Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

WEAR

ELSEVIER
Wear 195 (1996) 232-240

Development
University

of the dry sand/rubber

wheel abrasion test

A.N.J. Stevenson I, I.M. Hutchings


of Cambridge, Department of Materials Science and Metallurgy, Pembroke Street, Cambridge CB2 SQZ, UK
Received 29 November 1995; accepted 22 February 1996

Abstract
A rubber wheel abrasion tester with some advantages over the design described in ASTM (36.5 has been designed and characterised. In the new design the specimen was held horizontally. It was therefore possible to determine the actual mass Bow rate of abrasive abrading the specimen and consequently the number of particles in the contact zone and the load carried by each particle. The abrasive feed system was designed to improve control over the conditions in the contact region. The performance of the apparatus was evaluated in tests on low carbon (AISI 1020) steel, with wheels made from cast polyurethane and chlorobutyl rubber. The influences of load, sliding speed and rubber hardness were investigated. Temperature rises of wheel, specimen and sand, and the friction force were measured; the main mechanism of heat dissipation was through the continual flow of fresh abrasive past the specimen.
Keywords: Abrasion test; Friction force; Load; Sliding speed; Rubber hardness

1. Introduction The rubber wheel abrasion test (RWAT) as described in ASTM standard G65 [ l] is commonly used to evaluate the abrasive wear behaviour of materials under three-body conditions. The apparatus is shown schematically in Fig. 1. In the test, a plane specimen is loaded against the rim of a rotating rubber wheel; sand is fed into the gap between the wheel and specimen and is carried past the specimen and thus abrades it. The conditions in the test have been described as low stress scratching abrasion [ 21. The test is widely used to rank materials for components that will be subjected to low stress abrasion in service: e.g. agricultural tools, chutes and hoppers in ore processing plant, and construction equipment. Good correlation is generally found between the test results and field experience [ 3,4]. Haworth was the first worker to use the rubber wheel abrasion test in its present form [ 51. The apparatus consisted of a wheel 9 in (228.6 mm) in diameter and 0.5 in ( 12.7 mm) wide, with a rubber rim. Haworth used rubber to maintain the contact pressure as the specimen wore, because he had noted that the pistons of slurry pumps were made from rubber in order to maintain pressure as the liner wore. The specimen was held against the vertical edge of the wheel and was 3 in (76.2 mm) long and 1 in (25.4 mm) wide, twice the width Present address: Tenmat Ltd., Bowdon House, Ashburton Trafford Park, ManchesterM17 IRU, UK.
Road West,

Abrasive hoppa

Fig. 1, Schematic diagram of the rubber wheel abrasion described in ASTM G65 [ 1 ]

testing apparatus

of the wheel. The wheel had shallow grooves distributed around the perimeter, perpendicular to the flat faces, to pick up dry abrasive or slurry from a tray through which it ran. Intermediate moisture contents could not be used because the wheel was unable to pick up and carry abrasive particles which were not either dry or incorporated into a liquid slurry. A variant of Haworth s design was manufactured and marketed by the Fargo Tool Co. of Detroit, USA [6], and led to

0043-1648/96/$15.00 0 1996 Elsevier Science S.A. All rights reserved PllSOO43-1648(96)06965-7

A.N.J. Stevenson. I.M. Hutchings / Wear 195 (1996) 232-240

233

considerable standardisation. Ottawa sand was adopted as a standard abrasive. Early machines, however, had a critical flaw: the pivot bearing was not in the plane containing the friction force. As a result, the normal force on the specimen was significantly influenced by the frictional force acting on the specimen when the machine was running. Borik [ 7-91 used a Fargo machine which was modified to carry a 6 in (152.4 mm) diameter wheel with a 0.5 in ( 12.7 mm) thick neoprene rubber rim, resulting in a wheel 7 in (177.8 mm) in diameter. He used an abrasive slurry which was stirred by paddles on the wheel, and paid special attention to the behaviour of the rubber in the test. He found that the hardness of the rubber was important, but so too was surface finish. After considerable use the rubber appeared glazed and exhibited grooves and ridges parallel to the sliding direction. Dressing an old wheel by grinding it in a lathe dramatically increased the specimen wear rate [ 91; examination showed that abrasive particles had not become embedded in the rubber, and it was concluded that the dressed wheel had a texture particularly suited to holding the abrasive grains which did not allow them to roll. Borik noted that wear rate decreased as the test continued and ascribed this either to the change in area of the wear scar, or to the change in the surface finish of the rubber. Regrinding the wear specimen to produce a flat surface had practically no effect on the continuous decline in the wear rate. A procedure was adopted of running-in the specimen for 5000 wheel revolutions prior to quantitative tests and of dressing the rubber wheel before every test. A series of reproducibility tests led Borik to claim that a 6% difference in wear rate between specimens was significant at the 95% confidence level [ 81. Tucker and Miller [lo] employed conditions similar to those later standardised in ASTM G65. They justified the use of 200-300 p,rn sand particles at a feed rate of 130 g min- by pointing out that in this range the wear rate is insensitive to variations in abrasive size and feed rate. In a series of tests on AISI 1020 steel they found that the results depended on the testing increment. They then decided that attempts to directly relate weight loss data from one test increment schedule to another should be avoided [ lo]. In 198 1 Avery [ 61 wrote a comprehensive paper discussing the RWAT, and discussed the need to keep tight control over all variables to ensure consistency of results. He argued, however, that abrasive feed rate is not an important parameter provided that an even layer of sand is maintained across the specimen face. As the feed rate increases the load on each particle decreases, but more particles abrade the specimen. This suggestion was later disputed by Huard et al. [ 1 l] who showed that the wear rate of AISI 1020 steel increased by 20% when the sand feed rate was increased from 100 g min- to350gmiK . Avery also studied the effect of rubber hardness, finding [ 61 that abrasive wear increases exponentially with its [the rubber] hardness, all other variables being kept constant. This exponential relationship was also found by Borik [ 71. Neoprene rubber hardens with ageing, but chlorobutyl rubber

is supposedly less prone to this. Both types of rubber soften due to frictional heating in the tests, and it was suggested that long runs should therefore be avoided. Huard et al. [ 1 l] concentrated on the influence of the abrasive particles and compared the behaviour with the standard Ottawa sand to that with three foundry sands with rounded particles. The particle shape was characterised; the foundry sand particles were closer to spherical than the Ottawa sand and were larger. The wear rate of AISI 1020 steel decreased with increasing abrasive size, which was attributed to a decrease in particle angularity. They also found that sand particles sticking to the wheel had a significant effect on the results. The particles adhering to the wheel were generally larger and more angular than the average. Brushing the wheel during the test to remove adhering particles reduced the wear rate by = 5%. ASTM Standard G65 was first published in 1980, by which time much of the above characterisation work had been carried out, and there was already widespread use of the conditions that were later standardised. There have been few changes to the test method since it was first written. Indeed Huard s work [ 111 is perhaps the only publication after 1981 to include significant consideration of the behaviour of the apparatus. There are however some drawbacks with the test method. The specimens are relatively large (the recommended size being 76.2 mm long by 25.4 mm wide by 12.7 mm thick) and it is recommended that only one test be conducted on each specimen. The test is of quite long duration; with a highly wear resistant material 5000 wheel revolutions are prescribed (ASTM procedure A), although it would in practice be possible to record a significant mass loss from a much shorter test. Many wear resistant materials are expensive and difficult to machine, so it would ideally be preferable to perform several short tests on one specimen, rather than conduct long duration tests on several large specimens. The standard specifies the type of abrasive to be used: SO70 mesh (212-300 cl.rn) Ottawa sand, although other rounded quartz sands may be used. Whilst rounded quartz sand is a commonly occurring practical abrasive, it must be borne in mind that there will be many situations where significantly different abrasives will be encountered in service and tests may be required with these. The ASTM test involves rather large quantities of abrasive, typically 3-4 kg per test, and for some purposes a test requiring the use of a smaller quantity of abrasive would be desirable. In some contexts, the measurement of the friction force during abrasion testing will provide useful information for use in design. The ASTM standard specifies the composition and mechanical properties (indentation hardness) of the rubber wheel; the possibility of substituting other elastomers with similar elastic properties might be attractive if the wheels are thus more readily fabricated or renovated. The objectives of the present work were to investigate modifications to the dry sand rubber wheel abrasion test in order to address these criticisms.

234

A.N. J. Stevenson, IM. Hutchings / Wear 195 (1996) 232-240

2. Experimental method and materials The apparatus developed in the present work is illustrated in Fig. 2 and incorporates a rubber rimmed wheel with the dimensions specified in ASTM standard G65. In other respects, however, the apparatus differs significantly from the standard method. The test specimen is held horizontally, and a load cell is incorporated to measure the dynamic friction force on the specimen. The specimen is loosely constrained in a yoke, and the normal load is transferred to the specimen by contact with a fixed ball pressing against its upper surface. This allows the specimen to float freely on the wheel. Loading is achieved by adding weights to the arm directly above the specimen. The abrasive feed system shown in Fig. 2 was designed to give a constant feed rate and produce an even monolayer of particles on the wheel. The sand feed rate is controlled by a rotating drum with a shallow groove on its surface, driven by a variable speed motor. The width of this groove matches the internal diameter of the base of a sand hopper in contact with the drum. Sand flows freely down the inclined aluminium chute at angles steeper than = 40 from the horizontal. Inevitably, some abrasive is lost from the edge of the wheel, but this can be quantified by collecting it separately from the

sand which passes through the specimen contact. The quantity of abrasive actually passing the specimen can be determined in this way, which is not possible with the ASTM design. ASTM standard G65 stipulates that the wheel should consist of a steel disc 203.2 mm in diameter and 12.7 mm wide with a chlorobutyl rubber tyre 12.7 mm thick bonded to the rim. A wheel conforming to this specification was used for some tests. Additionally, wheels were fabricated with polyurethane elastomer rims in the following way. Steel blanks 203.2 mm in diameter and 12.7 mm thick were prepared by drilling shallow holes in the circumference and then grit blasting to provide a key for the rubber. Polyurethane casting compounds were used (Monothane grades A40 to A80, Compounding Ingredients Ltd, Preston, UK) to provide a series of rubber rims with different hardnesses. Each blank was placed in a simple brass mould as shown in Fig. 3, and liquid urethane monomer poured into the mould at a temperature of 70 C, at which it flowed viscously. Curing was carried out for 8 h at 135 C in an oven in accordance with the manufacturer s recommendations, and the mould then cooled in the oven to room temperature. Polyurethane rubbers with four different hardnesses in the range 46 to 84 IRHD were used; most of the abrasion tests

Rotating

drum

wheel with rubber rim

Yoke (shown partly

Fig. 2. Schematic diagram of the redesigned apparatus.

A.N. J. Stevenson. I.M. Hutchings / Wear 195 (1996) 232-240

235

Fig. 4. The effect of applied load on the wear rate of 1020 steel. A polyurethane rubber wheel of hardness 61 IRHD was used. The sliding speed was 0.6 m SC throughout and the supplied sand feed rate was 1.0-1.1 g SC . A testing increment of 7 1.8 m ( 100 wheel revolutions) was used.

tapped holes for bolts used to remove cast wheel Fig. 3. Dimensions of mould used to cast tubber-rimmed wheels.

were carried out with a polyurethane wheel of hardness 61+ 1 IRHD at 21 C, within the range specified in ASTM G65 (viz. 58 to 62 IRHD) . The density of this polyurethane after curing was 1210+ 10 kg md3. Its glass transition temperature (T,) and specific heat capacity (C,) were determined by differential scanning calorimetry at a heating rate of lOKmin_ . Tg was measured as - 23.8 C. The specific heat capacity was 1790 J kg- K-i at 20 C androse linearly with temperature to 1960 J kg- K- at 80 C. In contrast, Tg for the chlorobutyl rubber was measured as - 55.8 C, and its hardness was 64 IRHD at 25 C.

01 0 0.25 0.5 0.75 Sliding speed (m s-9 Fig.

-I

5. The effect of sliding speed on the wear rate of 1020 steel. A polyurethane rubber wheel of hardness 61 IRHD was used. The applied load was 98.1 N and the supplied sand feed rate was 1.0-1.1 g s- . A testing increment of 71.8 m ( 100 wheel revolutions) was used.

3. Experimental results 3.1. Influence of specimen load on wear rate To study the effect of normal load on wear rate a specimen of 1020 steel was used 25.4 X 28 X 10 mm in size. This was run-in against the polyurethane wheel with a hardness of 61 IRHD until the wear scar was 22 mm long. A linear sliding speed of 0.6 m s - was used, and the load was varied between 24.5 and 122.6 N. The wear rate was measured at sliding distance increments of 71.8 m (i.e. 100 wheel revolutions). The sand feed rate was maintained at a constant value of 1.0 g s-i. The results are shown in Fig. 4. The wear rate increased with load essentially linearly. 3.2. Influence of sliding speed on wear rate A series of tests was performed with run-in mild steel specimens on the polyurethane wheel with a hardness of 61 IRHD rotating at different speeds. The sand feed rate was maintained at approximately 1.1 g s _ and the applied load was kept constant at 98.1 N. The wear rate as a function of sliding speed is shown in Fig. 5. Over the range investigated

it increased with sliding speed to a maximum decreased at high speeds.

value and then

3.3. Influence of rubber hardness

Experiments were performed with four different polyurethane wheels with hardnesses in the range 46 to 84 IRHD. Mild steel specimens, all from the same strip of AISI 1020 steel, were run-in against the as-cast wheels to give equal length wear scars on each specimen. The wear rate was then determined against each wheel in terms of mass loss per unit sliding distance. The sliding distance increments were 359 m (i.e. 500 wheel revolutions). There was no noticeable difference in the surface finish of the wheels either before or after the tests. A difference in the curvature of the wear scars was noticed, the hardest wheel giving a wear scar with the greatest curvature. The results are shown in Fig. 6 and show an increase in specimen wear rate with rubber hardness. The chlorobutyl rubber wheel conforming to ASTM standard G65 was also used in these tests. The wear rate of mild steel against this wheel under the conditions described above was 599fQ kg m-i, considerably greater than the values measured with the polyurethane wheels.

236

A.N.J. Stevenson. I.M. Hutch&v/

Wear 195 (1996) 232-240

..._ ...._.
40 50 60 70 80 90

1.67 g s- (theory)

Rubber hardness (IRHD)

(a)

0.25

0.5

0.75

Fig. 6. The effect of rubber hardness on the wear rate of 1020 steel. The applied load was 98.1 N, the sliding speed was 0.6 m SC and the supplied sand feed rate was 1.0-I. 1 g s- I, A testing increment of 359 m (500 wheel revolutions) was used.

Sliding speed (m s- )

3.4. The specimen-abrasive packing fraction

contact: load per particle and

The layer of abrasive particles between the specimen and the wheel plays a central role in the wear process. The packing density of the particles in this layer is determined by two main factors: the rate at which the abrasive is fed onto the wheel and the wheel speed. Some particles will fall off the sides of the wheel as it contacts the specimen; this amount can be measured and the true abrasive flow rate past the specimen calculated. A quantity termed the packing fraction cf, can be used to describe the density of the abrasive layer. This can be defined by f = 1 - e, where e is the voidage andf is the volume occupied by the abrasive particles divided the total volume in the contact region. For example, a close-packed monolayer of spheres would fill 60.5% of space, and for this situation f= 0.605. The number of particles instantaneously in the contact zone (N,) can be estimated by measuring the net mass flow rate of abrasive past the specimen (IV) and making an assumption about the particle size distribution. If all the particles have the same size, then:

0.1 t (b)
01
I I 1

.. .. . .. 1 0.25 0.5 0.75

I .67 g s. (theory)

Sliding speed (m s-r) Fig. 7. The effect of sliding speed and abrasive feed rate on (a) load per particle and (b) packing fraction. The lines relate to the simple model discussed in the text.

W feed is less than W,im then it can be assumed that all the abrasive supplied to the contact region passes through it and W= Wf_,. If W,,, > W,im then the flow will be choked with W= W,,, and excess sand will fall off the wheel before it can be drawn into the contact. This is the situation usually used in testing. These assumptions, together with Eq. ( 1) and Eq. (2)) and taking W,im as the net abrasive flow rate at the slowest sliding speed, give rise to the theoretical lines plotted in Fig. 7. 3.5. Frictional force The coefficient of friction p (tangential drag force divided by normal load) on the steel specimen was recorded during the abrasion tests. p was generally low (typically = 0.3) at the start of each test but increased rapidly reaching a reasonably constant value of so.36 to 0.40 in approximately 20 revolutions, As the wheel rotated there was a small (no more than 5%) cyclic variation in the frictional force which was attributed to differences in texture around the wheel. The friction coefficient was independent of the applied load (in the range 245122.6 N) and was independent of the sliding speed (in the range 0.06-0.9 m s- ) ; 90-150 p,rn silica sand was used as the abrasive and 1020 steel was used for the specimens. p was also effectively independent of wheel hardness for the ASTM chlorobutyl rubber or the polyurethane rubber (in the range 46 to 84 IRHD). In additional tests in which a much harder specimen was used (a high chromium

(1)
where x is the contact length between the rubber wheel and the specimen, u is the relative sliding speed (m SC ) and mP is the mass of one particle. The packing fractionfis given by

.fJvdbp

(2)

where d is the particle diameter, b is the wheel width ( 12.7 mm in this case) and p is the particle material density. Fig. 7 shows experimental measurements and theoretical predictions for the effect of wheel speed on the packing fraction and the load per particle, at two different sand feed rates. It was found in practice that there was a limiting flow rate past the specimen that could be achieved, W,im. This then gives rise to a limiting packing fraction V;,,) and a limiting load per particle. If the abrasive feed rate from the hopper,

A.N.J. Stevenson, I.M. Hutchings / Wear 195 (1996) 232-240

231

white cast iron with H,, = 8.14 GPa) it was also observed that the value of /1 did not change. 3.6. Temperature rise during abrasion testing

Number of revolutions
0 100 200 3al 400

0.271 m s ,

A thermocouple was used to monitor the temperatures of steel specimens during abrasion testing, and was inserted into a small hole drilled into the back of the specimen, close to the centre of the wear scar. The temperature was recorded as a function of the number of revolutions of the wheel (readily converted to time or sliding distance) and the effects of load and sliding speed were studied. Fig. 8 shows the effects of load and sand feed rate on the specimen temperature rise at a constant sliding speed of 0.6 m s-l. In each case the temperature of the specimen rose rapidly at first; the rate of increase then decreased and in some cases a steady state was reached. The rate of increase and the final temperature increased with load at constant sand feed rate. Fig. 9 shows the effect of sliding speed on specimen temperature rise at a constant applied load and sand feed rate. The temperature rise generally increased with sliding speed. Fig. 10 shows the influence of the rubber hardness. Although there is little variation in the results, in general the harder wheels led to a greater temperature rise. In separate tests the temperature of the rubber wheel was monitored by pressing a fine thermocouple onto the wheel immediately at the end of each test, with the results shown in Fig. 11. The sand temperature was also monitored during the tests with a thermocouple held in the sand falling from the wheel after it had passed through the contact zone. The results are shown in Fig. 12. 3.7. Variation of rubber hardness with temperature Since both the specimen and the rubber wheel heat up during an abrasion test, the effect of temperature on the 2 L b
400 ! _,_...-.-:: . .._ .._... -----u ------_.-.-._.-. ___._,___. 78.5 98.1 127.5 > 78.5 98.1 127.5 ) 0.50 0.18 0.80 0.29 t jf

0.382 m s- 0.494 m s-1 0.6cn m 6-11 o 1 , , , , , 1 ----_-_ 0.713ms-t

50

loo

150

200

250

Sliding distance (m)

Fig. 9. The effect of sliding speed on the temperature of a 1020 steel specimen. A polyurethane rubber wheel of hardness 61 IRHD was used, the applied load was 98.1 N and the net sand feed rate was past the specimen was 0.9 g SC .
Number of revolutions

Rubber hardness (IRHD) -46 .._ -_.. . . .._.-... -_-_62 73 83

00 50 100 150 200 250 300 350

Sliding distance(m)

Fig. 10. The effect of rubber hardness on the temperature of a 1020 steel specimen in a wear test. The applied load was 98.1 N, the sliding speed was 0.6 m SC and the net sand feed rate past the specimen was 0.9 g s- .

B
.z
2 x C

15

Number

of revolutions
200 I 300 I _,_._.-.

0
70 60g .f 1 F 50 4. 30 _ _

1lM I

10

chlorobutyl

rubber

,A

_./

_./

_,_,_._.-.-~-.-~-.--

,,.,.,. ~:;~~~~~~~~~:~_~ .___. _____.............-.

l!Jl

200

300

400

500

:;... ,:=::.l.. ;;,.,:/ ; +;. M. , /C : 10 . 0 0 I 50 I 100

Number of revolutions

I 150

I 200 250

Fig. 11. The variation in rubber temperature during a test. A polyurethane rubber wheel of hardness 61 IRHD was used, the sliding speed was 0.6 m SC , the applied load was 98.1 N and the net sand feed rate past the specimen was 0.9 g SC .

Sliding distance (In)

Fig. 8. The variation in specimen temperature with sliding distance, applied load and net abrasive feed rate. A polyurethane rubber wheel of hardness 61 IRHD was used and the sliding speed was 0.6 m SC .

mechanical properties of the rubber was also investigated. The hardness of the polyurethane rubber with a room temperature hardness of 61 IRHD was measured as a function of temperature. The results are shown in Fig. 13. The chloro-

238

A.N.J. Stevenson, I.M. Hutchings / Wear 195 (I 996) 232-240

cblorobutyl fubber I 200

0 0 Fig. 12. The variation conditions as Fig. 11).

I I 300 400 500 Number of revolutions in sand temperature leaving the wheel (same test

I 100

cblorobutyl rubber

polyurethane rubber

20

30

40

50

60

70

Tempwahre ( C) Fig. 13. The effect of temperature on the hardness of a cast polyurethane rubber and a chlorobutyl (ASTM standard) rubber.

butyl rubber wheel conforming to ASTM G65 was also tested in this manner with the results shown in Fig. 13; this rubber showed much less change in hardness with temperature.

4. Discussion A linear relationship between wear rate and load was found (Fig. 4)) as predicted by simple models of abrasive wear (e.g. [ 12,131) . Ellis and Armstrong [ 141 have also reported that the wear rate is linearly proportional to load in similar rubber wheel abrasion tests except at very high loads, when the wear rate increases more strongly with load. Haworth [5] also noted that the wear rate at high loads was greater than that expected from a simple linear relationship, although he used only two loads (89 and 178 N) in his tests. Avery [6] reported work performed by Borik with wheels with Shore Durometer hardnesses of A42 and A66. A tool steel with hardness 65 HRc was used as the testspecimen and the load was varied from 3 1 to 237 N. For the harder wheel the relationship between wear rate and load was not linear, with the wear rate increasing less rapidly than the load. For the softer wheel Avery claimed that the wear rate actually decreased as the load was increased and attributed this to an increasing area of contact between the wheel and the specimen as the load was increased.

Wear rate might be expected to be independent of the sliding speed (Fig. 5). The simple theory used to generate the theoretical lines in Fig. 7 suggests that the product of load per particle and packing fraction should be a constant, and independent of sliding speed. This correlates with Avery s concept of opportunity vs. severity [ 61, which he used to explain the unimportance of the sand feed rate. The product of the load per particle and the packing fraction is indeed constant, approximately 4.2 mN, and independent of the sliding speed and the sand feed rate. There is however some discrepancy between the theory and the experimentally determined values, and this is greater at the higher sliding speeds. After descending the inclined feed chute the sand hits the wheel and must change its velocity (in both magnitude and direction) ; as the wheel speed increases this leads to a greater loss of sand at the point of transfer from chute to wheel. Since variations in the packing fraction and load per particle are self-compensating they cannot be used to explain the behaviour observed in Fig. 5. The friction coefficient was effectively constant and independent of sliding speed and it does not therefore seem likely that at low sliding speeds more rolling of the abrasive particles takes place. The observed dependence of wear rate on sliding speed is more probably accounted for by variation of the mechanical properties of the rubber with strain rate and temperature. As Fig. 9 shows, the specimen temperature rise, and therefore the rubber temperature rise, increases with sliding speed. The rubber will therefore soften more as the sliding speed increases and so a lower wear rate would be expected on the basis of the results in Fig. 6. At low sliding speeds, the strain rate associated with the indentation of the abrasives into the wheel surface will be lower than at higher sliding speeds, and as a result of this the modulus of the rubber will increase as the sliding speed increases. The maximum in the data in Fig. 5 may thus be explained in terms of the superposition of the effects of strain rate and temperature. The wear rate was found to increase with rubber hardness (Fig. 6)) as reported previously by both Avery [ 61 and Borik [ 71. Borik [ 71 presented results for a wide range of different materials tested with three wheels of different hardnesses, in the range A50 to A70 Shore Durometer and wrote the weight loss was found to be some exponential function of the rubber hardness. Fig. 6 shows similar results for polyurethane rubbers with hardnesses in the range 46 to 84 IRHD. Use of the ASTM chlorobutyl rubber wheel resulted in a much higher specimen wear rate than a polyurethane wheel with similar room temperature hardness. Part of this difference can be accounted for by the temperature sensitivity of the hardness of the two materials, since significant heating of the wheels occurred in the tests. Since the polyurethane had a glass transition temperature of = - 24 C, at room temperature it would have been in a transitional state in which conformational changes are slow and true rubber-like behaviour is observed only at low strain rates. The chlorobutyl rubber in contrast had a T, of = - 56 C and although its hardness (modulus) decreased with increasing temperature over the

A.N.J. Stevenson, I.M. Hutchings/

Wear 195 (1996) 232-240

239

temperature range studied the effect was smaller than for the polyurethane (Fig. 13). At 40 C, for example, a temperature which can be reached by the wheel at the end of a test (as shown in Fig. 1 1 ), the chlorobutyl wheel would have been significantly harder than the polyurethane, and would therefore have resulted in a significantly greater wear rate. It is not clear, however, that the effect of temperature alone can explain the whole difference in specimen wear rate between the rubbers shown in Fig. 6; it is probable that strain rate sensitivity also played a role. Nearly all the work done at the specimen-abrasive particle interface is converted into heat. At steady state themechanical work done ( pPv, where P is the applied normal load and u is the linear sliding speed) can be equated to thermal losses due to conduction, convection etc. from the wheel and specimen and the amount of heat energy input into the sand, which is given by W ATsandCsand (where W is the sand mass flow rate, A Sand is the temperature rise above ambient, and Csand is the specific heat capacity of the sand). Csandwas measured calorimetrically as 860 + 60 J kg- K- . From the data presented in Figs. 10-12, it is clear that once a steady state is reached most of the frictional work done is dissipated in heating up the sand which typically rises to = 23.5 C above ambient (for a sliding speed of 0.6 m s- and a normal load of 98.1 N). Under these conditions, for example, the thermal energy entering the sand amounted to = 84% of the total power input. There was a small cyclic variation in the temperature of the rubber wheel during each test which was detected by inserting a thermocouple just under the surface of one part of the wheel. As the thermocouple passed the specimen it rapidly rose in temperature by = 3 C and then cooled during the rest of the rotation. The mean wheel temperature increased progressively until thermal losses balanced the heat input. The steady state wheel and specimen temperature are determined by the frictional power input (pPu) and the heat losses, the main one of which is the thermal energy carried away by the sand flow. The results in Fig. 9 and Fig. 10 can thus be explained. Although steady state conditions were not reached in all cases within the duration of the test, the general trends are clear. The temperature rise increased with sliding speed (Fig. 9) and with applied load (Fig. 8) at constant sand feed rate. Since decreasing the sand feed rate reduces the amount of heat carried away, the specimen (and the rubber) will then reach higher temperatures (Fig. 8). ASTM standard G65 recommends frequent dressing of the wheel surface. In the course of this and other work a single polyurethane wheel with a hardness of 61 IRHD was used repeatedly over a period of two years with no obvious surface roughening or need for dressing. The condition of the wheel was monitored by performing wear tests on a standard material, a 3C-25Cr white cast iron. The wear rate of this material under nominally identical test conditions increased by 20% over the two years; this may be attributed to a change in wheel surface finish, but may also be due to a small change in the viscoelastic properties of the rubber as the wheel aged. All

the tests described in the present paper were performed over a much shorter time, so that the effects of wheel ageing can be ignored. For quantitative comparisons of wear resistance between materials, the use of a standard material as a control, as recommended in the ASTM method, is clearly advisable. The rubber wheel abrasion tester used in this work has some advantages over the ASTM design. The sand feed is better controlled, and offers some economy in the amount of abrasive used. Mounting the specimen horizontally allows the amount of abrasive that actually abrades the specimen to be accurately determined. The load per particle and the packing density of the abrasive within the contact region between the wheel and specimen can be measured. The design also incorporates a load cell to measure the friction force. With the test conditions commonly employed (load, 98.1 N; sliding speed, 0.6 m s- ) steady thermal conditions were usually reached in = 200 wheel revolutions. If this is to be regarded as an initial transient then long testing increments cannot be avoided, and the test conditions in this steady state should be reported. However there is no evidence that the wear rates of different materials change by different proportions when the wheel hardness is changed, and there is thus no reason why shorter test durations, sufficient to cause a measurable mass loss, cannot be used; this would have the advantage that several tests may be conducted on one specimen with negligible change in the length of the wear scar. Whatever test regime is chosen, it is important that after each test the apparatus is allowed to cool to room temperature before starting the next test.

5. Conclusions 1. For low carbon steel specimens the wear rate increased linearly with normal load, up to a maximum load of 122.6 N. 2. The rubber hardness decreased as the temperature was increased for both the polyurethane and chlorobutyl (ASTM standard) rubber wheels, but the effect was greater for polyurethane. 3. The specimen wear rate increased with the rubber hardness for wheels with hardnesses in the range 46 to 84 IRHD. A chlorobutyl rubber wheel gave a much higher wear rate than a polyurethane wheel of the same hardness. This was partly attributed to a difference in the change in rubber hardness with increasing temperature but strain rate sensitivity may also be important. 4. A maximum was seen in the dependence of wear rate on sliding speed. This was attributed to a balance between strain rate effects increasing the modulus of the rubber, and heating effects reducing the modulus. Maximum wear rate occurred at a sliding speed of approximately 0.5 m s-i. 5. The friction coefficient was found to be independent of sliding speed and applied load. It was also independent of the rubber hardness and rubber type.

240

A.N.J. Stevenson, I.M. Hutchings / Wear 195 (1996) 232-240

Temperature rises in the test could be large and sufficient to cause significant changes in the mechanical properties of the rubber. The main mechanism of heat loss was heating of the continual flow of fresh abrasive through the contact. Typically, this accounted for more than 80% of the total mechanical power input once steady state conditions had been reached. Polyurethane rubber wheels were easily produced in the laboratory by casting, and frequent dressing was not necessary to maintain stable experimental conditions.

Acknowledgements A.N.J. Stevenson thanks EPSRC and British Steel, Swinden Laboratories for financial support via a CASE studentship. The authors are grateful to Dr Charles Setterlield for conducting the DSC analysis and to J.S. Ellis, British Steel, for advice and helpful discussion.

[8] F. Borik, Using tests to define the influence of metallurgical variables on abrasion, Met. Eng. Q., 12 (1972) 33-39. [9] F. Borik, Testing for abrasive wear, in R.G. Bayer (ed.), Selection and Use of Wear Tests for Metals, ASTM STP 615, ASTM, Philadelphia, PA, 1976, pp. 30-44. [ IO] R.C.J. Tucker and A.E. Miller, Low stress abrasive and adhesive wear testing, in R.G. Bayer (ed.), Selection und Use of Wear Tests for Metals, ASTM STP 615, ASTM, Philadelphia, PA, 1976, pp. 68-90. [ 1 I] G. Huard et al., The effects of size and shape of abrasive particles on the measurement of wear rate using a dry sand rubber wheel test, in K.C. Ludema (ed.), Proc. Int. Con& on Wear of Materials, ASME, New York, 1987, pp. 689-699. [ 121 E. Rabinowicz, L.A. Dunn and P.G. Russell, A study of wear under three body conditions, Wear, 4 (1961) 345-355. [ 131 I.M. Hutchings, Tribology: Friction and Wear of Engineering Materials, Edward Arnold, London, 1992. [ 141 J.S. Ellis and B.M. Armstrong, Evaluation of Wear Resistant Materials for Use in Minerals Handling Systems, British Steel Rep. SL/CS/P/ 13/90/o, 1990.

Biographies A.N.J. Stevenson: graduated from Cambridge University in 1992 with a first degree in Materials Science and Metallurgy, and was awarded his PhD from the same university in 1996 for a study of the abrasive and erosive wear of weld-hardfacing alloys. He is currently working as a development engineer in the ceramics industry. I.M. Hutchings: graduated in 197 1 with a BA degree in physics from the University of Cambridge, and was awarded his PhD from Cambridge in 1975 for work on the erosion of ductile metals by solid particle impact. He has carried out research into many aspects of wear by erosion (including erosion-corrosion) and abrasion, and problems of impact, surface deformation and high strain-rate mechanical properties. Since 1977 he has been a staff member in the Department of Materials Science and Metallurgy, University of Cambridge. His current research interests include friction and wear, and in particular the relationship between material microstructure and wear rates and mechanisms, for all types of materials. He is the author of a textbook, Tribology: Friction and Wear of Engineering Materials, published in 1992, and was awarded the Tribology Silver Medal in 1994.

References
[ 1] ASTM G65-91, Standard test method for measuring abrasion using the dry sand/rubber wheel apparatus in Annual Book of ASTM Standards Volume 03.02, ASTM, Philadelphia, PA, pp. 247-259. [2] H.S. Avery, Classification and precision of abrasion tests in W.A. Glaeser. K.C. Ludema and S.K. Rhee (eds.), Proc. In?. Conf: on Wear of Materials, ASME, New York, 1977, pp. 148-157. [ 31 P.A. Swanson, Comparison of laboratory and field abrasion tests, in K.C. Ludema (ed.), Proc. Int. ConJ on Wear of Materials, ASME, New York, 1985, pp. 519-525. [4] P.A. Swanson, Comparison of laboratory abrasion tests and field tests, in A.W. Ruff and R.G. Bayer (eds.), Tribology; Wear Test Selection for Design and Application, ASTM STP 1199, ASTM, Philadelphia, PA, 1993, pp. 80-99. [5] R.D. Haworth, Jr., The abrasion resistance of metals, Trans. Am. Sot. Met., 41 (1949) 819-869. [6] H.S. Avery, An analysis ofthe rubber wheel abrasion test in S.K. Rhee, A.W. Ruff and K.C. Ludema (eds.), Proc. Inf. Co@ on Wear of Materials, ASME, New York, 1981, pp. 367-378. [7] F. Borik, Rubber wheel abrasion test, SAE Trans., 79 (1970) 21452154.

You might also like