Evaluation of Remaining Service Life of Aged Rails Hiroo Kataoka 1 , Yuya Oikawa 1 , Tadashi Deshimaru 1 , Noritsugi Abe 2 1 Railway Technical Research Institute, Tokyo, Japan, 2 Nihonkido Kogyo Co., Tokyo, Japan Abstract Rail maintenance represents an enormous expense in the operation of railways, and to prevent rail failure, rails in Japan are replaced periodically. However, as vehicles have become lighter and rail joint fractures have decreased in recent years, the cycle of rail replacement can be extended. The authors therefore reviewed the rail replacement period focusing on cracks from rail foots on continuous welded rail (CWR) and star cracks at the boltholes of fishplated joints. In this study, fatigue tests were carried out on laid rails, and their remaining life was evaluated. The average gross tonnage of the test rails was 380 MGT for the CWR of Shinkansen tracks, 540 MGT for CWR, and 330 MGT for the fishplated joints of meter gauge lines. At the same time, a dynamic stress analysis model of a fishplated joint was developed to calculate stress under a range of track conditions. As a result of evaluating the remaining rail life, it was found that the rail replacement period could be extended. It is, however, necessary to remove longitudinal rail surface irregularities at welds by grinding, and maintenance is essential for severe wear between fishplates and rails at fishplated joints. 1. Introduction Rail maintenance represents an enormous expense in the operation of railways, and to prevent rail failure, rails in Japan are replaced periodically. However, as vehicles have become lighter and rail joint fractures have decreased in recent years, the cycle of rail replacement can be extended. There are a number of different types of rail fracture. Squart and head-check frequently occur on site, and can be detected by ultrasonic or visual inspection. However, cracks occurring at rail foots other than in a position right under the rail web are difficult to detect. The inspection of star cracks at the boltholes of fishplated joints may also be overlooked because of battered rail ends and difficulties in adjusting the detection range. The rail replacement period was therefore reviewed, focusing on cracks from rail foots on continuous welded rail (CWR) and star cracks at the boltholes of fishplated joints. In the past, the authors performed bending fatigue tests on new rails and evaluated the service life of CWR and fishplated joints[1][2]. A dynamic analysis model of a welded joint was established to calculate bending stress, and was applied to evaluate the life of CWR. In this study, fatigue tests were carried out on laid rails, and their remaining life was evaluated. At the same time, a dynamic stress analysis model of a fishplated joint was developed to calculate stress under a range of track conditions. This is composed of a beam model to calculate dynamic wheel/rail contact forces and a solid model to calculate the stress field from these contact forces. The remaining life of welded joints with boltholes was also evaluated. These rails were once used with fishplated joints, and were later welded to form CWR. 2. Evaluation of service life of laid rails of CWR 2.1 Bending fatigue test on laid CWR To estimate fatigue life of rails, it was necessary to clarify the relationship between stress and the number of cycles to failure through fatigue testing. The authors carried out bending fatigue tests on laid rails (Fig.1), which undergo an average accumulated passing tonnage of 380 million gross tons (MGT) for the CWR of Shinkansen tracks and 540 MGT for the CWR on meter gauge lines. The results in Fig.2 show that the number of weld cycles to failure is in the same order as the base material of the rail; these values were lower than those for new rails. The number of weld cycles of gas-pressure welding, enclosed arc welding, thermit Fig. 1 Bending fatigue test 1300 150 unit:mm
Transcript
Evaluation of Remaining Service Life of Aged Rails
1Railway Technical Research Institute, Tokyo, Japan, 2Nihonkido Kogyo Co., Tokyo, Japan
Abstract
Rail maintenance represents an enormous expense in the operation of railways, and to prevent rail failure, rails in Japan are replaced periodically. However, as vehicles have become lighter and rail joint fractures have decreased in recent years, the cycle of rail replacement can be extended. The authors therefore reviewed the rail replacement period focusing on cracks from rail foots on continuous welded rail (CWR) and star cracks at the boltholes of fishplated joints. In this study, fatigue tests were carried out on laid rails, and their remaining life was evaluated. The average gross tonnage of the test rails was 380 MGT for the CWR of Shinkansen tracks, 540 MGT for CWR, and 330 MGT for the fishplated joints of meter gauge lines. At the same time, a dynamic stress analysis model of a fishplated joint was developed to calculate stress under a range of track conditions. As a result of evaluating the remaining rail life, it was found that the rail replacement period could be extended. It is, however, necessary to remove longitudinal rail surface irregularities at welds by grinding, and maintenance is essential for severe wear between fishplates and rails at fishplated joints.
1. Introduction
Rail maintenance represents an enormous expense in the operation of railways, and to prevent rail failure, rails in Japan are replaced periodically. However, as vehicles have become lighter and rail joint fractures have decreased in recent years, the cycle of rail replacement can be extended. There are a number of different types of rail fracture. Squart and head-check frequently occur on site, and can be detected by ultrasonic or visual inspection. However, cracks occurring at rail foots other than in a position right under the rail web are difficult to detect. The inspection of star cracks at the boltholes of fishplated joints may also be overlooked because of battered rail ends and difficulties in adjusting the detection range. The rail replacement period was therefore reviewed, focusing on cracks from rail foots on continuous welded rail (CWR) and star cracks at the boltholes of fishplated joints. In the past, the authors performed bending fatigue tests on new rails and evaluated the service life of CWR and fishplated joints[1][2]. A dynamic analysis model of a welded joint was established to calculate bending stress, and was applied to evaluate the life of CWR. In this study, fatigue tests were carried out on laid rails, and their remaining life was evaluated. At the same time, a dynamic stress analysis model of a fishplated joint was developed to calculate stress under a range of track conditions. This is composed of a beam model to calculate dynamic wheel/rail contact forces and a solid model to calculate the stress field from these contact forces. The remaining life of welded joints with boltholes was also evaluated. These rails were once used with fishplated joints, and were later welded to form CWR.
2. Evaluation of service life of laid rails of CWR
2.1 Bending fatigue test on laid CWR To estimate fatigue life of rails, it was necessary to clarify the relationship between stress and the number of cycles to failure through fatigue testing. The authors carried out bending fatigue tests on laid rails (Fig.1), which undergo an average accumulated passing tonnage of 380 million gross tons (MGT) for the CWR of Shinkansen tracks and 540 MGT for the CWR on meter gauge lines. The results in Fig.2 show that the number of weld cycles to failure is in the same order as the base material of the rail; these values were lower than those for new rails. The number of weld cycles of gas-pressure welding, enclosed arc welding, thermit Fig. 1 Bending fatigue test
1300
150
unit:mm
welding and flash butt welding are also in the same order. Surface roughness on the rail base due to corrosion was assumed to have a large effect on the remaining life. However, as the number of rail cycles on Shinkansen tracks is somewhat different from that on meter gauge lines, the S-N curves of laid rails on Shinkansen and meter gauge lines were calculated differently. Based on the test results, an S-N curve expressing the remaining life of laid rails at a fracture probability of 50% was obtained using weighted Probit analysis suitable for small-sample fatigue data sets. To calculate the S-N curve, the authors added 20 items of data representing the actual breaking stress of flash butt welds on Shinkansen tracks or thermit welds on meter gauge line. An S-N curve at an arbitrary fracture probability was then calculated from the one at 50% by assuming a constant standard deviation. Fig.3 shows the S-N curve and fatigue data. Miner’s rule was adopted and modified so that the slope in the area where the stress amplitude was under the fatigue limit became half that in the stress area over the fatigue limit. As the rails for testing had different histories in terms of accumulated tonnage and surface irregularity, the test data was corrected to even out the accumulated tonnage as per the flow chart shown in Fig.4.
2.2 Method of estimating the service life of laid CWR Service life was estimated using the S-N curve for laid rail welds obtained in Section 2.1 as per the flow chart shown in Fig.5. Fig.6 shows the relationship between accumulated tonnage and rail surface irregularity in a 100mm chord, measured using digital surface irregularity-measuring equipment before fatigue tests were performed. The accumulated tonnage reached the order of hundred of million, meaning that the growth rate of surface irregularity obtained from the irregularity measurement results was reliable. Here, the growth rates of surface irregularity per 100 MGT of train passage on meter gauge line were assumed as 0.1mm for enclosed arc welding and 0.05mm for other kinds of weld. The growth rate of irregularity on
End Yes
No Agree with previous S-N curve ?
Calculating each specimen
Calculating tentative S-N curve
i
iiii L
ttNN −=∆ Compensated number
of cycles ti :Accumulated tonnage, :Averaged accumulated tonnage it
Pulsative stress amplitude Si Number of cycles Ni
Modified number of cycles iii NNN ∆+=*
Estimation of S-N curve
Remaining life Ri
Fig. 4 Flow chart of estimating S-N curve base on averaged accumlated tonnage
Fig. 3 S-N curve of laid rail of CWR at 50% fracture probability
╘╡
₧═
▓ ⁿ ▓ ⁿ▌ ⁿ ▌ ⁿ
Meter gauge line S=-136log10N+1083 Fatigue limit 232 N/mm2
Shinkansen tracks was assumed as 0.2mm per 100 MGT of train passage, based on the results of previous studies [1]. Cumulative fatigue damage was calculated for different track conditions and vehicle types on Shinkansen and meter gauge lines using equations developed in the past to estimate rail base bending stress in consideration of surface irregularity [1]. Surface irregularity was assumed to grow in proportion to the accumulated tonnage and be reduced by periodical rail grinding.
2.3 Results of the estimated service life of laid CWR Overall fatigue life was estimated by adding the accumulated tonnage of the specimen to the remaining life estimated in section 2.2. Fig.7 shows the estimated fatigue life. In this figure, the lifespan was estimated for ballasted track with ballast mats on a Shinkansen line and for ballasted track assuming enclosed arc welds surface growth under loose sleeper conditions on a meter gauge line. From the estimated life at a probability of 0.01%, the existing 60kg rail replacement period of 600 MGT on Shinkansen lines can be extended by 300 MGT by grinding 0.1mm per 100 MGT of accumulated tonnage. Similarly, from the estimated life at a probability of 0.1%, the existing 60kg and 50kgN rail replacement period of 800 MGT and 600 MGT on meter gauge line can be extended by 300 MGT by grinding 0.1mm per 100 MGT of accumulated tonnage. In practice, extension of the replacement period by 100 or 200 MGT is recommended. The results validate proposals to extend the replacement period by grinding the rail surface on the basis of past studies.
Fig.5 Flow chart of evaluateing fatigue life of CWR
Surface irregularity of rail welds (each welding type)
Rail bending stress (track structure)
wheel load
Accumulated fatigue damage = 1?
End
No
Yes
Thermal stress
Calculating accumulated fatigue damage Fatigue damage law (modified Miner’s rule Fatigue strength of rail welds (S-N curve)
Growing of surface irregularity
Surface grinding
100 mm
Fig.6 Relationship between acculmulated tonnage and 100 mm chord surface irregularity
▌ ╧ ╩
╖╩
Growth rate: 0.1 mm/100 MGT
Accumulated tonnage (100 MGT)
100
mm
cho
rd s
urfa
ce
irreg
ular
ity o
f rai
l (m
m)
Weld type
Growth rate: 0.05 mm/100 MGT
100 mm chord surface irregularity
Rail surface
3. Evaluation of the service life of jointed rails
3.1 Bending fatigue tests on jointed rails Jointed rails, which undergo an average accumulated passing tonnage of 330 million gross tons (MGT), were subjected to bending fatigue tests as shown in Fig. 8. The tests produced cracks at the edge of the holes extending at a 45-degree angle, as shown in Fig.9. Based on the test results, an S-N curve expressing the remaining life of jointed rails at a fracture probability of 50% was obtained by weighted Probit analysis, similar to the case with rail welds covered in section 2.1. However, the differing accumulated tonnage of the test rails was not taken into account. To calculate the S-N curve, ten data items of the actual rail breaking stress of 1500 N/mm2 were added. Fig. 10 shows the S-N curve and fatigue data. 3.2 Dynamic stress analysis model for jointed rails To evaluate rail stress at joints under a range of track conditions, a dynamic stress analysis model (Fig.11) consisting of a beam model to calculate dynamic force and a solid model to calculate stress from this force was developed. The model was validated using the test results obtained [3]. 3.3 Estimated service life of jointed rails The remaining fatigue life of used jointed rails was evaluated using the modified Miner’s rule as a
Fig. 8 Bending fatigue test of fishplated joint rail
Unit : mm 10
130 77 350
Fig. 9 Example of rail broken in fatigue test
Crack
Solid model calculating dynamic stress
Fine mesh model around bolt holes
Beam model calculating dynamic force
Rail Fishplate
Sleeper
V
Rail
Sleeper
Unsprung mass
Railpad
Bogie
Car body (Calculating stress from displacement results of the previous analysis)
Fig. 11 Dynamic stress analysis model of jointed rail
Fig. 10 S-N curve of laid jointed rail at 50% fracture probability
╘ ╡
₧═
╚ ⁿ
S=-146logN+1407 Fatigue limit 444N/mm2
Standard deviation 41N/mm2
Puls
ativ
e st
ress
am
plitu
de
2㰰a
(N/m
m2 )
Number of cycles
Failure No failure 50 % S-N curve
cumulative fatigue damage law. Life expectancy was estimated for ballasted track under normal state, loose sleeper and worn fishplate conditions. In the case of 50kgN rail, field data measured in the past was used to estimate the stress values. In the case of 60kg rail, as the stress values in the field data were relatively small due to minor rail irregularities in the field, they were corrected in consideration of the difference of these irregularities. Estimated stress amplitudes and the sum of the remaining fatigue life and the average accumulated passing tonnage of 330 MGT are shown in Figs. 12 and 13. The fatigue life at a probability of 0.1% under normal and loose sleeper conditions was over 820 MGT for 50kgN rail and 1,210 MGT for 60kg rail. These values are greater than the rail life until replacement adopted by Japanese railway companies, thus enabling extension of the replacement period. However, the remaining fatigue life in the worn fishplate scenario was as short as approximately 10 MGT, suggesting that maintenance of the worn fishplate is important. The estimation was based on the results of tests for a worn fishplate scenario artificially created by inserting 1-mm cuts in the plates. Thus, the estimated life using actual wear conditions may be different from that outlined above. At the same time, the bending fatigue life of jointed rails using the S-N curve calculated in section 2.1 was over 690 MGT for 50kgN rail and over 890 MGT for 60kg rail. Bending fatigue must therefore be taken into consideration when evaluating the service life of jointed rails.
4. Evaluation of the service life of welded joints with boltholes
4.1 Method of estimating the service life of welded joints with boltholes The service life of welded joints with boltholes on meter gauge line was estimated through the following procedures. These rails were once used as fishplated joints and were later welded to make CWR. (a) The rail base bending stress and bolthole stress in stages used as fishplated and welded joints were evaluated for different track conditions and vehicle types. (b) The S-N curves of the rail welds and jointed rails outlined in sections 2.1 and 3.1 were used for the rail base bending stress and bolthole stress respectively to calculate the cumulative rail base and bolthole fatigue damage. Surface irregularity was assumed to grow in proportion to the accumulated tonnage in stages used as fishplated and welded joints, and was also assumed to be reduced by periodical rail grinding in its latter stages. As many welded joints with boltholes were made using thermit welding, the growth rate of 100mm of chord surface irregularity on non-treated rail was taken as 0.05mm per 100 MGT (as prescribed in section 2.2) to calculate bending stress. Bending stress of end hardened rail weld was evaluated using an analytical model. A dynamic stress analysis model of a welded joint with boltholes (similar to the fishplated joint model shown in Fig. 11) was developed. The solid model is shown in Fig. 14. These models were validated in field tests, the results of which are shown in Figs. 15 and 16.
Fig. 12 Estimated stress amplitudes around bolt holes
₧═
┴ ╘ ╩ ╘ ╓ ╘┴ ╘ ╩ ╘ ╓ ╘
≤ ╘ ╒ ┌ ┬ ▄ 㱀
high rail on curve
Stre
ss a
mpl
itude
(N
/mm
2 )
Normal Worn Loose state fishplate sleeper
Limited express, 50kgN Local train, 50kgN Locomotive, 50kgN Limited express, 60kg Local train, 60kg Locomotive, 60kg
Fig. 13 Example of evaluated fatigue life of bolt holes
evaluated remaining life
average of tonnage of test rails
┴ ╘ ╩ ╘ ╓ ╘┴ ╘ ╩ ╘ ╓ ╘
┌⌡╜
─▌
≤ ╘ ╒ ┌ ┬ ▄ 㱀
high rail on curve
Normal Worn Loose state fishplate sleeper
Fatig
ue li
fe o
f bol
t hol
e (M
GT)
Limited express, 50kgN Local train, 50kgN Locomotive, 50kgN Limited express, 60kg Local train, 60kg Locomotive, 60kg
4.2 Evaluation of the service life of welded joints with boltholes Comparing the remaining life for a crack at a rail foot with that at a bolthole, the former was estimated to be shorter (Fig. 17). The replacement period for the former was therefore examined by evaluating the remaining life for a crack at a rail foot. Here, the growth rate of surface irregularity on meter gauge line was assumed to be 0.05mm per 100 MGT of train passage for non-treated and end hardened rail welds, and the grinding period thickness was assumed as 0.05mm or 0.1mm per 100 MGT of accumulated tonnage after welding. The stress of end hardened rail welds was estimated from analytical results. From the estimated life expectancies shown in Fig. 18, the existing 60kg and 50kgN rail replacement period on meter gauge line can be extended by 100 or 200 MGT by grinding 0.1mm per 100 MGT of accumulated tonnage, as in the case of CWR. However, as estimates for welds with boltholes in end hardened rail with large surface irregularity were shorter, extension of the replacement period is confined to rails whose surface irregularity is small when welding and is grinded periodically.
5. Conculusion
To evaluate the fatigue life of rails, tests were carried out and dynamic stress analysis models were developed. The fatigue life of CWR, fishplated joints and welded joints with boltholes was evaluated, with the results outlined below. (1) In the fatigue tests on laid rails, the average gross tonnage of the test rails was 380 MGT for CWR on Shinkansen tracks, 540 MGT for CWR and 330 MGT for fishplated joints on meter gauge lines. The results from welded rails showed that the number of weld cycles to failure was in the same order as those of the rail’s base material. The values were lower than those for new rails.
(Whole model)
(Magnified figure of rail welds with boltholes)
Fig. 14 Stress analysis model of welded joint with boltholes
(a) Analysis Fig. 15 Rail base stress
20N/mm2
╫ ╩
₧═
-1000 0 1000 Distance (mm)
40
20
0
Stre
ss(N
/mm
2 )
-1000 0 1000 Distance (mm)
(b) Measurement
50N/mm2
╫ ╩
₧═
10050
0 -50
-100 -1000 0 1000
Distance (mm)
-1000 0 1000 Distance (mm)
(a) Analysis Fig. 16 Rail bolt hole stress
Stre
ss(N
/mm
2 ) (b) Measurement
(a) 50kgN rail (b) 60kg rail Fig. 17 Comparison of fatigue life between for bolt hole and rail base bending stress (Growth rate of surface irregularity : 0.1 mm/100 MGT)
┌ ₧ ╠ ▌ ╧ ╩ ▀
㩠⌡╜
┌ ₧ ╠ ▌ ╧ ╩ ▀
㩠⌡╜
Accumulated tonnage at welding (100 MGT)
Accumulated tonnage at welding (100 MGT)
Fatig
ue li
fe (1
00 M
GT)
Fatig
ue lif
e (1
00 M
GT)
Non-treated rail / bolt hole Non-treated rail / base bending
End hardened rail / bolt hole End hardened rail / base bending
(2) Dynamic stress analysis models of fishplated joints and welds with boltholes were developed to calculate stress under a range of track conditions. These models were validated using field test data. (3) The remaining life of welded joints with boltholes was evaluated. These rails were once used as fishplated joints and were later welded to make CWR. Comparing the remaining life for a crack at a rail foot with that at a bolthole, the former was estimated to be shorter. (4) Evaluation of the remaining life of rails showed that the rail replacement period could be extended, although it is necessary to remove longitudinal rail surface irregularities at welds by grinding. Maintenance is essential for severe wear between fishplates and rails at fishplated joints. Consequently, several Japanese railway companies have extended the replacement period for rails laid in CWR segments.
References
[1] K. Takao, N. A be. “Estimation of service life of rail welding joint concerning thermal stresses”, In Proceedings of World Congress of Railway Research ‘97, pp. 331-337, (1997).
[2] H. Kataoka, et al., “Evaluation of Service Life of Jointed Rails”, Quarterly Report of RTRI, Vol. 43, No. 3, pp. 101-106, (2005)
[3] H. Kataoka, et al., “Dynamic Analysis of Stresses and Evaluation of Service Life of Jointed Rails”, Quarterly Report of RTRI, Vol. 46, No. 4, pp. 250-255, (2005)
[4] Satou, Y. and Satou, Y., “Life of Rai”, Railway Technical Research Report, No.476, (1965) (in Japanese).
[5] Knothe, KL. and Grassie, S. L., “Modelling of Railway Track and Vehicle/Track Interaction at High Frequencies”, Vehicle System Dynamics, 22, pp.209-262, (1993).
[6] Thomas, J. and Abbas, B. A. H., “Finite Element Model for Dynamic Analysis of Timoshenko Beam”, Journal of Sound and Vibration, 41, pp.291-299, (1975).
Fig. 18 Fatigue life of Welded rail with boltholes (a) Non-treated rail (b) End hardened rail
50kgN rail Growth rate of surface irregularity:0.05mm/100MGT Probability at fracture:0.1%
Fatig
ue li
fe (1
00 M
GT)
Fatig
ue li
fe (1
00 M
GT)
Grinding thickness : No grinding 0.05mm/100 MGT 0.1 mm/100 MGT
