- Email: [email protected]
Metallurgical improvement of rail for the reduction of rail-wheel contact fatigue failures
319
Wear, 144 (1991) 319328
Metallurgical improvement of rail for the reduction rail-wheel contact fatigue failures
of
K. Sugino, H. Kageyama and C. Urashima Yawata R & D Ikzboratmy, Central R & D Bureau, Nippon Steel Corporation, KitaKyushu (Japan)
A. Kikuchi Plant Engineering
& Technology Bureau, Nippon Steel Corpmation,
Tokyo (Japan)
Abstract One of the important factors in prolongs the service life of rails is the prevention or retardation of the occurrence of shelling and transverse defects in heavy haul railways. This paper describes a rne~~~c~ ~vest~ation of rails which failed in service and gives a preliminary calcuhktion of the stress behaviour in order to acquire a deeper understanding of the effect of non-metallic inclusions, hardness level and its suitable distribution in the rail head from the standpoint of rail production.
1. Introduction Improvement of efficiency in railway transportation demands further increases in speed and the maximum axle load of trucks [ 11. In order to solve these technical challenges great efforts towards the improvement of rail steels have been made [Z-8]. One of the important problems is that of prolonging the service life of rails through the prevention or retardation of the occurrence of rolling contact fatigue defects, especially shelling and transverse defects (TDs), in heavy haul railways 19, lo]. In our experience it seems that the occurrence of both defects is controlled by almost identical metallurgical factors as far as their initiation is concerned. These metallurgical factors are considered to be non-metallic inclusions, hardness and residual stress in the rail head. In this paper we report a number of investigations which were conducted in order to obtain information which might lead to the improvement of these factors from the viewpoint of rail production.
2. Experimental procedures 2.1. ~et~~l~rg~~al
investigation
of shelling and TDs
From the standpoint of quality control of rail steel through production, the composition, size and volume of non-metallic inclusions around the 0043-1648/91/53.50
0 Elsevier Sequoia/Printed in The Netherlands
320
Di.stsnee
Fig. 1. Sampling position for non-medic
from
rail
head
surface
inclusion counting.
Fig. 2. Typical hardness patterns in rail head for stress calculation, high strength rail: -, case 1; ---, case 2; -.-, Std C. TABLE 1 Chemical composition and inclusion sizes of rail for running test at FAST Element
c
Si
Mn
P
S
Cr
Al
N
0
H
Amount (wt.%, *ppm)
0.81
0.22
0.89
0.014
0.001
0.02
0.006
33*
32*
0.6*
Total length of A&O, cluster: less than 650 Frn. Unit length of MnS inclusion: less #an 50 pm (average).
initiation point of shelling and TDs in failed raiis in a North American heavy haul railroad were investigated metaIlurgicaIIy in detail. Results by other researchers [ 1 l] showed that stringer-type ahunina clusters were mainly observed. In order to count ahunina clusters of over 100 pm, which is the limiting size of practical detection, a semiautomatic counting method was developed and used [ 121. Figure 1 shows the observed area of 20 mm by 10 mm beneath the gauge corner surface of the rail head where shehing and TDs often occur. The total length of ahunina clusters of 14 failed and 10 new raiIs which had been co~erci~y produced by several rail miIls was measured as an index from the s~dpo~t of quality control of rail production. 2.2. Eflect of rail hardness on shelling and TL; ocmrrme In order to clarify the effect of rail hardness level on shelling and TD occurrence, two classes of rails with low and high hardness were produced from the same melt of steel. One was as-rolled Std C rail with a hardness of 250 HB and the other was heat-treated high strength (hardness) rail with an average hardness of 345 HB Table 1 shows the chemical composition and inclusion sizes of the melt. The two classes of rails were sent to the Facility for Accelerated Service Testing (FAST) at the Transportation Test Center and submitted to the heavy haul running test (defect occurrence and growth (DOG) test).
321
2.3. meet
of hardness distribution
on intern& stress in rail head
From the standpointof rail production, it is importantto know the most suitable hardness ~bution in the rail head for the prevention of shelling and TD occurrence. On the other hand, it is clear that the initial hardness distribution in the rail head is changed by the repeated applied loads due to wheel passing, resulting in the induction of tensile stress in it [ 131. As a preliminary step, the stress behaviour in the rail head during several cycles of constant applied load on the gauge comer surface was calculated. Referring to an actual rail production, two types of hardness distribution in the rail head, cases 1 and 2, were chosen as shown in Pig. 2, because they are often seen practically. Prerequisites for the calculation are shown in Fig. 3.
3. Resnlts 3.1. Ntmmtetallic inclusions in failed rails Fourteen rail samples cut from the failed rails on the North American heavy haul rail track were metalhngicahy investigated. Much attention was paid to the relation between the initiation point of shelling and the non-
I.
: 132RE
Rail
Std 0 rail
132RE Length
: 5OOma (Case
2.
Wheel
: $844~.
3.
Load
: Normal 20 ton
Load
: 6 thes
repetition
Load center
: FEM : MARC.
Constitutive
8 nodes
analysis
Metallurgical
by van condition
properties
modulus
Poisson ratio
haxabedron
Isw :Elastic-plastic
Mises
Young’s
and
10 ton >
: by Ifertz’ equation
Calculation
Constraint
1 and 2 )
Load distribution
Software
5.
rail
Straight tread
( Tangential
4.
6.
High strength
: 2.06 X 10’ MPa : 0.3
Yield stress . Work hardening
: Experimental
Fig. 3. Prerequisites for internal stress calculation by finite element analysis.
322
metallic inclusions. From this investigation the results can be summarized as follows. (1) There was no correlation between the number of failed rails and the chemical compositions of aluminium, oxygen and other elements forming inclusions in the rail steels, as shown in Fig. 4. Although an increase in cleanliness of rail steel is important for the prevention or retardation of shelling and TD occurrence, it is not necessarily satisfied by a decrease in delusion-fo~g elements during the steel-making process. (2) Shelling and TDs were clearly confirmed to initiate at rather large stringer-type alumina clusters in two of the rail samples. Therefore the size of alumina clusters among the various inclusions should be kept as small as possible. (3) In spite of intensive observation by an optical microscope, no inclusions could be found at the vitiation point of shelling and TDs in the remainder of the failed rails, whereas by the quantitative counting it was found that they included a comparatively large number of alumina clusters, as shown in Fig. 5. Therefore new rails, plotted as E and S in the same range of total length of alumina clusters, would fail after a similar period in service when installed on the same track line. (4) In the case of the failed Std C rails it is also noted that the rails with shelling and TDs included alumina clusters of total length at least 2000 pm at a specific location and area beneath the gauge corner of the rail head. On the other hand, in the case of the high strength rails (G, H, X and Y) this value appears to be about 5000 pm. This implies that the detrimental effect of alumina clusters on shelling and TD occurrence depends greatly on the hardness of the rail head. This has certainly been confirmed in many heavy haul railways as an empirical fact [ 14 1.
m
Fatigue
Mn(x10-2wt%) Fig.
4. Chemical
damaged rails
S(X 10-3wt%) analysis
AE(XlO-“~3%)
of manganese,
sulphur,
0(X103wt@
aluminium and oxygen in rail samples.
323
0
LO
10
20
30
40
60
60
Fig. 5. Measurement of alumina clusters in raii heads: 0, fractured rails; a, crack-detected rails; 0, new rails.
(5) There was no evidence of a detrimental effect of MnS on shelling and TD occurrence. This appears to fit the general view in the rolling contact fatigue problem [ 151.
3.2. Eflect of hardness in rail head on shelliq and TD occurrence In order to confirm the sensitivity of non-metahic inclusions to shelling and TD occurrence in rails of different hardness levels, both Std C and high strength rails produced from the same melt have been submitted to the rail performance experiment (DOG test) at FAST with many kinds of rails produced by other rail miIls. According to Reiff [ 161 and a private communication of interim results at the passing tonnage of 120 MGT (million gross tons) at FAST, it was found that in the Std C rail, shelling and TDs occurred after 60-90 PvIGT, whereas in the high strength rail they had not occurred at least up to 120 MGT. The failed Std C rails were metallographically investigated in detail, but alumina clusters or other inchrsions could not be confirmed at or near the initiation point on the fracture surface of shelling and TDs except for one sample. It was also recognized that all samples investigated showed a similar mode of failure concerning the location and direction of the internal crack and the hardness change in the matrix near it. Referring to the above results, it is considered that:
324
(1) a decrease in volume and size of MnS inclusions scarcely affects the prevention or retardation of shelling and TD occurrence; (2) strengthening of the rnet~~gi~al structure surrounding inclusions, especially alumina inclusions, considerably reduces their detrimental effect; (3) in the case of the hardness level of the Std C rail, shelling and TDs are suggested to initiate from very small alumina inclusions which cannot be perfectly controlled by the present steel-refining process. Figure 6 shows typical examples of the failed Std C rails on FAST track. The failure initiated as shelling and grew to TDs. Almost all the shells investigated occurred near the boundary of the work-hardening zone owing to wheel load, as noted by Steele and Reiff [ 11, although it was not confirmed that there was a cyclic softening zone around the shell. This proves that plastic deformation in the rail head due to repeated wheel loads is closely related to the initiation of shelling and TDs. 3.4. Suitable hardness in rail head When a hardening process is applied to the rail head, the hardness is usually highest on the surface of the rail head and decreases inwards. Nippon Steel developed an in-line slack-quenching process for high strength rail production in 1987 [8]. This process has the characteristic of making the hardening zone in the rail head much deeper than normal reheating processes do. Therefore, from the standpoint of rail production, control of the hardness distribution in the rail head is very important as well as refinement of the molten steel to decrease non-meta~c inclusions, p~icul~ly alumina clusters. As a rough estimation of a suitable hardness distribution for the prevention or retardation of shelling and TD occurrence, the effect of two typical types of hardness pattern, cases 1 and 2, in the head of high strength rail on plastic deformation was calculated by a three-dimensional finite element analysis, although the prerequisites for the calculation are not necessarily
135 810 15 20 25 30 35 Distanoe from surface 6m>
Fig. 6. Fracture surface and hardness distributionof faiIed rails after running test at FAST: 0, 3-0833; A., 25-0301; 0, 25-1290.
325
well fitted for all the actual conditions. These hardness patterns are shown in Fig. 2 and have often been seen in actual production. l?igure 7 shows the effect of the hardness patterns on the e~~v~ent stsess (von M&es) with depth in the head of high strength rail after a repeated (four to six times) applied load of 20 t. Adopting the residual stress as an
-+I 6
0
e
G
_.+.. 7
400
200
__
07
0
2
4
6 Distance
8
--
10
from rail
12
head surface
14
16
18
20
(smr>
7. Effect of hardness diitribution on von Mies equivalent stress beneath load centre at gauge comer: 1, yield stress; 2, initial stress; 4 and 6, residual stress after fourth and sixth load cycle respectively; 3, 5, 7, stress under fourth, fifth and sixth applied load respectively. Fig.
326
index for plastic deformation, it is noted that the amount and depth of this in case 2 is larger than in case 1 in spite of the harder surface zone. Theoretical studies on the initiation and propagation of shelling and TDs in North America have been summarized by Steele [ 171. It is not necessarily easy to confirm which stress mainly controls shelling and TD occurrence in cases of actual failure. Following the precedent, an attempt was made to calculate the stress or stress range induced in a particular direction in the rail head under conditions close to actual wheel load and shape, size and mechanical properties of the rail, expecting that the stress should reveal a positive value and a maximum at the depth of shelling and TD occurrence. Figure 8 shows the residual stress in the direction of loading beneath the gauge corner surface under the conditions for calculation shown in Fig. 3. Assuming that the repetition of a positive stress affects the initiation of shelling, it becomes effective below a depth of about 6 mm at the rail gauge corner. The stress of case 1 is smaller than that of case 2. Therefore it is preferable to adopt the hardening process for preventing the rapid decrease in hardness with depth in the rail head. It is also important to keep the rail head surface as hard as possible. According to another interim analysis of failed 132RE high strength rails by a running test at an axle load of 27 t in a heavy haul railway, shelling and TDs occurred at alumina clusters of over 400 pm which were located at a depth of about 6-12 mm in the rail head after 500 MGT. This result
-400 0
2
4
6
3 Distance
10
12
from surface
14
16
18
20
(78x1
Residual stress in direction of loading (normal load 20 t, six repeated loadings): 0, case 1; @, case 2.
Fig. 8.
327
seems to be qualitatively in good agreement with the calculation described above. F’urther study should be done for a more quantitative analysis. 4. Conclusions Considering the results from these few experiments, we can set out the following basic guidelines for the production of rails for heavy haul railways: (1) adopt a steel-making process which gives as large a decrease in long ahunina clusters as possible; (2) keep the present cleanliness level of MnS inclusions; (3) adopt a hardening process which makes the hardness difference between the surface and the internal region of the rail head as small as possible, while maintaining high hardness on its surface. Acknowledgments
The authors would like to thank Dr. R. K. Steele and Dr. R. P. Reiff at AAR for an offer of rail samples with useful information. A portion of this work was done with Mr. H. W. Newell at Norfolk Southern Corporation. We also thank the Nippon Steel Corporation for permission to publish. References 1 R. K. Steele and R. P. Reitf, Rail: its behavior and relationship to total system wear, Proc. 2nd Int. Heavy Haul Railway Cor& 1982 82-HH-234, pp. 227-276. 2 S. Marich and P. Curcio, Development of high-strengthalloyed rail steel suitable for heavy duty application, ASl’!M Spec. Tech. PubL 644, 1976, pp. 167-211. 3 J. D. Young, United Kingdom development of rails rolled from continuously cast blooms, ASi% Spec. Tech. PubL 644, 1976, pp. 256-284. 4 Y. E. Smith and F. B. Fletcher, Alloy steels for high-strength,as-rolled rails, ASTM Spec. Tech. PubL 644, 1976, pp. 212-232. 5 W. Heller, E. Koerfer and H. Schmedders, Production of special-grade naturally hard rails in Germany and experience gained in operation, Proc. 2nd Iti. Heavy Haul Railway Co& 1982, 82-HH-18, pp. 170-177. 6 K. Sugino, H. Kageyama and H. Masumoro, Development of weldable high-strength steel rails, Proc. 2nd Iti. Heavy Haul Railway Co& 1982, 82-HH-20, pp. 187-198. 7 W. H. Hodgson, Y. K. Yates and R. K. Preston, The development of a second generation of alloy steel rails for heavy haul applications, Proc. 2nd ITU. Heavy Haul Railway COT& 1982, 82-HH-22, pp. 207-215. 8 K. Sugino, H. Kageyama, T. Suzuki, K. Fukuda, H. Yoshitake, Y. Makino and M. +hii, Development of in-line heat treated DHH rails, Proc. 4th Iti. Heavy Had Railway Cm, 1989, pp. 41-45. 9 S. Marich, J. W. Cottam and P. Curcio, Laboratory investigation of transverse defects in rails, Proc. 2nd Heavy Haul Railways Coqf, 1978,The Institutionof Engineers, Australia, pp. 1-13. 10 H. Ghonem, J. Kalousek, D. H. Stone and E. E. Laufer, Aspects of plastic deformation and fatigue damage in pearlitic rail steel, Proc. 2nd Int. Heavy Haul Railway Cm& 1982, 82-HH-31, pp. 339-349.
328 11 C. G. ChipperfIeld and A. S. Blicblau, Modelhng rolling contact fatigue in rails, Rail Int., 15 (1) (1984) 25-51. 12 K. Sugino, H. Kageyama and H. W. Newell, Detection method for harmful inclusions in rail steels, Proc. AREA, Bull., 89 (716) (1988) l-20. 13 B. N. Leis and R. C. Rice, Rail fatigue resistance - increased tonnage and other factors of consequence, Proc. 2nd Int. Heavy Haul Railway Cm+, 1982, 82-HH-12, pp. 99-117. 14 S. Marich and U. Maass, Higher axle loads are feasible - economics and technology agree, Pre-ConJ Proc. 3rd Int. Heavy Haul Railway CM, 1986, pp. l-14. 15 K. Sugino, K. Miyamoto, M. Nagumo and K. Aoki, Structural alterations of bearing steels under rolling contact fatigue, Trans. Iron Steel Inst. Jpn., 10 (1970) 98-111. 16 R. P. Reiff, Report from FAST - recent test experiences, Transpmtatim Research Record 1174, Transportation Research Board, NRC, 1988. 17 R. K. Steele, Recent North American experience with shelling in railroad rails (Part 2), Proc. AREA Bull., 90 (723) (1989) 395-408.
Appendix
A: Discussion
of paper
Qmxtion (Mr. Mutton): Please elaborate on the possible effect of reducing the level of MnS inclusions on these high strength rail steels: would there be an increase in the susceptibility to hydrogen embrittlement/flaking? Answer (Dr. Sugim): If the sulphur content is reduced there may indeed be a greater possibility of hydrogen embrittlement or “shatter cracks”. On the other hand, large MnS inclusions are detrimental to fatigue. I think we should keep present levels of MnS.
Questicm (Dr. Steele): What is the effect of sulphur content on dry wear in rail steel? Answer (Dr. Clayton): I would expect the wear rate to increase with an increase in sulphur content, but there is little conclusive evidence at present.
Wear, 144 (1991) 319328
Metallurgical improvement of rail for the reduction rail-wheel contact fatigue failures
of
K. Sugino, H. Kageyama and C. Urashima Yawata R & D Ikzboratmy, Central R & D Bureau, Nippon Steel Corporation, KitaKyushu (Japan)
A. Kikuchi Plant Engineering
& Technology Bureau, Nippon Steel Corpmation,
Tokyo (Japan)
Abstract One of the important factors in prolongs the service life of rails is the prevention or retardation of the occurrence of shelling and transverse defects in heavy haul railways. This paper describes a rne~~~c~ ~vest~ation of rails which failed in service and gives a preliminary calcuhktion of the stress behaviour in order to acquire a deeper understanding of the effect of non-metallic inclusions, hardness level and its suitable distribution in the rail head from the standpoint of rail production.
1. Introduction Improvement of efficiency in railway transportation demands further increases in speed and the maximum axle load of trucks [ 11. In order to solve these technical challenges great efforts towards the improvement of rail steels have been made [Z-8]. One of the important problems is that of prolonging the service life of rails through the prevention or retardation of the occurrence of rolling contact fatigue defects, especially shelling and transverse defects (TDs), in heavy haul railways 19, lo]. In our experience it seems that the occurrence of both defects is controlled by almost identical metallurgical factors as far as their initiation is concerned. These metallurgical factors are considered to be non-metallic inclusions, hardness and residual stress in the rail head. In this paper we report a number of investigations which were conducted in order to obtain information which might lead to the improvement of these factors from the viewpoint of rail production.
2. Experimental procedures 2.1. ~et~~l~rg~~al
investigation
of shelling and TDs
From the standpoint of quality control of rail steel through production, the composition, size and volume of non-metallic inclusions around the 0043-1648/91/53.50
0 Elsevier Sequoia/Printed in The Netherlands
320
Di.stsnee
Fig. 1. Sampling position for non-medic
from
rail
head
surface
inclusion counting.
Fig. 2. Typical hardness patterns in rail head for stress calculation, high strength rail: -, case 1; ---, case 2; -.-, Std C. TABLE 1 Chemical composition and inclusion sizes of rail for running test at FAST Element
c
Si
Mn
P
S
Cr
Al
N
0
H
Amount (wt.%, *ppm)
0.81
0.22
0.89
0.014
0.001
0.02
0.006
33*
32*
0.6*
Total length of A&O, cluster: less than 650 Frn. Unit length of MnS inclusion: less #an 50 pm (average).
initiation point of shelling and TDs in failed raiis in a North American heavy haul railroad were investigated metaIlurgicaIIy in detail. Results by other researchers [ 1 l] showed that stringer-type ahunina clusters were mainly observed. In order to count ahunina clusters of over 100 pm, which is the limiting size of practical detection, a semiautomatic counting method was developed and used [ 121. Figure 1 shows the observed area of 20 mm by 10 mm beneath the gauge corner surface of the rail head where shehing and TDs often occur. The total length of ahunina clusters of 14 failed and 10 new raiIs which had been co~erci~y produced by several rail miIls was measured as an index from the s~dpo~t of quality control of rail production. 2.2. Eflect of rail hardness on shelling and TL; ocmrrme In order to clarify the effect of rail hardness level on shelling and TD occurrence, two classes of rails with low and high hardness were produced from the same melt of steel. One was as-rolled Std C rail with a hardness of 250 HB and the other was heat-treated high strength (hardness) rail with an average hardness of 345 HB Table 1 shows the chemical composition and inclusion sizes of the melt. The two classes of rails were sent to the Facility for Accelerated Service Testing (FAST) at the Transportation Test Center and submitted to the heavy haul running test (defect occurrence and growth (DOG) test).
321
2.3. meet
of hardness distribution
on intern& stress in rail head
From the standpointof rail production, it is importantto know the most suitable hardness ~bution in the rail head for the prevention of shelling and TD occurrence. On the other hand, it is clear that the initial hardness distribution in the rail head is changed by the repeated applied loads due to wheel passing, resulting in the induction of tensile stress in it [ 131. As a preliminary step, the stress behaviour in the rail head during several cycles of constant applied load on the gauge comer surface was calculated. Referring to an actual rail production, two types of hardness distribution in the rail head, cases 1 and 2, were chosen as shown in Pig. 2, because they are often seen practically. Prerequisites for the calculation are shown in Fig. 3.
3. Resnlts 3.1. Ntmmtetallic inclusions in failed rails Fourteen rail samples cut from the failed rails on the North American heavy haul rail track were metalhngicahy investigated. Much attention was paid to the relation between the initiation point of shelling and the non-
I.
: 132RE
Rail
Std 0 rail
132RE Length
: 5OOma (Case
2.
Wheel
: $844~.
3.
Load
: Normal 20 ton
Load
: 6 thes
repetition
Load center
: FEM : MARC.
Constitutive
8 nodes
analysis
Metallurgical
by van condition
properties
modulus
Poisson ratio
haxabedron
Isw :Elastic-plastic
Mises
Young’s
and
10 ton >
: by Ifertz’ equation
Calculation
Constraint
1 and 2 )
Load distribution
Software
5.
rail
Straight tread
( Tangential
4.
6.
High strength
: 2.06 X 10’ MPa : 0.3
Yield stress . Work hardening
: Experimental
Fig. 3. Prerequisites for internal stress calculation by finite element analysis.
322
metallic inclusions. From this investigation the results can be summarized as follows. (1) There was no correlation between the number of failed rails and the chemical compositions of aluminium, oxygen and other elements forming inclusions in the rail steels, as shown in Fig. 4. Although an increase in cleanliness of rail steel is important for the prevention or retardation of shelling and TD occurrence, it is not necessarily satisfied by a decrease in delusion-fo~g elements during the steel-making process. (2) Shelling and TDs were clearly confirmed to initiate at rather large stringer-type alumina clusters in two of the rail samples. Therefore the size of alumina clusters among the various inclusions should be kept as small as possible. (3) In spite of intensive observation by an optical microscope, no inclusions could be found at the vitiation point of shelling and TDs in the remainder of the failed rails, whereas by the quantitative counting it was found that they included a comparatively large number of alumina clusters, as shown in Fig. 5. Therefore new rails, plotted as E and S in the same range of total length of alumina clusters, would fail after a similar period in service when installed on the same track line. (4) In the case of the failed Std C rails it is also noted that the rails with shelling and TDs included alumina clusters of total length at least 2000 pm at a specific location and area beneath the gauge corner of the rail head. On the other hand, in the case of the high strength rails (G, H, X and Y) this value appears to be about 5000 pm. This implies that the detrimental effect of alumina clusters on shelling and TD occurrence depends greatly on the hardness of the rail head. This has certainly been confirmed in many heavy haul railways as an empirical fact [ 14 1.
m
Fatigue
Mn(x10-2wt%) Fig.
4. Chemical
damaged rails
S(X 10-3wt%) analysis
AE(XlO-“~3%)
of manganese,
sulphur,
0(X103wt@
aluminium and oxygen in rail samples.
323
0
LO
10
20
30
40
60
60
Fig. 5. Measurement of alumina clusters in raii heads: 0, fractured rails; a, crack-detected rails; 0, new rails.
(5) There was no evidence of a detrimental effect of MnS on shelling and TD occurrence. This appears to fit the general view in the rolling contact fatigue problem [ 151.
3.2. Eflect of hardness in rail head on shelliq and TD occurrence In order to confirm the sensitivity of non-metahic inclusions to shelling and TD occurrence in rails of different hardness levels, both Std C and high strength rails produced from the same melt have been submitted to the rail performance experiment (DOG test) at FAST with many kinds of rails produced by other rail miIls. According to Reiff [ 161 and a private communication of interim results at the passing tonnage of 120 MGT (million gross tons) at FAST, it was found that in the Std C rail, shelling and TDs occurred after 60-90 PvIGT, whereas in the high strength rail they had not occurred at least up to 120 MGT. The failed Std C rails were metallographically investigated in detail, but alumina clusters or other inchrsions could not be confirmed at or near the initiation point on the fracture surface of shelling and TDs except for one sample. It was also recognized that all samples investigated showed a similar mode of failure concerning the location and direction of the internal crack and the hardness change in the matrix near it. Referring to the above results, it is considered that:
324
(1) a decrease in volume and size of MnS inclusions scarcely affects the prevention or retardation of shelling and TD occurrence; (2) strengthening of the rnet~~gi~al structure surrounding inclusions, especially alumina inclusions, considerably reduces their detrimental effect; (3) in the case of the hardness level of the Std C rail, shelling and TDs are suggested to initiate from very small alumina inclusions which cannot be perfectly controlled by the present steel-refining process. Figure 6 shows typical examples of the failed Std C rails on FAST track. The failure initiated as shelling and grew to TDs. Almost all the shells investigated occurred near the boundary of the work-hardening zone owing to wheel load, as noted by Steele and Reiff [ 11, although it was not confirmed that there was a cyclic softening zone around the shell. This proves that plastic deformation in the rail head due to repeated wheel loads is closely related to the initiation of shelling and TDs. 3.4. Suitable hardness in rail head When a hardening process is applied to the rail head, the hardness is usually highest on the surface of the rail head and decreases inwards. Nippon Steel developed an in-line slack-quenching process for high strength rail production in 1987 [8]. This process has the characteristic of making the hardening zone in the rail head much deeper than normal reheating processes do. Therefore, from the standpoint of rail production, control of the hardness distribution in the rail head is very important as well as refinement of the molten steel to decrease non-meta~c inclusions, p~icul~ly alumina clusters. As a rough estimation of a suitable hardness distribution for the prevention or retardation of shelling and TD occurrence, the effect of two typical types of hardness pattern, cases 1 and 2, in the head of high strength rail on plastic deformation was calculated by a three-dimensional finite element analysis, although the prerequisites for the calculation are not necessarily
135 810 15 20 25 30 35 Distanoe from surface 6m>
Fig. 6. Fracture surface and hardness distributionof faiIed rails after running test at FAST: 0, 3-0833; A., 25-0301; 0, 25-1290.
325
well fitted for all the actual conditions. These hardness patterns are shown in Fig. 2 and have often been seen in actual production. l?igure 7 shows the effect of the hardness patterns on the e~~v~ent stsess (von M&es) with depth in the head of high strength rail after a repeated (four to six times) applied load of 20 t. Adopting the residual stress as an
-+I 6
0
e
G
_.+.. 7
400
200
__
07
0
2
4
6 Distance
8
--
10
from rail
12
head surface
14
16
18
20
(smr>
7. Effect of hardness diitribution on von Mies equivalent stress beneath load centre at gauge comer: 1, yield stress; 2, initial stress; 4 and 6, residual stress after fourth and sixth load cycle respectively; 3, 5, 7, stress under fourth, fifth and sixth applied load respectively. Fig.
326
index for plastic deformation, it is noted that the amount and depth of this in case 2 is larger than in case 1 in spite of the harder surface zone. Theoretical studies on the initiation and propagation of shelling and TDs in North America have been summarized by Steele [ 171. It is not necessarily easy to confirm which stress mainly controls shelling and TD occurrence in cases of actual failure. Following the precedent, an attempt was made to calculate the stress or stress range induced in a particular direction in the rail head under conditions close to actual wheel load and shape, size and mechanical properties of the rail, expecting that the stress should reveal a positive value and a maximum at the depth of shelling and TD occurrence. Figure 8 shows the residual stress in the direction of loading beneath the gauge corner surface under the conditions for calculation shown in Fig. 3. Assuming that the repetition of a positive stress affects the initiation of shelling, it becomes effective below a depth of about 6 mm at the rail gauge corner. The stress of case 1 is smaller than that of case 2. Therefore it is preferable to adopt the hardening process for preventing the rapid decrease in hardness with depth in the rail head. It is also important to keep the rail head surface as hard as possible. According to another interim analysis of failed 132RE high strength rails by a running test at an axle load of 27 t in a heavy haul railway, shelling and TDs occurred at alumina clusters of over 400 pm which were located at a depth of about 6-12 mm in the rail head after 500 MGT. This result
-400 0
2
4
6
3 Distance
10
12
from surface
14
16
18
20
(78x1
Residual stress in direction of loading (normal load 20 t, six repeated loadings): 0, case 1; @, case 2.
Fig. 8.
327
seems to be qualitatively in good agreement with the calculation described above. F’urther study should be done for a more quantitative analysis. 4. Conclusions Considering the results from these few experiments, we can set out the following basic guidelines for the production of rails for heavy haul railways: (1) adopt a steel-making process which gives as large a decrease in long ahunina clusters as possible; (2) keep the present cleanliness level of MnS inclusions; (3) adopt a hardening process which makes the hardness difference between the surface and the internal region of the rail head as small as possible, while maintaining high hardness on its surface. Acknowledgments
The authors would like to thank Dr. R. K. Steele and Dr. R. P. Reiff at AAR for an offer of rail samples with useful information. A portion of this work was done with Mr. H. W. Newell at Norfolk Southern Corporation. We also thank the Nippon Steel Corporation for permission to publish. References 1 R. K. Steele and R. P. Reitf, Rail: its behavior and relationship to total system wear, Proc. 2nd Int. Heavy Haul Railway Cor& 1982 82-HH-234, pp. 227-276. 2 S. Marich and P. Curcio, Development of high-strengthalloyed rail steel suitable for heavy duty application, ASl’!M Spec. Tech. PubL 644, 1976, pp. 167-211. 3 J. D. Young, United Kingdom development of rails rolled from continuously cast blooms, ASi% Spec. Tech. PubL 644, 1976, pp. 256-284. 4 Y. E. Smith and F. B. Fletcher, Alloy steels for high-strength,as-rolled rails, ASTM Spec. Tech. PubL 644, 1976, pp. 212-232. 5 W. Heller, E. Koerfer and H. Schmedders, Production of special-grade naturally hard rails in Germany and experience gained in operation, Proc. 2nd Iti. Heavy Haul Railway Co& 1982, 82-HH-18, pp. 170-177. 6 K. Sugino, H. Kageyama and H. Masumoro, Development of weldable high-strength steel rails, Proc. 2nd Iti. Heavy Haul Railway Co& 1982, 82-HH-20, pp. 187-198. 7 W. H. Hodgson, Y. K. Yates and R. K. Preston, The development of a second generation of alloy steel rails for heavy haul applications, Proc. 2nd ITU. Heavy Haul Railway COT& 1982, 82-HH-22, pp. 207-215. 8 K. Sugino, H. Kageyama, T. Suzuki, K. Fukuda, H. Yoshitake, Y. Makino and M. +hii, Development of in-line heat treated DHH rails, Proc. 4th Iti. Heavy Had Railway Cm, 1989, pp. 41-45. 9 S. Marich, J. W. Cottam and P. Curcio, Laboratory investigation of transverse defects in rails, Proc. 2nd Heavy Haul Railways Coqf, 1978,The Institutionof Engineers, Australia, pp. 1-13. 10 H. Ghonem, J. Kalousek, D. H. Stone and E. E. Laufer, Aspects of plastic deformation and fatigue damage in pearlitic rail steel, Proc. 2nd Int. Heavy Haul Railway Cm& 1982, 82-HH-31, pp. 339-349.
328 11 C. G. ChipperfIeld and A. S. Blicblau, Modelhng rolling contact fatigue in rails, Rail Int., 15 (1) (1984) 25-51. 12 K. Sugino, H. Kageyama and H. W. Newell, Detection method for harmful inclusions in rail steels, Proc. AREA, Bull., 89 (716) (1988) l-20. 13 B. N. Leis and R. C. Rice, Rail fatigue resistance - increased tonnage and other factors of consequence, Proc. 2nd Int. Heavy Haul Railway Cm+, 1982, 82-HH-12, pp. 99-117. 14 S. Marich and U. Maass, Higher axle loads are feasible - economics and technology agree, Pre-ConJ Proc. 3rd Int. Heavy Haul Railway CM, 1986, pp. l-14. 15 K. Sugino, K. Miyamoto, M. Nagumo and K. Aoki, Structural alterations of bearing steels under rolling contact fatigue, Trans. Iron Steel Inst. Jpn., 10 (1970) 98-111. 16 R. P. Reiff, Report from FAST - recent test experiences, Transpmtatim Research Record 1174, Transportation Research Board, NRC, 1988. 17 R. K. Steele, Recent North American experience with shelling in railroad rails (Part 2), Proc. AREA Bull., 90 (723) (1989) 395-408.
Appendix
A: Discussion
of paper
Qmxtion (Mr. Mutton): Please elaborate on the possible effect of reducing the level of MnS inclusions on these high strength rail steels: would there be an increase in the susceptibility to hydrogen embrittlement/flaking? Answer (Dr. Sugim): If the sulphur content is reduced there may indeed be a greater possibility of hydrogen embrittlement or “shatter cracks”. On the other hand, large MnS inclusions are detrimental to fatigue. I think we should keep present levels of MnS.
Questicm (Dr. Steele): What is the effect of sulphur content on dry wear in rail steel? Answer (Dr. Clayton): I would expect the wear rate to increase with an increase in sulphur content, but there is little conclusive evidence at present.