Effect of interlamellar spacing on the wear resistance of eutectoid steels under rolling-sliding conditions

Wear, 135 (1990)

369 - 389

369

EFFECT OF INTERLAMELLAR SPACING ON THE WEAR RESISTANCE OF EUTECTOID STEELS UNDER ROLLING-SLIDING CONDITIONS* P. CLAYTON and D. DANKS Oregon Graduate

Center, Beauerton,

OR 97006-l

999 (U.S.A.)

Summary

The influence of the mean true interlamellar pearlite spacing on the wear behavior of eutectoid steels under rolling-sliding line contact has been investigated. A spacing range of 118 - 450 nm was achieved by carrying out isothermal heat treatments on two medium carbon steels, one a plain carbon steel and the other containing small additions of chromium and molybdenum. The relationship between wear rate and interlamellar spacing (S) was not affected by the small compositional differences but did depend on hertzian contact pressure. A progressive pattern of behavior was observed in which the wear rate was proportional to So.47 at 1220 N mme2 and to 5’1.33 at 500 N mm-2. The mechanism of wear generated by the high slide-roll ratio used in these experiments, the nature and influence of the initial breakin period, and the reproducibility of test data are also addressed.

1. Introduction

The work reported in this paper is part of a research program with several goals. On one level it can be viewed as an applied research project for the railroad industry. The aim is to generate meaningful rail and wheel steel wear data using a simple laboratory test, thus obviating the need to conduct all experiments in the field or in complex rigs employing full-size or scaled model specimens. Wear data for existing steels are required for models of track deterioration which are being developed to assist the industry in its much needed efforts to reduce costs and improve competitiveness [l]. Rail wear is one of the major rail deterioration processes, and rail costs are a significant factor in maintenance budgets. It is important, therefore, to obtain the best economic performance from the range of steel rails currently available. It is also necessary to evaluate the potential of new developments and address the questions of how close the present heat-treated eutectoid steels are to the limits of wear resistance for pearlitic steels and what benefits can be derived from bainitic microstructures. *Paper presented at the International U.S.A., April 8 - 14,1989

Conference on Wear of Materials, Denver, CO,

0043-1648/90/$3.50

0 Elsevier Sequoia/Printed

in The Netherlands

370

Implicit in the program, then, is the need to understand the relationship between wear behavior and microstructure. However, this represents a separate goal and is being pursued beyond the immediate needs of the railroad program because it is seen as an important element in building an understanding of wear. In particular, the metallurgist needs some systematic and clear relations to enable wear properties of materials to be introduced to students at a less advanced level than at present. Another element of the work which is being amplified beyond the immediate needs of the applied research program is the nature of the wear processes that occur in rails and in laboratory tests under rolling-sliding contact. Understanding the basic wear mechanisms and relating them to existing ideas on wear processes is seen as the cornerstone of fundamental wear research, and it is intended that this work will make a useful contribution to that effort. The investigation so far has involved testing old standard carbon rail steel X21, Table 1, over a wide range of contact pressures and slide-roll ratios in an Amsler wear testing machine [2]. All three types of wear identified by Bolton and Clayton [3] were observed, and by comparing the worn surfaces of the specimens with those of the gauge face of rails removed from a curve at the Facility for Accelerated Testing (FAST) in Pueblo, Colorado, it was shown that the highest slide-roll ratios generated the type of wear that occurs in the field. The second stage of the work focused on confirming the relevance of the laboratory test. Four different rail steels tested at FAST were removed TABLE

1

Chemical

composition

Old standard carbon X21 FAST

of steels c

Cr

MO

Mn

Si

S

P

Cu

Ni

V

0.75

0.02

0.02

0.98

0.25

0.03

0.04

0.07

0.09

0.004

0.72

0.02

0.01

0.97

0.21

0.02

0.005

0.14

0.06

0.002

0.67 0.73 0.71

0.82 0.70 0.55

0.02 0.24 0.21

0.93 0.67 0.61

0.61 0.27 0.29

0.02 0.02 0.02

0.005 0.005 0.005

0.04 0.02 0.25

0.06 0.02 0.10

0.14 0.002 0.002

0.63

0.14

0.05

0.88

0.17

0.01

0.005

0.29

0.13

0.002

0.71

0.57

0.21

0.88

0.41

0.02

0.005

0.26

0.10

0.002

0.77 0.70

0.08 0.02

0.04 0.001

0.66 0.72

0.33 0.53

0.04 0.02

0.03 0.03

0.08 0.05

0.08 0.03

0.005 0.005

rails

Old standard carbon X29 MnSiCrV CrMo I CrMo II Heat-treated Improved standard carbon A CrMo B Wheels Wl w2 _---

---_---_

371 TABLE 2 Wear data for four rail steels Rail steel

Old standard carbon X29 MnSiCrV CrMo I CrMo II

FAST

Laboratory

tests at PO = 1220 N mme2

Average Relative wear rate (mm MGT-’ )” zi:ance

Wear rate

Relative wear resistance

Average wear rate (X10’ mm rev-’ )

Relative wear resistance

ugly-*-1 rolled)

0.226

1.00

1354

1 .oo

538

1.00

0.173 0.112 0.127

1.31 2.02 1.78

729 634 654

1.86 2.14 2.07

413 319 323

1.30 1.69 1.67

MGT, million gross tons.

from the test site and laboratory specimens made from the rail sections. The results of the experiments, which are reported in detail elsewhere [4], are summarized in Table 2. Roller wear rates were evaluated in terms of weight loss and diameter loss. Diameter loss was determined from the average of diameter measurements taken at three circumferential locations in the center of the wear path. The correlation between the field and laboratory data is good, particularly when wear is assessed on the basis of diameter loss. This similarity in the relative wear resistances of the four steels has been taken as strong evidence that the laboratory test produces data which are relevant to the field. Further, it is considered that the relevance stems directly from matching the wear processes as closely as possible. In the present paper, the laboratory test procedure has been used to investigate the effect of pearlite interlamellar spacing on the wear resistance of pearlitic steels without grain boundary such as proeutectoid ferrite. To create a range of spacings, two rail steels of different chemical composition have been subjected to isothermal heat treatments. A total of 13 steel conditions, including the as-rolled ones, with a range of interlamellar spacings from 118 to 450 nm have been tested. The effect of spacing on wear rate has been determined at five different contact pressures. The effect of interlamellar spacing has received attention in previous investigations. Bhattacharya [5] showed that under sliding contact fine pearlite provides a better wear performance than intermediate or coarse pearlite, a qualitative result confirmed by Heller and Schweitzer [6] for rolling-sliding contact with a low slide-roll ratio of 0.7%. Clayton [7] investigated the sliding wear behavior of a number of pearlitic and ferriticpearlitic steels in which the microstructure was quantified in terms of mean free path in ferrite defined by h, = f,-d + f,,*2S, where f, and fp are the volume fractions of ferrite and pearlite, respectively, 2s is the mean uninterrupted ferrite path in pearlite, after Gensamer [8], S is the mean true interlamellar spacing, and d is the mean linear intercept in ferrite. The relationship derived between wear rate and this microstructural parameter suggests

372

that for fully pearlitic steels, the wear rate should be proportional to S”.s7. Shen [ 91 investigated the sliding wear of eutectoid and hypereutectoid steels in a cross cylinder machine. Although there was a trend of reduced wear rate with decreasing interlamellar spacing, the relationship was difficult to define even though the spacing range was wide, ranging from 100 to 476 nm. The large variability encountered in the wear rate data could account for this. A reasonably well-defined relationship between wear rate and interlamellar spacing is to be expected in view of the well-established links between flow stress and spacing [lo], the relationship between hardness and flow stress [ 111, and the influence of hardness on the wear behavior of pearlitic steels [6, 7, 121. While the central focus of the paper is the influence of microstructure on wear rate, it became necessary during the course of the work to consider two other aspects of wear behavior. These were the variability in wear rates measured under ostensibly identical experimental conditions and the nature of the break-in period and its influence on the subsequent steady-state wear rates.

2. Experimental

details

Rolling-sliding wear tests were carried out with an Amsler wear testing machine. The specimen configuration is shown in Fig. 1. The rail steel top rollers were 35.0 mm in diameter and the lower wheel steel rollers 45.2 mm to give a nominally constant slide-roll ratio of 35%. The increase in slip from the beginning to the end of a test as a result of decreasing roller diameters was not more than 10%. The maximum initial hertzian contact pressure PO can be controlled by adjusting the applied load and calculated from the relation [ 131

r-

35 mm

Roll

Steel

II

: 452mm

L

Wheel

Steel

1111

Fig. 1. Amsler

wear specimen

geometry.

373

PO = 0.418(LE/R)*'2 where E is Young’s modulus, l/R = l/R1 + 1/R2, with R, and R2 being the roller radii of the top and bottom rollers, and L is the load per unit contact length. In all tests, the contact length, that is the roller width at the contact plane, was 5 mm. The upper rollers were machined from bars taken longitudinally from rail heads and the bottom rollers were taken from the rims of unused wheels. The chemical compositions of the steels used are given in Table 1. The new standard carbon rail steel was used to produce the heattreated series A microstructures and the chromium-molybdenum steel was used to produce series B. The unlubricated wear tests were run with a jet of dried compressed air directed onto the rollers to minimize heating effects. The rollers were cleaned with soap and water and rinsed with methanol prior to testing. Wear was measured by periodically removing the rollers and determining both weight loss and changes in diameter. Usually, the first data points were taken soon after the surfaces had been observed to have broken in, but for one test the wear rate during the break-in period was established more carefully by making several stops in the early part of the test. Wear rates were calculated from the plots of weight or diameter loss against revolutions, which were usually essentially linear after break-in had occurred and are expressed in terms of weight loss or diameter loss per meter of rolling. The isothermal heat treatments were carried out by austenitizing at 925 “C a steel bar, 38 mm in diameter and about 27 cm in length, machined from a rail head, before dropping it into a salt bath, measuring 35 cm in depth and 18 cm in diameter, and holding it at one of a number of constant temperatures. The duration of the soaking in the bath was chosen to ensure complete transformation to pearlite. Determination of mean true interlamellar spacings was carried out using a scanning electron microscope following the method of Vander Voort and Roosz [14] employing a minimum of ten fields. The heat-treated series of rollers was tested against W2 wheel steel (Table 1) over a range of contact pressures from 500 to 1220 N mme2 at 35% slip and a lower roller rotational speed of 200 rev mini. Several repeat tests were carried out with an old standard carbon rail steel used in some previous experiments [2] running against W2 wheel steel (Table 1). These tests were run at only two contact pressures of 900 and 1220 N mmP2. All hardness measurements were taken using the appropriate Rockwell hardness scale and converted to Brine11 hardness numbers (BHN).

3.Results The interlamellar spacings and hardnesses resulting from the various isothermal heat treatments are given in Table 3 together with the microstructural characterization of the two as-received rail steels. In one case, 50 fields were taken, but with little change in the average spacing or standard deviation

374 TABLE

3

Heat treatments,

Sample designation

Old standard carbon X21 Old standard carbon X29

hardnesses,

and interlamellar

Isothermal treatment temperature C’C 0’))

Time (h)

(as received)

-

(as received) (as received) (as received)

-

A0 A31 A32 A33 A34 A35 A4 A55 A55 A55 A55 A6 A7 BO Bll B12 B13 B14 B15 B3 B4 B5 B6 B7 Wl w2

(as received) 604 (1120)

-

-

-

626 (1158) 669 (1237) -

1 21

a~ is standard

BHN

- -

MnSiCrV CrMo I CrMo II

0.5

-

1 3.5 0.75

3.6 1.5 46 24 2 -

Mean true interlamellar spacing (nm)

2oa

22

236

252

26

27

265

225

17

33 36 34

311 336 322

151 143 156

10 12 22

26 26 25 27 26 25 22 -

258 258 252 265 258 252 236

34 33 33 34 35 35 32 28 24

322 311 311 322 328 328 305 270 247

188 183 199 201 229 266 220 469 458 437 438 352 118

17 12 26 26 20 34 28 35 63 52 66 40 22 12 36 22 18 12 14 16 27 47 27 24 29 11

190

-

597 (1106) 659 (1218) 715 (1319) 704 (1300) 650 (1202) (as received) (as received)

Hardness -___

RC

(as received)

655 (1211) 196 (385) (as received) 485 (905)

spacings

190

195

20 22 32

205 228 236 301

-

166 141 150 122 122 140 170 231 391 472 315 214 162

deviation.

of those determined with only ten fields; ten fields were therefore considered adequate. A possible source of variation in spacing is along the length of the heattreated bar. Measurements on all the rollers from two bars, A3 and Bl, showed (Table 3) that for the Bl treatment, variation in spacing was not a problem, whereas the A3 treatment caused a 31% difference in spacing from one end of the bar to the other. Because variation in spacing did not appear to be a problem for the remaining heat treatments, spacings were determined

375

with a single roller while three hardness measurements were taken at different locations on the side of each individual roller. The variation in interlamellar spacings around the perimeter of a single A5 roller measured at four different positions was found to be only 7%. Some examples of the microstructures are shown in Fig. 2. The only dubious feature of the microstructures produced is the partially spheroidized cementite observed in B5, to a lesser extent in B6 and to a small degree in A5. B5 is estimated visually to contain 75% spheroidized carbide, B6 about 30% and A5 less than 10%. The results of the wear tests using the A and B series of specimens are given in Table 4 in terms of wear rate. All the wear rates in Table 4 were TABLE

4

Heat-treated

rail wear rates (pg m-l)

PO (N mm-*)

for A and B series

500

700

900

1080

1220

I)

33400 34900 49200 67200 39800 18100

66200 63500 70700 85900 80400 55800

101200 130100 149800 113900 66800

141800 140500 179400 212100 152800 102700

I)

8100 3100 3000 2300 5100 5700

16500 5700 4600 6900 7800 15300

14500 12600 10400 10200 19400

32200 20000 17400 14300 14100 30800

7000 5400 20100 38100 18000

17000 21700 49300 84200

43100 47700 94800 135100 84600 123500 142300

63500 90200 132100 164600 91300 137800 171700

10900 10500 14900 11600 4000 5200 9100

16100 18500 20500 14000 4300 8100 12700

A series rail A0 A3 A4 A5 A6 A7

6900 10500 14700 (Type

W2 wheel on A series rail A0 A3 A4 A5 A6 A7

900 900 1500 (Type

B series rail BO Bl B3 B4 B5 B6 B7

3200 (Type I) (Type I) 11000 9200 15600 14100

45500

84900 88300

W2 wheel on B series rail BO Bl B3 B4 B5 B6 B7

800

(Type 1) (Type 1) 1500 600 900 1300

1500 1800 4200 4100 1200 3800

4600 5600 7900 7500 3700 6800

(b)

(a)

(d)

(e)

k) Fig. 2. Typical (h) B6.

(h) microstructures:

(a) Bl,

(b) A3,

(c) Wl,

(d) A4,

(e) A5,

(f) A6,

(g) B5,

311

calculated from the slope of the linear regression line through the final part of the wear curve. In Type III wear, this is the wear rate after the break-in of the roller surfaces. Three tests resulted in Type I wear, Table 4, and these were excluded from the analysis. The wear rates are plotted against contact pressure in Figs. 3 - 6. As observed previously, the wheel steel wears at a lower rate than the rail steel at this slip level. For reasons not understood, it appears that the smaller roller always wears at the faster rate in this type of test. The plots are all non-linear with an accelerating wear rate as contact pressure is increased, in contrast to the linear plots presented previously [ 2, 31. From the regression data for these curves presented in Table 5, it is seen that the response of the wear rate to increasing contact pressure is not the same for all materials. The exponent is always greater than two, however, and since for line contact, contact pressure is proportional to the square of

500

700

900 Contact

Pressure

Fig. 3. Wear rate of top roller against

1100

I300

(N/mm2)

contact

pressure,

test series A.

200 I80 160 140 120 100 80 60 40 20 0 500

700 Contact

900 Pressure

Fig. 4. Wear rate of top roller against

1100

I300

(N/mm’)

contact

pressure,

test series B.

378

32 28 24

16

300

700

900

Contact

Pressure

1100

1300

(N/mm’)

Fig. 5. Wear rate of bottom roller against contact pressure, test series A.

:

16

z ,E

14

Fn” eb K ?az a z cr L : 3

12 IO 8 6 4 2 0 500

700

900 Contact

Pressure

II00

1300

(N/mm*)

Fig. 6. Wear rate of bottom roller against contact pressure, test series B.

the load, the relations indicate a slight acceleration in wear rate with increasing load. The effect of interlamellar spacing on wear rate at the five different contact pressures is presented in Figs. 7 - 11. The results for the two steel compositions cannot be separated, and the best fit line has been drawn using the combined data (Table 6). B5, which was consistently out of line with the other data points, was omitted from the regression. The volume fraction of carbide and its composition, thus, has little, if any, effect on the wear rate for a given interlamellar spacing. Table 6 shows that the relation between wear rate and interlamellar spacing is not the same at all contact pressures. Indeed, there is a clear progression in the relation such that the influence of interlamellar spacing on the wear rate is less significant at higher contact pressures than at lower ones. However, it has to be noted that the correlation coefficients are low.

379 TABLE

5

Regression

data for wear rate (pg m-l)

= a(PO)b, A and B series Wheel

Rail a

b

r

a

b

r

A0 A3 A4 A5 A6 A7

1.36 x 1O-3 1.22 x 10-S 7.43 x 10-s 6.7 x lo-* 1.35 x10-3 6.93 x lO@

2.599 3.28 3.049 2.095 2.618 2.977

0.99 0.97 0.96 0.94 0.99 0.94

7.24 x 1O-4 3.18 x lo-’ 9.84 x lo-’ 1.18 x 1OVj 7.48 x 1O-4 1.74 x 10-s

2.482 3.498 3,315 3.278 2.365 3.995

0.99 0.99 0.98 0.98 0.99 0.96

BO Bl B3 B4 B5 B6 B7

1.26 x 1O-6 3.02 x lo-” 3.45 x 10-e 6.2 x lo-’ 2.92 x 1O-4 2.34 x 1O-3 3.75 x lo.-4

3.456 5.015 3.435 3.073 2.764 2.538 2.823

0.98 0.99 0.99 0.99 0.98 0.98 0.99

1.39 x 10-7 4.3 x 10-g 2.13 x 1O-5 2.32 x 1O-4 3.70 x 10-4 3.68 x 1O-4 3.075 x lo-”

3.575 4.090 2.911 2.534 2.301 2.370 2.474

0.98 0.99 0.99 0.99 0.99 0.99 0.99

TABLE

6

Wear rate and pearlite

interlamellar

spacing

regression

[wear rate (pg m-l)

= a(S)b ]

PO

a

b

r

1220 1080 900 700 500

10431 2138 788 7.97 5.51

0.47 0.70 0.80 1.51 1.33

0.64 0.73 0.67 0.79 0.81

200 F

:-_i~~::

$ 0

60-

2 =

40-

k+ a"

200 0

cl

I 400

1 200

Pearlite lnterlamellar Spacing (nml Fig. 7. Wear rate against interlamellar

spacing,

1220

N mm-*

contact

pressure.

380

_

2001

Pearllte

lnterlamellor

Fig. 8. Wear rate against

interlamellar

200

Spacing

(nm)

spacing,

1080

I

T!

,80

i

160 -

;

140 -

_

0 0

N mmP2

contact

pressure.

I

A Series B Series

E 9 po_

100

--

I20 ;j

c;

0

[

LL

I

0 0 Peorlite

Fig. 9. Wear rate against

0 0

e

/

200

400

lnterlamellar Spacmg (nml interlamellar

spacing,

900

N mm-*

contact

pressure.

A Series B series

40-

0” z -^

[L

30

-

1

0 0 Pearltte

Fig.

I

200

10. Wear rate against

lnterlamellar

interlamellar

400 Spacing

(nm)

spacing,

700

N mm-*

contact

pressure.

381

Pearllte

lnterlamellar

Fig. 11. Wear rate against

1.4

0

0

Spacing

interlamellar

(nm)

spacing,

contact

pressure.

Roll Roller Wheel Roller

1.2

+g 1.0

500 N mm-

A -

9 0.8

-

;

0.6

-

F g

0.4

-

0.2

-

YI

: -I

Revolutions

Fig. 12. Weight loss during

the break-in

period

of type

behavior.

An example of a typical Type III wear curve is shown in Fig. 12. It was generated for one of the A7 tests from several measurements made in the initial break-in period. The dramatic change in wear rate from 7100 E.cgm-i to 66 800 pg m-l is associated with the break up of the roller surface. This can be understood by reference to Fig. 13 which shows the surface topography of the roller used to generate the curve in Fig. 12, at the point at which the surface is partially broken in. The reproducibility of Type III wear data as determined with the old standard carbon rail steel X21 used in previous investigations [2] is given in Table 7. The scatter is greater at the higher contact pressure. At 1220 N mmP2, two standard deviations for the rail steel represent 21% of the mean, whereas for the wheel steel they represent only lo%, and at 900 N mm-*, 8% and ll%, respectively. The duration of the break-in period can vary considerably (Table 7); however, there is no relation between the length of the break-in and the subsequent steady-state wear rate.

382

Fig. 13. Amsler TABLE

roller surface

after approximately

50 rev during break-in.

7

Wear rate reproducibility

Test no.

Rail wear rate (pg m-l rolled) -_I__-

Break-in interval

Wheel wear rate

(rev)

(1.18m-l

--

rolled)

1220

900

1200

900

1220

900

1 2 3 4 5 6 7

136400 162200 140700 153900 137600 141000 116100

63600 61700 66000 65600 66000 69600

2400 100 200 200 100 300 100

3600 4100 5800 320 600 200

14000 13300 13800 13100 12400 12600 12200

6900 6100 7100 7100 6900 7100

x (I

141100 14540

65420 2660

13100 692

6870 388

4. Discussion The problem of variability in wear test data for identical materials under the same contact conditions has recently been addressed by Wallbridge and Dowson [15]. They rightly point out that it is an important issue that receives far less attention than it deserves with very few examples of investigators carrying out repeat tests. In the current work, the laboratory tests have

383

a maximum scatter of only *21% of the arithmetic mean at the highest contact pressure used, compared with the several orders of magnitude reported by Wallbridge and Dowson [15] and the MO% recorded by Shen [9] on isothermally heat-treated 1078 steel tested under sliding conditions. The scatter in the rolling-sliding wear data appears satisfactory, therefore, and the plots of wear rate against contact pressure should give a reasonably accurate picture. The scatter in the FAST wear data has been quoted as +20% [16]. In view of these results and the close correlation obtained between the relative wear resistances determined in full-and small-scale tests, Table 2, the evidence suggests that the laboratory test procedure produces data that are relevant to field performance. This result has been achieved by taking care to ensure that the wear process is the same in the two cases. The greatest part of the wear damage suffered in the steady-state part of this process stems from the grooving caused by the interpenetration of the two surfaces in contact. This damage can occur only when the surfaces have become very rough as a result of the break-in period. This early part of the test produces an accumulation of protrusions by debris accretion, effectively creating a third body problem [ 171, and from the definitions of wear laid down by the OECD group for wear [X3], it seems reasonable to interpret the steady-state part of Type III wear as one of self-abrasion. The abrasive particles are relatively soft with a hardness not much higher than that of the roller surface. The presence of the built-up areas has a very significant effect on the sliding interaction distance in any given contact. From the length of the grooves in the roller surfaces, it is seen that this distance is about 4 mm, whereas the overall length of the contact patch in the rolling direction should only be 0.4 mm for a contact pressure of 1220 N mm -2. Examinations indicate that the majority of grooves were made in a single encounter as evidenced by the continuous nature of the scar. The effect of the potential sliding interaction distance on the mode of wear damage has been raised previously [3] and is considered to be a significant factor in determining the type of wear generated. This is an aspect of rolling-sliding wear which requires further attention in the context of understanding the wear mechanisms and the factors controlling them. It is apparent, then, that the role of the break-in period is to generate the necessary debris mounds to cause the abrasive process to occur. While this aspect of Type III wear has not been studied in any depth so far, it is clear from Fig. 13 that the end of the break-in period is associated with the break-up of the roller surface in a restricted location. This suggests that a single event might be responsible. Further, the variation in the duration of the break-in suggests that it is very chance dependent. However, the scatter in the subsequent steady-state wear rate is not related to the statistics of the break-in regime. Even though the scatter in the wear rate data is shown to be acceptable with respect to other data, it does, nonetheless, make it difficult to determine the exact mathematical relation between wear rate and microstructural

384

1220

I80

N/mm’ 2

0 0

160 140 120 100 80 60 40 20 0’

E t 0

I

200

I

I

400

Peorllte lnterlomellor Spocmg (rim) Fig. 14. Scatter bands for X21 standard carbon rail steel superimposed interlamellar spacing for series A and Bat 1220 N mm-’ and 900 N mm-*

on wear rate us. contact pressure.

parameters. By superimposing the scatter band for the repeat data from the old standard carbon rail steel on the wear rate US. interlamellar spacing plot for the A and B series at the appropriate spacing (Fig. 14) it is seen that the variability of the wear rate may account for much of the scatter in this relation. From the regression data for the wear rate against interlamellar spacing plots, it is seen that the relations are not the same at all the contact pressures. Furthermore, in determining the best fit, it is apparent that steel B5 consistently fell below the relation for the other steels. This also happens to a lesser extent for B6 and A5. Although further data would be required to establish any trend, these are the three steels in which spheroidized carbide is observed. The suggestion that this would enhance wear resistance gains little support from the literature [5, 19 - 221 since previous investigations with spheroidized structures have indicated poorer wear properties with respect to fully lamellar structures in both dry sliding and abrasive wear by hard abrasives. Figure 15 compares the wear rates based on dimensional loss with those taken from weight loss for the B series as a function of interlamellar spacing. Although the dimensional wear rate of B5 is still low for the spacing associated with it, the discrepancy is lower than for weight loss. It is possible, therefore, that part of the unexpectedly good wear resistance in weight loss terms results from deformation without fracture. Another possible reason for the apparent anomalous behavior could lie in the characterization of a mostly spheroidized microstructure by interlamellar spacing. Figure 16, however, shows that even in terms of wear rate against hardness the B5 results are inconsistent with those of the other steels. Since the unexpectedly good wear resistance of the B5 steel occurred at each contact pressure, it is very unlikely that the results can be explained on the basis of scatter. Within the experimental limitations of the current work, it would appear that the difference in chemical compositions of the two steels of

385 180

,

,

1

,

160

I

,

,

,

1

I

,

1

,

,

I

,

0

0

0

El5

60

6’

0

5 40

0 0

Weight Loss (pg/m) Diameter Loss (mm/m

x 10s)

20

1’1’1’1’1’1”“’

0

220

140

Pearlite

300

lnterlomellar

460

360

Spacing

(nm)

Fig. 15. Comparison of wear rates based on weight loss and diameter loss for test series B as a function of pearlite interlamellar spacing.

240 0

o

1220

D

700

N/mm2 N/mm2 1

Hardness

(BHN)

Fig. 16. Wear rate vs. hardness for both test series A and B at 1220 N mmP2 and 700 N mm -’ contact pressure.

0.08% carbon, 0.43% chromium, 0.06% molybdenum and the effects that these have on the volume fraction of carbides, the thickness of the carbides and the composition of the carbides have no detectable influence on the wear rate of pearlite. In view of the small differences in chemistry of the two steels, this might be expected. For example, according to Pickering [23] the difference in the cementite plate thicknesses would be expected to be less than 20%, which is of the same order as the scatter in the wear data. There is very little existing data with which to compare the relations derived for the effect of interlamellar spacing on wear rate. In pin-on-disk tests with a 100 kg load, Clayton found that the wear rate was proportional to S”e9’. This fits well with the current work where wear rate is proportional to S”*47’1*51depending on the contact pressure. More indirectly, it is possible to compare the effect of hardness on wear rate and the relation between hardness and interlamellar spacing with that

386

for other mechanical properties, such as flow stress. Mutton [24] using a large scale test rig found about a two times drop in wear rate with an increase in hardness from 250 to 350 HV30 (22 - 36 R,). Hodgson et al. [12] found a similar effect using a twin disk machine, and in the current work, the results at 1220 N mm--*, Fig. 16, also fit this pattern. In terms of the relation between interlamellar spacing and mechanical properties, it is found [lo] from extensive studies that yield strength shows a dependency on So.“‘.‘. In the current work, the relation between hardness and interlamellar spacing, shown in Fig. 17, reveals a dependency on S”.42. As far as can be ascertained, therefore, the relations derived in this work are reasonably consistent with previously published data, Figure 18 shows the relation between wear rate and interlamellar spacing at two contact pressures when other rail steel data are included with the A and B series. The validity of this plot depends on the effect of the type

100

200 Peorlbte

Fig. 17. Hardness

300 lnterlomellar

us. pearlite

interlamellar

, Interlomellar

(nm)

spacing

I

for A and B series steels.

1

400

200 Pearllte

500

400 Spacing

Spocmg

(nm)

Fig. 18. Wear rate against pearlite interlamellar spacing for several heat treatments at 1220 N mm-’ and 900 N mm-’ contact pressures.

rail chemistries

and

387

of wheel steel used in contact with the rail steel tested. While all the A and B series were run against the W2 wheel steel, Table 1, some of the other rail steel test data were derived from tests run against the softer wheel steel Wl. Since the overall relation between wear rate and interlamellar spacing is changed only slightly by the inclusion of the further rail steel data, it is possible that any effect the mating steel may have is overwhelmed by the overall scatter in the wear rate data. It is intended to extend the investigation by testing steels with interlamellar spacings down to the experimental limit of 70 nm. To do this, it might be necessary to employ a different experimental set up to carry out the isothermal heat treatments, perhaps along the lines of the work of Fegredo et al. [25]. However, given the scatter in the wear rate data and in the spacing measurements, it is unlikely that a more definitive relation can be determined without extensive test data.

5. Conclusions (1) The wear rate us. contact pressure curves for all of the 13 steel conditions tested were “best fitted” by a power law relation in which the index ranged from 2.1 to 5.0 indicating that wear rate is not a linear function of load. (2) The small compositional difference of 0.08% carbon, 0.43% chromium and 0.16% molybdenum between the two steels, isothermally heat treated to give a range of interlamellar spacings, did not affect the wear rate. (3) Seven repeat tests at a contact pressure of 1220 N mm-* produced a scatter in which two standard deviations represented no more than 21% of the mean, with the value being 11% for six tests at 900 N mm-*. (4) Even though the scatter in the wear data appears satisfactory compared with other recorded data, when combined with the variability in the microstructural characterization it is difficult to determine a definitive relation between wear resistance and microstructure. The correlation between the mean true interlamellar spacing and wear rate for a range of spacings of 118 to 450 nm was fitted by a power regression at each of the five contact pressures used, but the index progressed from 0.47 at 1220 N mm-* to 1.33 at 500 N mm-*. (5) The break-in period of Type III rolling-sliding wear initiates at a restricted location and from the variability of the duration of the period is apparently chance dependent. (6) The break-in generates a very rough surface dominated by mounds of material formed by debris accretion which increase the overall interaction distance in a single passage through the contact zone by an order of magnitude. The result is intermittent grooving of the roller surfaces in a wear process that can be viewed as one of self-induced mutual abrasion by soft abrasives.

388

Acknowledgments The authors would like to thank the Association of American Railroads for their financial support and Dr. R. K. Steele for his continued interest and encouragement.

References 1 W. B. Peterson, Presidential address to the American Railway Engineering Association, AREA Bulletin, 716, Vol. 89 (1988). 2 D. Danks and P. Clayton, Comparion of the wear process for eutectoid rail steels: field and laboratory tests, Wear, 120 (1987) 233 - 250. 3 P. J. Bolton and P. Clayton, Rolling-sliding wear damage in rail and tyre steels, Wear, 93 (1984) 145 - 165. 4 P. Clayton, D. Danks and R. K. Steele, Assessment of eutectoid rail steel wear performance using laboratory tests, Proc. Conf. Antiwear ‘88, September 20 22. The Royal Society, London, 1988. 5 S. Bhattacharya, Wear and friction in steel, aluminum and magnesium alloys I. Pearlitic and spheroidized steels, Wear, 61 (1980) 133 - 141. 6 W. Heller and R. Schweitzer, Hardness, microstructure and wear behavior of steel rails, Proc. 2nd Int. Heavy Haul Railways Conf., Colorado Springs, CO, 1982; American Association of Railroads, Chicago, IL, 1982, pp. 282 - 286. 7 P. Clayton, The relations between wear behavior and basic material properties for pearlitic steels, Wear, 60 (1980) 75 - 93. 8 M. Gensamer, E. B. Pearsall, W. S. Pellini and J. R. Low, The tensile properties of pearlite, bainite and spheroidite, Trans. ASM, 30 (1942) 983 - 1018. 9 D. Shen, Friction and wear of eutectoid and hypoeutectoid steels. In K. C. Ludema (ed.), Proc. Int. Conf. on Wear of Materials, Vancouver, April 14 - 18, 1985, American Society of Mechanical Engineers, New York, 1985. 10 D. J. Alexander and I. M. Bernstein, Microstructural Control of Flow and Fracture in P?arlitic Steels. In A. R. Mardo and J. I. Goldstein (eds.), Proc. Phase Transformations in Ferrous AZZoys, American Institute of Mechanical Engineers, New York, 1984. 11 D. Tabor, Hardness of solids, Rev. Phys. Technol., 1 (1970) 145 - 179. 12 W. H. Hodgson, J. K. Yates and R. R. Preston, The development of a second generation of alloy rail steels for heavy haul applications, Proc. 2nd Int. Heavy Haul Railway Conf., American Association of Railroads, Chicago, IL, 1982, pp. 207 - 215. 13 S. P. Timoshenko and J. N. Goodier, Theory of Elasticity. McGraw-Hill, New York, 1970, p. 419. 14 S. F. Vander Voort and A. Roosz, Measurement of the interlamellar spacing of pearlite, Metallogr., I7 (1984) 1 - 17. 15 M. C. Wallbridge and D. Dowson, Distribution of wear rate data and a statistical approach to sliding wear theory. In K. C. Ludema (ed.), Proc. Int. Conf. on Wear of Materials, Houston, 1987, American Society of Mechanical Engineers, New York, 1987. 16 R. Reiff, Defect and wear studies on premium and standard rails in rail wear experiment RME IV, O-35 MGT, Rep. TTC-030 (FAST-TN85), June, 1985, U.S. Department of Transporation. 17 M. Godet, The third body approach: a mechanical view of wear, Wear, 100 (1984) 437 18

- 452.

Glossary Wear

1980.

of Terms

Control

in Tribology, OECD. In M. B. Peterson American Society of Mechanical

Handbook,

and W. 0. Winer (eds.), Engineers, New York,

389

19 N. Prasad and S. D. Kulkarni, Relation between microstructural and abrasive wear of plain carbon steels, Wear, 63 (1980) 329 - 338. 20 D. M. Fegredo and J. Kalousek, The effect of spheroidization on the dry wear rates of standard C-Mn and Cr-Mo alloy rail steels. In K. C. Ludema (ed.), Proc. Znt. Conf. on Wear of Materials, Houston, 1987, American Society of Mechanical Engineers, New York, 1987. 21 M. A. Moore, The relationship between the abrasive wear resistance, hardness and microstructure of ferritic materials, Wear, 28 (1974) 59 - 68. 22 J. Larsen-Basse and K. G. Mathew, Influence of structure on the abrasion resistance of a 1040 steel, Wear, 14 (1966) 1461 - 1466. 23 F. B. Pickering, Physical Metallurgy and the Design of Steels, Applied Science Publishers, London, 1978, p. 96. 24 P. J. Mutton, The Influence of Microstructures on the Wear Behavior of Rail and Wheel Materials, Masters Thesis, University of Melbourne, Australia, 1985. 25 D. M. Fegredo, D. E. Parsons, W. A. Pollard and J. Nq-Yelim, The development of very hard and strong premium rails by controlled cooling procedures, Can. Met. Q., 22 (3) (1983)

385

- 395.