Download pdf - Influence of Rail Surface Roughness Formed by Rail ...bunken.rtri.or.jp/PDF/cdroms1/0006/2006/20000606040901.pdf · Influence of Rail Surface Roughness Formed by Rail Grinding

216216216216216 QR of RTRI, Vol. 47, No. 4, Nov. 2006

PAPERPAPERPAPERPAPERPAPER

Influence of Rail Surface Roughness Formed by Rail GrindingInfluence of Rail Surface Roughness Formed by Rail GrindingInfluence of Rail Surface Roughness Formed by Rail GrindingInfluence of Rail Surface Roughness Formed by Rail GrindingInfluence of Rail Surface Roughness Formed by Rail Grindingon Rolling Contact Fatigueon Rolling Contact Fatigueon Rolling Contact Fatigueon Rolling Contact Fatigueon Rolling Contact Fatigue

1. Introduction1. Introduction1. Introduction1. Introduction1. Introduction

Preventative rail grinding is currently becoming verypopular in Japan as a way of removing the surface layerof rail rolling contact fatigue (RCF) damage caused byrepetitive trainloads. In addition, curative rail grindingis carried out to remove the longitudinal rail surface ir-regularities such as rail corrugations and rail welds. Suchrail grinding is contributing to extended rail service livesand reduced rolling noise and vibration.

On the other hand, elastic-plastic stress analysis ofasperity contact has revealed the high degree of contactstress on the wheel/rail interface in terms of surfaceroughness 1). For instance, in cases where von Mises stresssurpasses the shear yield stress of rail materials, plasticdeformation may take place and cause cracks due toratcheting and other factors. Focused on rail roughnessfrom the point of view of rail grinding work efficiency, anappropriately high level of roughness has some advan-tage on grinding speed and ground thickness per cycle.However, an initially high level of roughness formed onthe rail surface by rail grinding poses a problem with re-gard to high-frequency rolling noise and has the poten-tial to cause RCF damage, as mentioned above. On theother hand, such high levels of initial roughness are usu-ally reduced to normal due to repetitive trainloads whenin operation. This poses two interesting questions. Oneis how long it takes or how much accumulated passingtonnage is required to reduce the initial level of rough-ness. The other concerns the degree of RCF damage ac-cumulated from the initial level of roughness to the timewhen trainloads have reduced it to normal roughness. Itis hence essential to study the optimal initial roughnessformed by rail grinding to avoid rolling noise and/or RCF,

and to discover the most appropriate and efficient way ofcarrying out grinding work.

In this study, we carried out some experiments usinga twin-disc rolling contact machine 2) to investigate thevariation in accumulated passing tonnage needed to settledown the initial roughness formed by rail grinding. Inaddition, we analyzed the rail discs, focusing on the plas-tic flow of the surface layer using an optical microscopeand the crystal axis density using X-ray diffraction. Thisreport describes the experimental results obtained by thetwin-disc rolling contact machine and the results of met-allurgical analysis carried out on the rail discs.

2. Repeating rolling experiments2. Repeating rolling experiments2. Repeating rolling experiments2. Repeating rolling experiments2. Repeating rolling experiments

2.1 T2.1 T2.1 T2.1 T2.1 Test machineest machineest machineest machineest machine

Figure 1 shows the twin-disc rolling contact machineadopted in the experiments. The wheel disc was set asthe driving side and the rail disc as the following side.The wheel disc (diameter: 300 mm, thickness: 50 mm) wasmade from the same materials as an actual wheel, andthe rail disc (diameter: 170 mm, thickness: 15 mm) wascut out from JIS 60kg rail and formed. Upon carrying outthe experiments, the contact surface of the wheel disc waspolished with #80 abrasive papers, and an Rz, the maxi-mum height of roughness, of 10 µm to 40 µm were ap-plied onto the contact surface of the rail disc (Fig. 2(a)).The direction of the roughness groove was formed paral-lel to the axis direction (Fig. 2(b)) referencing the actualroughness of the grinding marks.

H. CHEN, Ph.DH. CHEN, Ph.DH. CHEN, Ph.DH. CHEN, Ph.DH. CHEN, Ph.DSenior Researcher,

Track Dynamics Laboratory, Railway Dynamics Division

M. ISHIDAM. ISHIDAM. ISHIDAM. ISHIDAM. ISHIDAGeneral Manager,

JR Affairs, Marketing and Business Development Division

Based upon stress analysis that indicated that surface roughness causes higher con-tact pressure on contact surfaces, the question arises about whether the initial surfaceroughness formed by rail grinding may speed up the onset of rail rolling contact fatigue(RCF). In order to clarify whether initial surface roughness has an adverse effect on railRCF, the authors carried out experiments by means of a twin-disc rolling contact machineand then investigated the experimental results from several perspectives, such as surfaceroughness, plastic flow, as well as the hardness and axis density of crystals beneath therail surface. This paper describes the details of the experiments, the variation of rough-ness on contact surfaces that accompanies repeated rolling contact and the influence ofthe initial surface roughness on RCF.

KeywordsKeywordsKeywordsKeywordsKeywords: rail grinding, surface roughness, rolling contact fatigue, RCF, inverse pole fig-ure

217217217217217QR of RTRI, Vol. 47, No. 4, Nov. 2006

Load cell (4 units)

motor

motor

Torque sensor

Rotaryencoder

Rail disc

Wheeldisc

Rotaryencoder

D.C.

D.C.

Rotary meter

Universal joint

E.C.B

Radialload

Surfaceroughness

6mm

5mm

15mm

Ø170mm

(a) Roughness made on rail disc (b) Dimensions and roughness orientation of rail disc

Fig. 1 TFig. 1 TFig. 1 TFig. 1 TFig. 1 Twin-disc rolling contact machinewin-disc rolling contact machinewin-disc rolling contact machinewin-disc rolling contact machinewin-disc rolling contact machine Fig. 2 Rail disc dimensionsFig. 2 Rail disc dimensionsFig. 2 Rail disc dimensionsFig. 2 Rail disc dimensionsFig. 2 Rail disc dimensions

TTTTTable 1 able 1 able 1 able 1 able 1 TTTTTest arrangementsest arrangementsest arrangementsest arrangementsest arrangements

Test discsRoughness Rz

（μ m）

Contact pressure（MPa）

Rolling speed（km/h）

Slip ratio（％）

Repetitive cycles（cycles）

Passing tonnage（equivalent）（MGT）

Temperature & Humidity（℃）, 　　　（%）

　

　

750 30 0 136731 137 24, 　　＜ 30

No. 1-①　 32.98-②　 28.26-③　 16.30-④　 9.61

　

　

750 30 0 84463 84 25, 　　＜ 25

No. 2-①　 40.49-②　 39.97-③　 17.85-④　 9.02

　

　

750 30 0.2 138335 138 23, 　　＜ 32

No. 3-①　 31.63-②　 31.56-③　 9.92-④　 8.98

0

5

10

15

20

25

30

35

40

45

Repetitive cycles 103

Max

imum

rou

ghne

ss R

z ,

m

No. 1- 32.98 m

No. 1- 28.26 m

No. 1- 16.30 m

No. 1- 9.61 m

0 1 11.9 47.6 138.8

Initial roughness

0

5

15

15

20

25

30

35

40

45


Max

imum

rou

ghne

ss R

z ,

m

No. 2- 40.49 m

No. 2- 39.97 m

No. 2- 17.85 m

No. 2- 9.02 m

0 12.4 36.5 84.4

Initial roughness

0

5

10

15

20

25

30

35

40

45


Max

imum

rou

ghne

ss R

z ,

m

No. 3- 31.63 m

No. 3- 31.56 m No. 3- 9.92 m

No. 3- 8.98 m

Initial roughness

0 12.3 37.6 87.9 138.3

(a) Rail disc No. 1 (b) Rail disc No. 2

(c) Rail disc No. 3

Fig. 3 VFig. 3 VFig. 3 VFig. 3 VFig. 3 Variation of roughness with repetitive cycles on rail disc Nos.1 to 3ariation of roughness with repetitive cycles on rail disc Nos.1 to 3ariation of roughness with repetitive cycles on rail disc Nos.1 to 3ariation of roughness with repetitive cycles on rail disc Nos.1 to 3ariation of roughness with repetitive cycles on rail disc Nos.1 to 3


0

10

20

30

40

-20 0 20 40 60 80 100 120 140

No. 1- Slip ratio 0.0%



Max

imum

rou

ghne

ss R

z,

m


Repetitive cycles 12.3 103


No. 3-No. 3- No. 3-No. 3-

No. 3-No. 3- No. 3-No. 3-



No. 1-No. 1- No. 1-No. 1-

No. 1-No. 1- No. 1-No. 1-

(a) Rail disc No. 1 (b) Rail disc No. 3

2.2 Experimental arrangements2.2 Experimental arrangements2.2 Experimental arrangements2.2 Experimental arrangements2.2 Experimental arrangements

Table 1 describes the test arrangements. Nos. 1 to 3denote the rail disc numbers. ① to ④ show the location ofroughness (Rz) applied to a rail disc (Fig. 2(a)). The accu-mulated passing tonnage was calculated taking into con-sideration of the vertical axle load corresponding to anaxle load of 100kN. A constant maximum Hertzian pres-sure of 750 MPa was adopted, taking into account thecontact between a worn wheel and rail. In addition, a no-slip case and a slip (slip ratio 0.2%) were applied.

During the experiments, the roughness and hardnessof the contact surface were measured at 11,900 repetitivecycles corresponding to the number of wheel passes a dayon a revenue line of an annual 40 million gross tonnage(MGT: a parameter to define track damage caused bytrainloads, which is calculated as axle load multiplied bythe number of wheel passes) 3×11,900, 7×11,900 and11×11,900, respectively. After finishing the experiments,the plastic flow of the contact surface, its hardness, andcrystal axis density of crystals beneath the surface weremeasured using an optical microscope, hardness tester,and X-ray diffractometer.

3. Experimental results and discussions3. Experimental results and discussions3. Experimental results and discussions3. Experimental results and discussions3. Experimental results and discussions

3.1 V3.1 V3.1 V3.1 V3.1 Variation of roughness with repetitive cyclesariation of roughness with repetitive cyclesariation of roughness with repetitive cyclesariation of roughness with repetitive cyclesariation of roughness with repetitive cycles

Figure 3 indicates the variation of roughness withrepetitive cycles in terms of rail disc Nos.1 to 3. It can beseen that Rz decreased gradually in response to the in-crease of repetitive cycles. In addition, even the largestinitial roughness disc settled down to almost the samelevel of roughness as other smaller initial roughness discsafter 80,000 repetitive cycles.

Figure 4 shows a variation of roughness comparisonbetween rail discs Nos.1 to 3 with repetitive cycles. Witha slip ratio of 0.2%, rail disc No. 3 showed the largestvariation of roughness, which was reduced earlier thanthe rail discs without slip. The initial roughness reduc-tion trends in this study were roughly consistent withthe findings obtained so far on tracks in operational ser-vice.

In order to confirm the variation of roughness, Fig. 5shows the surface status of rail discs Nos.1 and 3 after11,900 repetitive cycles and 11×11,900 repetitive cycles.It is clear that the roughness of rail disc No.3 with slipreduced and almost vanished even at 11,900 cycles, whichwas earlier than that of No. 1 without a slip.

3.2 Effect of initial roughness on plastic flow3.2 Effect of initial roughness on plastic flow3.2 Effect of initial roughness on plastic flow3.2 Effect of initial roughness on plastic flow3.2 Effect of initial roughness on plastic flow

The plastic flow of rail discs No. 1- ① and No. 1- ④ ,which had different initial roughness but the same re-petitive cycles without slip, are shown in Fig. 6(a), thatof rail discs No. 1- ① and No. 2- ① , which had differentrepetitive cycles, are shown in Fig. 6(b). Furthermore, thatof rail disc No. 3- ① and No. 3- ④ , which had differentinitial roughness but the same repetitive cycles with slip,are shown in Fig. 6(c). The scale for all micro photos wasset at 400 times. In the case of rail disc Nos.1 and 2 with-out slip (Figs. 6(a) and 6(b)), almost no plastic flow wasidentified, irrespective of initial roughness. In the case of

Fig. 4 VFig. 4 VFig. 4 VFig. 4 VFig. 4 Variation of roughness comparison between railariation of roughness comparison between railariation of roughness comparison between railariation of roughness comparison between railariation of roughness comparison between raildisc Nos.1 to 3disc Nos.1 to 3disc Nos.1 to 3disc Nos.1 to 3disc Nos.1 to 3

Fig. 5 Investigation of surface status of rail disc Nos.1 and 3 with repetitive cyclesFig. 5 Investigation of surface status of rail disc Nos.1 and 3 with repetitive cyclesFig. 5 Investigation of surface status of rail disc Nos.1 and 3 with repetitive cyclesFig. 5 Investigation of surface status of rail disc Nos.1 and 3 with repetitive cyclesFig. 5 Investigation of surface status of rail disc Nos.1 and 3 with repetitive cycles


10m

No.1-No.1- No. 1-No. 1-RzRz 9.619.61 mRzRz 32.9832.98 m

No. 1-No. 1- No. 2-No. 2-Repetitive cyclesRepetitive cycles 138.8 138.8 10103 Repetitive cyclesRepetitive cycles 84.4 84.4 10103

No. 3-No. 3-No. 3-No. 3-RzRz 31.6331.63 m RzRz 8.988.98 m

(a) Optical micrographs of rail disc No. 1 (different initial roughness)

(b) Optical micrographs of rail disc Nos. 1 and 2 (different repetitive cycles)

(c) Optical micrographs of rail disc No. 3 (different initial roughness)

10m

10m

10m

10m

10m

Fig. 6 Investigation of plastic flow on rail disc Nos. 1 to 3Fig. 6 Investigation of plastic flow on rail disc Nos. 1 to 3Fig. 6 Investigation of plastic flow on rail disc Nos. 1 to 3Fig. 6 Investigation of plastic flow on rail disc Nos. 1 to 3Fig. 6 Investigation of plastic flow on rail disc Nos. 1 to 3

rail disc No.3 with a slip ratio of 0.2% (Fig. 6(c)), a slightplastic flow was apparent in their very thin layers. How-ever, the difference between various levels of initial rough-ness was not clear. From these metallurgical inspectionsof the rail discs on which repetitive loads of up to 130,000cycles were applied, the initial roughness formed by therail grinding is considered to have less effect on plasticflow.

3.3 Effect of initial roughness on hardness3.3 Effect of initial roughness on hardness3.3 Effect of initial roughness on hardness3.3 Effect of initial roughness on hardness3.3 Effect of initial roughness on hardness

We measured rail disc surface hardness under somerepetitive cycles during the experiments and measuredthe hardness of the rail discs beneath the surface afterthe experiments had been completed. A Shore-type hard-

ness tester was used in measuring the surface hardnessof the rail discs and a Knoop-type hardness (HK) testerused for measuring the hardness beneath the surface. Inadvance of the measurements with the Knoop-type hard-ness tester, it was calibrated with a standard test speci-men of 300 Hv (Hv: Vickers hardness) and 0.245N selectedas its applied load. The measurement depths decided uponwere at intervals of 20 µm up to 600 µm and intervals of0.1mm up to 1mm.

Figure 7 indicates the surface hardness Hv of rail discNo. 3 measured before, during and after the experiments.As shown in the figure, the surface hardness increasesfrom the beginning to 80,000 repetitive cycles. However,the surface hardness at 130,000 repetitive cycles is lessthan that at 80,000 cycles. The reason is related to the


precision of the hardness tester or the activation of ma-terials during repetitive rolling contact cycles.

Figure 8 depicts the variation of hardness beneaththe surface of rail disc No. 1-①, No. 1-④, No. 3-① and anew disc. Each datum plotted in the figures is the aver-age of three data measured at the same depth locationlaterally 2 mm apart. The hardness of rail disc No. 1-④with little initial roughness appears less than that of No.1-① as a whole, but the difference between the two lev-els of roughness is not clearly identified due to some datadispersion. The hardness of rail disc No. 3-① with slip is

250

260

270

280

290

300

0 20 40 60 80 100 120 140

No. 3- Initial roughness 32.98 m




Har

dnes

s, H

v


Fig. 7 Surface hardness variation of rail disc No. 3Fig. 7 Surface hardness variation of rail disc No. 3Fig. 7 Surface hardness variation of rail disc No. 3Fig. 7 Surface hardness variation of rail disc No. 3Fig. 7 Surface hardness variation of rail disc No. 3

Fig. 8 VFig. 8 VFig. 8 VFig. 8 VFig. 8 Variation of hardness beneath the surface of rail disc Nos. 1 and 3, plus new rail discariation of hardness beneath the surface of rail disc Nos. 1 and 3, plus new rail discariation of hardness beneath the surface of rail disc Nos. 1 and 3, plus new rail discariation of hardness beneath the surface of rail disc Nos. 1 and 3, plus new rail discariation of hardness beneath the surface of rail disc Nos. 1 and 3, plus new rail disc

slightly greater than that of rail disc No.1-① without slipin the case of the same roughness level. This agrees withthe findings obtained so far that indicate that the strainhardening of the surface to which slip is appliedprogresses more than that with no slip because a shearforce was applied. On the other hand, the hardness ofrail discs to which repetitive loads have been applied isgreater than that of a new rail disc, irrespective of slipbecause of strain hardening.

3.4 Effect of initial roughness on crystal axis density3.4 Effect of initial roughness on crystal axis density3.4 Effect of initial roughness on crystal axis density3.4 Effect of initial roughness on crystal axis density3.4 Effect of initial roughness on crystal axis density

The crystal axis density was measured using the X-ray diffraction of inverse pole figure method 3) to investi-gate the influence of initial roughness on RCF damage.An X-ray diffraction of inverse pole figure measurementgives information on the orientation of constituent crys-tals, from which it is considered possible to evaluate thedegree of RCF. The ratio of the X-ray diffraction inten-sity of a test piece to the intensity from the standardmaterial that has a random orientation is defined as thecrystallographic axis density. An axis density of 1 meansa condition in which there is no preferred orientation withrespect to the crystallographic axis.

In the measurements, Mo was used as the X-ray ori-gin at 60 kV and 200 mA and the target of crystallographicaxis was 222 since it is recognized as one of the potentialparameters to evaluate the degree of RCF damage accu-

Fig. 9 VFig. 9 VFig. 9 VFig. 9 VFig. 9 Variation of crystal axis density beneath the surface of rail disc No. 1ariation of crystal axis density beneath the surface of rail disc No. 1ariation of crystal axis density beneath the surface of rail disc No. 1ariation of crystal axis density beneath the surface of rail disc No. 1ariation of crystal axis density beneath the surface of rail disc No. 1

0.0

1.0

2.0

3.0

4.0

5.0

0 500 1000 1500 2000

Depth from surface, m

Axi

s de

nsity



222222

200

300

400

500

600

700

0 200 400 600 800 1000




New rail disc

Har

dnes

s, H

K

Depth from surface, m


mulated in rail surface layers in operational service 4).The inverse pole figure of the contact surface at a depthof 10 µm was measured first and later the measurementswere continued up to the depth of 2 mm with electro-so-lution polishing. Figure 9 shows an example of the mea-sured results of rail disc No.1 for crystallographic axis222. The axis density value in the very thin layer isgreater than 1, so that the order of the crystal plane isslightly recognized. However, the difference between vari-ous initial degrees of roughness is apparently unidenti-fied.

In the measured results of rails ground on tracks inoperational service 5), the axis density of the thin layer ofthe rail surface, where 26 MGT of accumulated passingtonnage were applied after grinding, was almost equiva-lent to 1. From the findings obtained by the actual railsamples 5), it may suggest that the minor damaged sur-face layer indicated in Fig. 9 would be removed while in-creasing repetitive loadings, and the axis density of thethin layer corresponding to 1.4 MGT of accumulated pass-ing tonnage in this study would be less or closer to 1 byincreasing repetitive loading cycles. In-depth studies intothis are anticipated. However, the possibility of minoreffects of initial surface roughness, grinding marks,formed by current rail grinding work on RCF damage wasidentified from the results of this study.

4. Conclusions4. Conclusions4. Conclusions4. Conclusions4. Conclusions

It was apparent that the surface roughness formedby rail grinding work was decreased and reduced to somenormal roughness level value after almost identical re-petitive loading cycles, irrespective of the initial rough-ness levels. In addition, based upon the investigated re-sults on the hardness, plastic flow, and rail disc crystal

axis density, after the initial roughness had settled downto almost the same level of normal rail roughness, thedifference of initial roughness was apparently unrecog-nized. Hence, the grinding marks formed by the currentrail grinding work have less potential to generate RCFcracks and make the cracks progressive. RCF damagecracks may mainly be generated after the initial rough-ness of the grinding mark has been reduced to a normalrail surface level.

5. Acknowledgements5. Acknowledgements5. Acknowledgements5. Acknowledgements5. Acknowledgements

The authors would like to express their sincere ap-preciation to Mr. Yukio Satoh and Dr. Kengo Iwafuchi fortheir very valuable discussions and help with the X-raydiffraction of inverse pole figure method.

ReferencesReferencesReferencesReferencesReferences

1) Kapoor, A., Frank, F. J., Wong, S. K. Ishida, M.: “Sur-face Roughness and Plastic Flow in Rail Wheel Con-tact,” Wear 253, pp. 257-264, 2002.

2) Ishida, M., Ban, T. and Moto, T.: “Development ofHigh-Precision Rolling Contact Machine and a FewInitial Test Results,” Proc. J-rail ’98, Japanese Soci-ety of Electrical Engineers, pp. 301-302 (in Japanese).

3) Nagashima, S.: “Texture Structure,” Maruzen, 1984.4) Satoh, Y. and Iwafuchi, K.: “Crystal Orientation

Analysis of Running Surface of Rail Damaged by Roll-ing Contact,” Wear 258, pp. 1126-1134, 2005.

5) Aabe, T. et al.: “Rolling Contact Fatigue Layer of RailUsed in Narrow Gauge Line,” Proc. J-rail ’04, Japa-nese Society of Electrical Engineers, pp. 53-54 (inJapanese).