Earlier in this section, it was argued that a sensible value of a was 0.4 and thus it can be seen that the maximum
value of (1 + a ) is less than the value of the particle aspect ratio
(ARp ) of 1.51 required to cause particle rolling. Given that particle
rolling has been experimentally observed indicates that there are
a number of assumptions which are not valid, such as the simple

2089

view that complex particle shapes can be expressed by the value of

ARp .
4.2.2. Consideration of the effect of applied load
In the previous section, the effect of changes in the sample hardness were considered, with no assessment of what was occurring
at the rubber wheelparticle interface (since it was assumed that
since situations of constant load were being considered, changes at
the particlerubber interface would be limited to effects of changes
in the lateral force). However, it was clear from Fig. 7 that the
observed friction coefcient would increase with increasing load
per particle. Fig. 8 shows the calculated values of observed coefcient of friction (p ) for the lowest and highest values of hardness
of the steels examined over the range of applied load per particle
employed in the experiments (see Tables 2 and 3). As in Fig. 7, the
data are presented for two values of R (see Fig. 6), namely 126 m
and 12.6 m. The reasons for the trends observed are the same as
they were for the trends in hardness, namely that for a spherical
indenter, the ploughing friction coefcient increases as the depth
of indentation increases. A similar increase in observed coefcient
of friction (p ) with increasing applied load has been observed in
the DSRW tests reported by Dube and Hutchings [26].

Fig. 7. Observed coefcient, p, of friction as a function of metal surface hardness

for a sphere grooving through a metallic surface for two applied loads as shown: (a)
sphere radius = 126 m and (b) sphere radius = 12.6 m. a assumed to be 0.4 in all
cases.

2090

S.M. Nahvi et al. / Wear 267 (2009) 20832091

Since it has been argued that low observed coefcients of friction favour particle sliding (grooving), particle sliding is thus shown
to be favoured by low applied loads per particle. This prediction
is in contrast to the observed phenomena, where quite the opposite is observed, namely that for low applied loads, particle rolling
becomes more favoured. These apparent inconsistencies can be
resolved by consideration of the equation which governs the transition between sliding (grooving) and rolling behaviour (Eq. (5)).
Sliding will occur when the observed friction coefcient (p ) is less
than that of the ratio (e/h) as shown in Fig. 5. As the observed friction coefcient for particle sliding (p ) increases with applied load,
then this implies that the ratio (e/h) must be increasing with load
more quickly so that the transition between rolling at low load and
sliding at high load observed for the lower hardness steels in Fig. 4
can be rationalised.
It has been shown that when there is adhesion between a rubber and a rigid body with tangential motion between them, there
will be a non-symmetric distribution of contact forces across the
contact area [32]. Such asymmetry of contact has between a stiff
sphere and a rubber in sliding has also been shown by a number
of workers [3339]. Fig. 9 (adapted from [35]) shows the type of

Fig. 9. Schematic diagram of the shape of the contact between a rubber and a hard
sphere in the presence of a tractional load: (a) low load per particle and (b) high
load per particle.

asymmetrical contact that might be expected. With this in mind,

as the load is increased, it can be seen that e increases in size and
that h decreases in size, both of which result in an increase in the
ratio (e/h). This increase in (e/h) with load must be more rapid than
the increase in p with load, and as such the observation of particle
motion changing from rolling at low loads to sliding at high loads
can be qualitatively understood. However, the change in shape of
the contact with applied load (as shown schematically in Fig. 9)
needs to be described quantitatively to allow a fuller understanding
of the particle motion in such contacts to be developed.
The change in shape of the contact between the particle and
the rubber wheel associated with increases in load may be more
signicant than might be expected by a consideration of the loads
themselves due to temperature effects in the rubber. Table 4 shows
that the steady temperature of the sample (which gives some indication as to the value of the temperature of the rubber) showed
signicant increases above room temperature, with the temperature increasing with applied load. The increase in temperature will
cause a decrease in the measured rubber hardness [24], and thus
to a decrease in elastic moduli [25] and to higher strains associated with the lateral and normal forces on the particles as shown
in Fig. 9.

Fig. 8. Observed coefcient, p, of friction as a function of applied load for a sphere
grooving through a metallic surface for two metal surface hardnesses as shown: (a)
sphere radius = 126 m and (b) sphere radius = 12.6 m. a assumed to be 0.4 in all
cases.

4.2.3. General observations

Using a range of steels with a wide range of hardness has shown
that under a standard set of conditions, particle motion through the
rubber wheeltestpiece contact depends upon the testpiece properties. Such a dependence must be of concern to users of the test,
since a robust test methodology will expose all materials to the
same conditions which themselves are independent of the material properties. In light of this, it is recommended that observation
of wear scars following exposure to the DSRW test and identication of the mode of particle motion through the contact is a routine
part of this test methodology.
Having highlighted the differences in the motion of particles, it
is also noted that the wear coefcients observed are not strongly
dependent upon the mode of particle motion through the contact.

S.M. Nahvi et al. / Wear 267 (2009) 20832091

This is in contrast to the very different wear coefcients associated commonly associated with three-body abrasion and two-body
abrasion, with the latter producing wear coefcients an order of
magnitude higher than the former. In the DSRW test, even if the
particles groove through the contact, they are able to orient themselves so that their attack angles on the testpiece are not high, thus
resulting in ploughing and wedging. This is in contrast to two-body
abrasion with xed abrasives (such as abrasive papers) where some
of the particles will have very high attack angles on the testpiece,
producing cutting wear with its associated high wear coefcients.
As such, there is a need to distinguish between xed-particle
grooving abrasion and free-particle grooving abrasion.
Development of a model has been shown to useful in attempting to understand the complexities of the DSRW test. Whilst the
contact of a particle with the metal testpiece can be modelled with
a quantitative set of rules, the details of the interaction of the particle with the rubber wheel is less easy to model. As such, to be
able to fully model this latter contact would produce a clearer and
fuller understanding of particle motion; moreover, this could then
be developed to allow the attack angles of individual particles as a
function of particle shape to be understood. The large strains associated with the elastic indentation of the particles into the rubber do
not lend themselves to analytical solutions, and, as such, nite element analysis may be required to fully understand the mechanics
of the contact between the particle and the rubber wheel.
5. Conclusions
This work has shown that the movement patterns of abrasive
particles through the gap in the DSRW test is a function of both
applied load and hardness of the material under test. It has been
shown that particle sliding (grooving) is favoured by high applied
loads and by high substrate hardness. A model of the particle
motion in the contact has been developed, based upon the work
of Fang et al. [23]. The effect of hardness on particle motion is well
predicted by the model; however, the effect of the applied load on
particle motion is counter to that which is produced by the model,
where the experiments show that sliding (grooving) was favoured
by high loads. The mechanics of the contact between the particle
and the rubber wheel has been addressed by modifying the model
to consider the effect of increased applied load (and thus friction)
on the moment equations upon which the model is built. This modication to the model is qualitative in nature, and in the future, a
robust model which describes the motion of particles in the contact
must be developed which in particular addresses the detail of the
particlerubber wheel contact mechanics.
References
[1] A. Misra, I. Finnie, A review of the abrasive wear of metals, Journal of Engineering Materials and TechnologyTransactions of the ASME 104 (1982)
94101.
[2] I.M. Hutchings, TribologyFriction and Wear of Engineering Materials, Edward
Arnold, London, 1992.
[3] Standard test method for measuring abrasion using the dry sand/rubber wheel
apparatus, ASTM standards, G65-04, 2004.
[4] R.D. Haworth, The abrasion resistance of metals, Transactions of the American
Society for Metals 41 (1949) 819869.
[5] Standard test method for conducting wet sand / rubber wheel abrasion tests,
ASTM standards, G105-02, 2007.

2091

[6] H. Avery, An analysis of the rubber wheel abrasion test, in: S.K. Rhee, A.W. Ruff,
K.C. Ludema (Eds.), International Confederation on Wear of Materials, ASME,
New York, 1981, pp. 367378.
[7] J.A. Hawk, R.D. Wilson, J.H. Tylczak, O.N. Dogan, Laboratory abrasive wear
tests: investigation of test methods and alloy correlation, Wear 229 (1999)
10311042.
[8] J.H. Tylczak, J.A. Hawk, R.D. Wilson, A comparison of laboratory abrasion and
eld wear results, Wear 229 (1999) 10591069.
[9] J.D. Gates, Two-body and three-body abrasion: a critical discussion, Wear 214
(1998) 139146.
[10] R.I. Trezona, D.N. Allsopp, I.M. Hutchings, Transitions between two-body and
three-body abrasive wear: inuence of test conditions in the microscale abrasive wear test, Wear 229 (1999) 205214.
[11] T.O. Mulhearn, L.E. Samuels, The abrasion of metals: a model of the process,
Wear 5 (1962) 478498.
[12] J.M. Challen, P.L.B. Oxley, An explanation of the different regimes of friction and
wear using asperity deformation models, Wear 53 (1979) 229243.
[13] S. Wirojanupatump, P.H. Shipway, A direct comparison of wet and dry abrasion
behaviour of mild steel, Wear 235 (1999) 655665.
[14] S. Wirojanupatump, P.H. Shipway, Abrasion of mild steel in wet and dry conditions with the rubber and steel wheel abrasion apparatus, Wear 239 (2000)
91101.
[15] P.H. Shipway, A mechanical model for particle motion in the micro-scale abrasion wear test, Wear 257 (2004) 984991.
[16] L. Fang, Q.D. Zhou, Y.J. Li, An explanation of the relation between wear and
material hardness in three-body abrasion, Wear 151 (1991) 313321.
[17] J. Larsen-Basse, V. Premaratne, Effect of relative hardness on transitions in abrasive wear mechanisms, in: Proceedings of International Conference on Wear of
Materials, ASME, New York, 1983, pp. 161166.
[18] J.L. Xuan, I.T. Hong, E.C. Fitch, Hardness effect on three-body abrasive wear
under uid lm lubrication, Journal of TribologyTransactions of the ASME
111 (1989) 3540.
[19] Y. Kusano, I.M. Hutchings, Modelling the entrainment and motion of particles in
a gap: application to abrasive wear, Proceedings of the Institution of Mechanical
Engineers Part JJournal of Engineering Tribology 217 (2003) 427433.
[20] K. Adachi, I.M. Hutchings, Wear-mode mapping for the micro-scale abrasion
test, Wear 255 (2003) 2329.
[21] J.A. Williams, A.M. Hyncica, Abrasive wear in lubricated contacts, Journal of
Physics DApplied Physics 25 (1992) A81A90.
[22] J.A. Williams, A.M. Hyncica, Mechanisms of abrasive wear in lubricated contacts, Wear 152 (1992) 5774.
[23] L. Fang, X.L. Kong, J.Y. Su, Q.D. Zhou, Movement patterns of abrasive particles
in three-body abrasion, Wear 162 (1993) 782789.
[24] A.N.J. Stevenson, I.M. Hutchings, Development of the dry sand rubber wheel
abrasion test, Wear 195 (1996) 232240.
[25] Rubber, Vulcanised or thermoplasticdetermination of hardness (hardness
between 10 IRHD and 100 IRHD), British Standard ISO 48 (2007) 1-19.
[26] N.B. Dube, I.M. Hutchings, Inuence of particle fracture in the high-stress and
low-stress abrasive wear of steel, Wear 235 (1999) 246256.
[27] A.A. Torrance, Modelling abrasive wear, Wear 258 (2005) 281293.
[28] S. Jacobson, P. Wallen, S. Hogmark, Fundamental-aspects of abrasive wear studied by a new numerical-simulation model, Wear 123 (1988) 207223.
[29] T.W. Hwang, S. Malkin, Upper bound analysis for specic energy in grinding of
ceramics, Wear 231 (1999) 161171.
[30] G. Pintaude, D.K. Tanaka, A. Sinatora, The effects of abrasive particle size on the
sliding friction coefcient of steel using a spiral pin-on-disk apparatus, Wear
255 (2003) 5559.
[31] J. Goddard, H. Wilman, A theory of friction and wear during the abrasion of
metals, Wear 5 (1962) 114135.
[32] Q.V. Bui, J.P. Ponthot, Estimation of rubber sliding friction from asperity interaction modeling, Wear 252 (2002) 150160.
[33] M. Barquins, A.D. Roberts, Rubber friction variation with rate and
temperaturesome new observations, Journal of Physics DApplied Physics
19 (1986) 547563.
[34] A. Schallamach, How does rubber slide? Wear 17 (1971) 301312.
[35] M. Barquins, Sliding friction of rubber and Schallamach wavesa review, Materials Science and Engineering 73 (1985) 4563.
[36] E.L. Deladi, M.B. de Rooij, D.J. Schipper, Modelling of static friction in rubbermetal contact, Tribology International 40 (2007) 588594.
[37] E.L. Deladi, M.B. de Rooij, E.G. de Vries, D.J. Schipper, Analytical and experimental investigation of the static friction regime for rubber-rigid ball contact,
Surface and Interface Analysis 38 (2006) 891893.
[38] M. Barquins, Friction and wear of rubber-like materials, Wear 160 (1993) 111.
[39] M. Barquins, Adherence, friction and wear of rubber-like materials, Wear 158
(1992) 87117.