Engineering Fracture Mechanics: Taizo Makino, Takanori Kato, Kenji Hirakawa

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Engineering Fracture Mechanics 78 (2011) 810–825

Contents lists available at ScienceDirect

Engineering Fracture Mechanics


journal homepage: www.elsevier.com/locate/engfracmech

Review of the fatigue damage tolerance of high-speed railway axles


in Japan
Taizo Makino a,*, Takanori Kato a, Kenji Hirakawa b
a
Corporate Research & Development Laboratories, Sumitomo Metal Industries, Ltd., Hyogo, Japan
b
Kyushu Polytechnic College, Fukuoka, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Railway axles are one of the most important components in railway systems since a fail-
Received 30 June 2009 safe design is not available. In the present paper, the fatigue tolerance of the high-speed
Received in revised form 22 December 2009 railway axle in Japan is reviewed. To maintain the safety, the fatigue strength of the axle
Accepted 24 December 2009
has been extensively studied. Theses case histories and consequent improvements in man-
Available online 4 January 2010
ufacturing process are presented. The crack propagation behavior of the induction hard-
ened axle is studied based on the fracture mechanics. Concerning the powered railway
Keywords:
axles, the fatigue design method in Japan is compared with that in Europe and the effect
Fatigue design
Railway axles
of the train velocity on the allowable load is discussed.
Fretting fatigue Ó 2010 Elsevier Ltd. All rights reserved.
Induction hardening
Carbon steel

1. Introduction

Fatigue failure of railway axle has been a source of difficulty for engineers since the railway service started in the early
part of the 19th century. Derailment caused by fatigue failure of railway axles is very rare in Japan. There have been no pas-
senger casualties since the inauguration of the high speed Shinkansen railway system in 1964. Railway axles are, however,
one of the most important components in railway systems since a fail-safe design is not available. In order to maintain the
safety of the high-speed railway system, a large number of investigations and experiments have been carried out by out-
standing research ever since, and many improvements have been made in the material, manufacturing, heat treatment
and design method. Axles in service are regularly checked by ultrasonic testing and magnetic particle inspection.
It was recognized that the design philosophy of Japanese high-speed railway axles is different from European design
methods. The critical parts for fatigue strength in Japanese high-speed railway axles are the press-fitted part such as the
wheel seat and the gear seat because the diameter ratio, (D/d), where d is the diameter of axle body and D is the wheel seat
diameter, is about 1.10. Whereas, in Europe, the diameter ratios are over 1.12 which make the fillet of axle body the critical
part for fatigue strength.
To increase the fatigue strength of the press-fitted part of the axle, which suffered from fretting fatigue, heat-treatment of
induction hardening method has been applied successfully in Japan. Once Shinkansen train has reached 600,000 km, an
inspection of the wheel seat using a magnetic particle by dismounting wheel form axle has been made. For several years after
the inauguration of the high-speed railway service of Shinkansen, minute fretting fatigue cracks at the wheel seat were
found on several axles inspected. However, no indication of fatigue crack by magnetic particle inspection has been detected
for over 20 years.

* Corresponding author. Tel.: +81 6 64895733; fax: +81 6 64895794.


E-mail address: makino-tiz@sumitomometals.co.jp (T. Makino).

0013-7944/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2009.12.013
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 811

Nomenclature

d diameter of axle body


D wheel seat diameter
D/d diameter ratio
V train velocity
aV vertical acceleration coefficient
aL horizontal acceleration coefficient
req Miner’s equivalent stress
KI mode I stress intensity factor
KII mode II stress intensity factor
Kh,max stress intensity factor in the maximum tangential stress direction
h angle of the maximum tangential stress
Kth threshold stress intensity factor
DKth threshold stress intensity factor range
a crack depth
rpm permissible stress
rw fatigue strength
Sf safety factor
FDL dynamic load factor
MX bending moment between rolling planes at the center of wheel seat
P1 vertical force on the more heavily loaded journal
P1,V vertical force accelerated by vertical vibration
P1,H increase of vertical force by the balance of rotary moment and the horizontal force.
Y1 wheel/rail horizontal force perpendicular to the rail on the side of the more heavily loaded journal
b half distance between vertical force input point on axle journals
s half distance between wheel rolling circles
R nominal radius of the rolling circle of a wheel
h1 height above the axle center of gravity of masses carried by the wheelset
F1 force exerted by the masses of the unsprung elements situated between the two wheels
m1g force on journals per wheelset
rb nominal bending stress at wheel seat
m load magnification factor
d0 bore diameter of hollow axle
Sf,IH safety factor for induction hardened axle

Subscripts
V vertical force
H horizontal force
JIS Japanese standard
EN European standard

In the present paper, the fatigue tolerance of the high-speed railway axle in Japan is reviewed. To maintain the safety, the
fatigue strength of the axle has been extensively studied. Theses case histories and consequent improvements in manufac-
turing process are presented. The crack propagation behavior of the induction hardened axle is studied based on the fracture
mechanics. Concerning the powered railway axles, the fatigue design method in Japan is compared with that in Europe and
the effect of the train velocity on the allowable load is discussed.

2. Fatigue design method of railway axle in Japan

The railway axle in Japan was designed for fatigue on the basis of a standard enacted in 1970. According to this old stan-
dard, the maximum velocity was 210 km/h. The standard [1] was revised in 1995, and a new SA class was provided that is
available for running at a velocity of 350 km/h. It was recognized that the actual bending stresses are very low in a high-
speed railway system because of adoption of high level maintenance of the track and the use of high running performance
bogies.
The fatigue design method in Japan specifies, as shown in Fig. 1, how to calculate the bending stresses incorporating with
the accelerated gravity force due to the vibration of the body. The axle diameter is, then, chosen so the maximum stress is
lower than the fatigue strength. This method is basically the same as the method specified by European Standard [2].
In Japanese Industrial Standard (JIS), railway systems are classified into two systems and the dependency of the acceler-
ation forces on the train velocity is standardized separately in each system as shown in Table 1 and Fig. 2. As shown in the
812 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

Fig. 1. Axle design method in Japan specifies [1].

Table 1
Classification of railway system and acceleration force.

Railway system Applied division Velocity, V (km/h) Vertical acceleration, aV Horizontal acceleration, aL
System 1 SA 200–350 0.0027 V 0.030 + 0.00060 V
A 150–280 0.0027 V 0.030 + 0.00085 V
System 2 A 60–160 0.0027 V 0.040 + 0.0012 V
<60 0.16 0.11
B 60–130 0.0052 V 0.060 + 0.0018 V
<60 0.31 0.17

SA: The especially improved track for the velocity 200–350 km/h.
A in system 1: Conventional high-speed railway line for the velocity 150–280 km/h.
A in system 2: The specially improved track in the conventional line.
B: Conventional track line.

figure, the acceleration force is less dependent on velocity than conventional low speed rail. The running tests were per-
formed at a velocity of over 250 km/h and it was recognized that the bending stresses are very low in the high-speed railway
system.
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 813

0.4

Horizontal acceleration
0.3 B

coefficient, αL
A SA

0.2 A

0.1
System 1
System 2
0
0 100 200 300 400
Velocity, V (kmh-1)

High-speed railway line: SA in System 1.... the specially improved tracks.


A in System 1 ..... the usual high-speed railway line.
Conventional railway line: A in System 2 .... the specially improved trucks
B in System 2.... the conventional track line

Fig. 2. Horizontal acceleration coefficient.

Fig. 3. Typical railway axle for Shinkansen [3]. (a) 0 Series solid axle (Maximum velocity: 210 km/h) and (b) 500 Series hollow axle (maximum velocity
300 km/h).

Fig. 4. Detail shape of press-fitted part.

3. Critical part of axle for fatigue

Fig. 3 shows the typical Shinkansen axle design [3]. Fig. 4 shows the detail shape of the press-fitted part. Table 2 shows
the diameter ratio and the fillet radius [3]. As shown in Figs. 3 and 4 and the values of Shinkansen in Table 2, the diameter
814 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

Table 2
Diameter ratio of Japanese and European axles [3].

Dimension Shinkansen Europe


0-Series 500-Series TGVa ICEa
Wheel-seat diameter D (mm) 209 194 212 200
Axle body diameter D (mm) 190 174 184 173
Diameter ratio D/d 1.10 1.11 1.15 1.16
Radius r (mm) 100 100 75/15 75/15
a
Typical example.

ratio, D/d, of wheel seat diameter D to body diameter d is about 1.1 below 1.2. As a result, the critical part of fatigue strength
of the axle becomes the press-fitted part such as the wheel seat and the gear seat because the fatigue strength of these parts
are lower than that of the other plain parts owing to the stress concentration from the press-fitted shape of axle and the
effect of fretting fatigue. Consequently, in order to increase the fatigue strength of the press-fitted part of the axle, induction
hardening of the axle was adopted.
On the contrary to this, the diameter ratio D/d of European railway axles, also shown in Table 2, is larger than 1.12. There-
fore, the critical part for fatigue of the European railway axle is not the wheel seat or the gear-seat but the fillet or the groove
of axles.

4. Countermeasure against fretting fatigue

4.1. Characteristic of fretting fatigue

Fretting fatigue is a typical high cycle fatigue due to the cyclic minute relative slip between the axle surface and the inner
surface of wheel hub, which initiates microcracks and wear, as shown in Fig. 5. As far as the cyclic relative slip continues,
cracks can initiate. It is necessary to carry out fatigue testing for more than 108 cycles to find the fatigue strength under fret-
ting conditions. Fig. 6 shows the results of fretting fatigue tests using press-fitted axle assemblies of 50 mm axle diameter
made of normalized carbon steel (0.38% C). As shown in the S–N curves, fatigue failures at lower cycles under higher stresses
occur at the fillet of axles, however, fatigue failures at the press-fitted part could occur at higher cycles of more than 107
cycles under the lower stresses.
Fig. 7 [4–6] shows the results of the rotating bending fatigue testing of the induction hardened press-fitted £40 mm axle
assemblies. In the test, to clear the effect of variable amplitudes on the fatigue crack initiation, each specimen was subjected
to variable stress amplitudes of two, three and ten step-levels. Stress amplitudes in almost test conditions were specified in
the range from 39 MPa to 118 MPa except one test condition. Here, upper limit (118 MPa) in this range is equivalent to fa-
tigue limit regarding minute crack initiation. The S–N curves are presented by the Miner’s equivalent stress req. As shown in
the figure, no axle was broken by fatigue at this level of stress amplitude. However, minute fretting fatigue cracks were ob-
served at more than 108 cycles. The depths of these cracks were less than 0.1 mm and are considered as non-propagating
fatigue cracks.
The size effect on the fatigue crack initiation of induction-hardened press-fitted axle is shown in Fig. 8. As shown in the
figure, the fatigue strength, concerning crack initiation, decreases as the size of the axle increased. The fatigue strength
against crack initiation of full scale axle is estimated as 60 MPa at 108 cycles [7,8]. It is noted that 108 cycles corresponds
to 250,000–300,000 km of train running.

Fig. 5. Fretting fatigue cracks at the wheel-seat.


T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 815

200
Fatigue fracure Fretting fatigue
* fracture

Bending stress, σa (MPa)


* *
*
*
150 *
*

r
ρ mm d mm
5.0 46.8
100 7.0 46.8
* Broken at fillet
6
10 107
Number of cycles to failure, Nf

Fig. 6. Typical features of fretting fatigue.

S-N curve under constant


200 stress amplitude
[MPa] eq
Equivalent stress

100
90 40mm press-fitted axle
80 , Two-level loadings
70 , Three-level loadings
60 , Ten-level loadings
Solid:Crack initiated
50 Open:Not initiated
40
105 106 107 108
Total number of cycles ni

Fig. 7. Effect of variable stresses on the fatigue strength of crack initiation.

300 Dia.40
Equivalent bending stress, (MPa)

Dia.150
200 Full-scale
(Dia.209)
Bending stress /

100

Fatigue Load
Constant Variable
50 Uncracked
Cracked

104 105 106 107 108


Number of cycles, N
Fig. 8. Size effect on the fatigue crack initiation of the induction hardened press-fitted axles.

4.2. Induction hardening of the axle

Shinkansen axles are induction hardened to increase the fatigue strength of the wheel seat and the gear seat by inducing
compressive residual stresses and by increasing the surface hardness. The axle for Shinkansen is made of 0.38% carbon steel.
After forging and machining, the entire axle length is induction hardened as shown in Fig. 9.
816 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

Fig. 9. Induction hardening method for Shinkansen axles.

400
Freq. Coil Temper
600 (kHz) turn -ing(K) 200
10 Single 436
Residual stress, (MPa)

10 Multiple 503
Hardness, HV

3 Multiple 473 0

400 -200
Freq. Coil Tempering
-400 (kHz) turn (K)
10 Single 436
10 Multiple 503
200 -600 3 Multiple 473

0 3 6 9 12 0 20 40 60
Depth below surface, (mm) Depth below surface, (mm)
(a) Hardness distribution (b) Residual stress distribution
Fig. 10. Hardness and residual distribution of the induction hardened axles.

The method of induction hardening has been amended to improve, in regard to, the residual stresses and hardness dis-
tribution [9]. Fig. 10 shows the change in the residual stress and hardness distribution by induction current frequency. As
shown in the figure, much improvement was obtained by lowering the current frequency. Table 3 shows the change in
the manufacturing method in terms of the frequency of the induction current, numbers of the heating turn coil and the tem-
pering temperature [9].
At the start of Shinkansen, the axles were made by specification O, as used in conventional railway locomotive axles. In
1970 the frequency of the current was changed from 10 kHz to 3 kHz to deepen the compressive residual stresses zone and
make it more compressive.
Finishing, such as grinding after the induction hardening, influences the surface residual stresses by local heating. Ini-
tially, in the specification of O and A in Table 3, the machining allowance of 0.5 mm was ground, but after 1965 the first
0.35 mm was machined by lathe turn and the final 0.15 mm has been ground by a softer grind-stone [7]. The tempering tem-
perature changes, which are shown in Table 3, were made expecting an improvement in decreasing the susceptibility to the
heating by the grinding. In 1990, the grinding processes were fully automatically controlled to avoid the local heating.
Fig. 11 shows the results of rotary bending fatigue testing using full size actual axle assemblies. The end of the wheel hub
was 6 mm over-hang from the wheel seat of the axle. As shown in the figure, the fatigue strength of quenched and tempered
axle is about 100 MPa, and the specification C induction hardened axle is 120 MPa, while the fatigue strength of specification
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 817

Table 3
Change in the manufacturing process of Shinkansen axles [9].

Specification Frequency (kHz) Coil turn Tempering temperature (K) Finishing


O 10 Single 463 Grinding (0.5 mm)
A 10 Single 503
B, C and D 10 Single 503 Grinding (0.15 mm) after Turning (0.35 mm)
E 10 6 503
F 10 6 473
G (1970) 3 6 473

200
Bending stress, σb [MPa]

150

100
Quench & tempered
Induction hardened: Spec. C
50 Induction hardened: Spec. G
Minute crack initiated
but not broken

106 107 108


Number of cycles
Fig. 11. Fatigue strength of the induction hardened press-fitted axles.

G axle is much improved, up to 180 MPa. The crack propagation behavior of the induction hardened axle will be studied in
following chapter.

4.3. Maintenance

The maintenance of Shinkansen axles is divided into two steps. After every 30,000 km of running, all axles are inspected
by an ultrasonic inspection using a hollow of the axles. After every 600,000 and 1,200,000 km of running, all the wheelsets
for Shinkansen are dismounted from the bogie, and axles are dismounted from the wheels [10]. The wheel seat is then sub-
jected to a magnetic particle inspection as well as an ultrasonic inspection. Crack depth, if any, is measured by polishing off
the surface of the axle until the crack is removed. When a crack deeper than 0.15 mm, are detected, the axle would be re-
placed. Regarding ballast impact, corrosion and surface flaw deeper than 0.15 mm, the same procedure would also be per-
formed. However, even if the detected crack is shallower than 0.15 mm, when a crack of any depth is detected again at the
same wheel seat on the next timing of inspection, the axle would be replaced.
The number of axles replaced, owing to the detection of minute fatigue cracks on the entire axle by the above inspections,
was none in 1991, four in 1992 and one in 1993 from a total number of ca. 20,000 axles inspected each year [11,12]. Since
then fatigue cracks have hardly been detected over 20 years.
Fig. 12 shows the relative number of the axle replacement since the start of Shinkansen due to the flaws at the wheel seat
detected by the inspection [13]. As shown in the figure, much improvement is seen as the manufacturing processes have
been improved.

5. Fracture mechanics approach on the crack propagation of induction hardened axle

Fracture mechanics is applied to the evaluation of the crack propagation behavior of the induction hardened axle.

5.1. Fatigue test of induction hardened axle with ring-shaped crack

Induction hardened axles with ring-shaped cracks were evaluated by rotating bending fatigue tests [14]. The material of
axle specimen is 0.38% carbon steel. The axle specimens were prepared by the following procedure. Firstly, ring-shaped crack
is pre-introduced into the ring-shaped notch of the cylindrical specimen shown in Fig. 13 by rotating bending fatigue test.
Next the outer surface layer including notch of specimen is removed by machining and the axle specimen is obtained. Then
818 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

Relative number of axles replaced


1.0

0.8

(1.0 : 8 %)
0.6

0.4
All axles in service
were replaced by
0.2
Spec.G axles

0
1975 1985 1995
Fig. 12. Relative number of replaced axles due to fretting fatigue crack [13].

62

51
0.5R 60
Fig. 13. Geometry of cylindrical specimen with ring-shaped notch.

Contact edge

1.7
Boss Ring-shaped crack
Axle
75

50
154

50

80
320

Fig. 14. Configuration of press-fitted axle specimen with ring-shaped crack.

ring-shaped crack of 0.5–0.6 mm depth remains at the surface of axle. Here the depth of crack is measured from fracture
surface obtained by following fatigue test. After the machining, the axle specimen is induction hardened and press-fitted into
the boss specimen made of wheel steel. Fitting clearance is determined with the contact pressure being 70 MPa. Fig. 14
shows the specimen configuration after being press-fitted. The ring-shaped crack is located at 1.7 mm inside from the con-
tact edge in the press-fitted region.
Rotating bending fatigue tests were carried out with the above-mentioned press-fitted axle specimen. Fig. 15 shows the
S–N diagram evaluated by nominal bending stress amplitude. Fractures of specimens were originated from pre-introduced
ring-shaped crack. Fatigue limit was determined as 261 MPa from Fig. 15.

5.2. Stress intensity factor of press-fitted axle

Stress intensity factor of ring-shaped crack existing in press-fitted axle specimen was calculated by finite element (FE)
analysis using ABAQUS. Fig. 16 shows the FE model and the boundary condition for analysis. The model consists of two parts,
i.e. boss and axle. Both parts are composed by axisymmetric solid elements with nonlinear, asymmetric deformation. Fitting
clearance is also considered in the model as the same in the actual specimen. Press-fit pressure of 70 MPa (nominal value)
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 819

400

Nominal stress amplitude (MPa)


350

300

250

200
a=0.5mm
150 a : crack depth
a=0.6mm
100
105 106 107 108
Number of cycles to failure N f

Fig. 15. S–N diagram for press-fitted axle specimen with ring-shaped crack.

Press-fitted surface (friction coefficient:0.6) Point load


Boss
Axle Rigid beam element

607
Crack

Fig. 16. FE model and boundary condition for analysis.

generates by defining the contact at press-fit surface on the analysis. Friction coefficient at the press-fitted surface is consid-
ered with a value of 0.6 [15,16]. A crack of 0.5 mm or 0.6 mm in depth is introduced on the press-fitted surface of axle. The
region around crack tip is meshed in a radiate shape for considering singularity. Neither contact at crack-face nor residual
stress are defined. The model is loaded by cantilever bending moment. Then the end face of the model, i.e. left-hand side of
axle and boss in Fig. 16 is restrained in axial direction. The end face of axle is also restrained in radial direction. The restric-
tion condition is considered as valid because the axle was not extracted from the boss at all after the fatigue test. Material
properties are specified as Young’s modulus of 205800 MPa and Poisson’s ratio of 0.3.
Modes I and II stress intensity factors KI, KII were calculated from the opening and shearing displacement between crack
face obtained by FE analysis. The stress intensity factor in the maximum tangential stress direction Kh,max was calculated by
using following equation based on mixed mode criterion by Erdogan–Sih [17]:
    
h h 3
K h;max ¼ cos K I cos2  K II sin h ð1Þ
2 2 2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 2
3K II  K I K I þ 8K II
cos h ¼ ð2Þ
K 2I þ 9K 2II

where h is the angle of maximum tangential stress to a circle around crack tip, if KII > 0, h < 0, if KII < 0, h > 0.
Fig. 17 shows the relationships between Kh,max and nominal bending stress. Solid lines indicate the stress intensity factor
of smooth cylindrical specimen, which is not press-fitted, obtained from handbook[18]. Plots by FE analyses are a little (1.03
times) higher than the solid lines in each crack depth. Reason why the increase of the stress intensity factor is very low is that
there is a gap of locations between the pre-introduced crack and the maximum stress. According to FE analysis result by a
model not defined crack, the maximum stress generates at the contact edge where the stress concentration factor (SCF) is
820 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

16
a=0.5mm,FEA
15
a=0.6mm,FEA 5mm
a=0.

m)
14 a : crack depth

Kθ,max, KI (MPa
6mm
13 a=0.

12
Smooth cylindrical specimen
11 (No press-fitted)

10
250 260 270 280 290 300 310
Nominal bending stress amplitude (MPa)
Fig. 17. Relationships between Kh,max and nominal bending stress amplitude.

2.66. On the other hand, SCF on the press-fit surface at 1.7 mm from the contact edge, equivalent to the location of crack, is
1.14. Obviously SCF below the press-fit surface is smaller than that on the surface because of stress decreasing in radial
direction.

5.3. Evaluation result by fracture mechanics

The fatigue test result of press-fitted axle with ring-shaped crack was evaluated by fracture mechanics. Fig. 18 shows the
relationship between Kh,max and the number of cycles to failure. The values of Kh,max in the figure are obtained from the above
analysis results. According to the figure, crack growth threshold equivalent to fatigue limit expressed by Kh,max is
p
12.6 MPa m. This value is supposed to be the threshold stress intensity factor Kth. Kth of the axle steel not being induction
p p
hardened, is 6.2 MPa m [4], 7.39 MPa m [16]. Kth of induction hardened press-fitted axle obtained from the fatigue test is
almost two times of Kth of the axle steel. The increase of Kth is caused by compressive residual stress due to induction
hardening.
On the other hand, Makino et al. evaluated the semi-elliptical cracks initiated in press-fitted axles after fatigue test by
fracture mechanics [15]. Fig. 19 shows the relationships between the threshold stress intensity factor range DKth and crack
depth a. In the figure, the previous results [15] are plotted as solid marks and the present result is plotted as an open mark.
Then DKth is defined as including the range under not only tensile but also compressive stress. Then KI under compressive
stress due to bending is assumed to be negative value considering that crack faces do not contact and overlap. Accordingly
p
the present result of DKth becomes 23.4 MPa m. The assumption of negative value of KI makes reasonable for considering
the opening behavior of small crack under compressive stress and discussing the effect of mean/residual stress on the prop-
agation behavior of small crack. According to this approach, the effect of residual stress apparently provides the change of
DKth. Despite of the difference of crack shape, the plot of the present result locates at the region extended from the tendency
of the plots for induction hardened axle from previous results in Fig. 19. DKth of induction hardened axle is about two times
higher than quench-tempered axle. This suggests that the induction hardened axle has high performance against fatigue
damage, even if it has some small defects.

15

14
m)
Kθ,max (MPa

13

12 a=0.5mm
a=0.6mm a : crack depth
11
105 106 107 108
Number of cycles to failure N f

Fig. 18. Relationship between Kh,max and the number of cycles to failure.
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 821

50
Quench-tempered axle
, Induction hardened axle

m)
10

Kth (MPa
5

1 Solid:Semi-elliptical crack[15]
Open:Ring-shaped crack
0.5
1 5 10 50 100 500 1000
Crack depth a ( m)

Fig. 19. Relationship between DKth and crack depth.

6. Fatigue design of the axles for Shinkansen

In this chapter, the safety factors of Shinkansen axles are calculated based on the European standards. Allowable stresses
specified in JIS are used for the calculation. Here ‘‘Allowable stress” in JIS is equivalent to ‘‘Fatigue limit” in European stan-
dards. In the present paper, ‘‘Allowable stress” in JIS is also called ‘‘Fatigue limit” to avoid confusion.

6.1. Fatigue limit and safety factor in JIS and European standards

Fatigue limits of axles are specified in JIS as well as in European standards. Fatigue limit of the axle for high-speed railway
is compared with those of European axles in Table 4. Here, fatigue limit is supposed to be lower than actual fatigue limit.
Permissible stresses, which are the maximum permissible amplitude stresses to be used in design, shown in the table are
derived from following equation:
rw
rpm ¼ ð3Þ
Sf
Critical part in JIS is wheel seat, on the other hand, that in European standard is fillet and groove of axle. From the table,
the permissible stresses at each critical part are in almost the same level.

6.2. The dynamic load factor used in JIS

The safety factor in JIS is specified 1.0 as the minimum value as shown in Table 4. However, the safety factor contained a
marginal factor, as explained as follows: as stated above, JIS specified the acceleration factors to account for the dynamic
effect due to the train vibration. Considering the high-speed train velocity V [km/h] being 200–350 km/h, the acceleration
factors become as follows:

Table 4
Comparison of the permissible stresses (powered solid axle with press-fit).

Standard
JIS EN13104
Material Induction hardened carbon steel EA1 N carbon steel EA4T alloy steel
Evaluation part Wheel seat
Fatigue limit rw 147 MPaa 120 MPa 145 MPa
Safety factor, Sf b 1.0 1.5 1.66
Permissible stress, rpmc 147 MPa 80 MPa 87 MPa
Evaluation part Fillet and groove of axle
Fatigue limit, rw Not specified. 200 MPa 240 MPa
Safety factor, Sfb Not specified. 1.5 1.66
Permissible stress, rpmc Not specified. 133 MPa 144 MPa
a
Described as ‘‘Allowable stress” in JIS.
b
The minimum value.
c
The maximum value.
822 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

aV ¼ 0:0027V ð¼ 0:540—0:945Þ ð4Þ


aL ¼ 0:030 þ 0:00060V ð¼ 0:15—0:24Þ ð5Þ
where aV is the vertical acceleration coefficient and aL is the horizontal acceleration coefficient.
Outline of design method in JIS, shown in Fig. 1, is described more specifically. Fig. 20 shows the dimensions of wheelset
in European standard [2]. According to JIS (System 1 SA), the bending moment in the rolling plane, Mx can be converted by
using the dimension of the figure and be expressed as follows [1]. For the studied axle this could be near the center of the
wheel seat. However, the crucial location is not in this location. Strictly, the area at the wheel seat near to the axle body
should be studied since fretting would occur at this location.
  
m1 g h1
MX ¼ m ð1 þ aV Þ þ aL m1 g  ðb  sÞ þ aL Rm1 g
2 2b
m1 g h1 m1 g
¼ mð1 þ aV Þðb  sÞ þ maL ðb  sÞ þ maL Rm1 g ð6Þ
2 b 2
where m1g is the force on journals per wheelset, m is load magnification factor for powered axle. m is defined as 1.2–1.3 for
powered axles with parallel cardan driving devices. The first term in Eq. (6) means the bending moment by the vertical force
MX,V and the second and third terms mean the moment by the horizontal force MX,H, which are expressed as follows:
m1 g
MX;V ¼ mð1 þ aV Þðb  sÞ ð7Þ
2
h1 m1 g
MX;H ¼ maL ðb  sÞ þ maL Rm1 g ð8Þ
b 2
On the other hand, the bending moment at the wheel seat under static load MX,st is estimated as follows:
m1 g
MX;st ¼ ðb  sÞ ð9Þ
2
If the dynamic load factor FDL is specified as the ratio of MX,V and MX,st, FDL is expressed as follows:
M X;V
F DL ¼ ¼ mð1 þ aV Þ ð10Þ
M X;st
When the dimensions are taken as shown in Table 5, and substituting Eqs. (4) and (5) into Eq. (10), FDL becomes 1.85–2.53.
Moreover, MX,H becomes the range of 0.115m1g to 0.200m1g.

6.3. The dynamic load factor used in European standard

The bending moment between rolling planes, MX, at the center of wheel seat can be expressed as follows according to
European standards [2]:

b b h1

P1 F1 P2
H
yi
R
Y1 Y2
s s
Q1 Q2

Fig. 20. Dimensions of wheelset in European standard [2].

Table 5
Dimensions of the axles.

2b (m) 2s (m) h1 (m) 2R (m)


2.1 1.5 1.47 0.86
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 823

MX ¼ P1 ðb  sÞ þ RY 1 ð11Þ
P1 and Y1 are expressed as follows [2]:
    
h1 h1
P1 ¼ 0:625 þ 0:0875 m1 g ¼ 0:625m1 g þ 0:0875 m1 g ð12Þ
b b
Y 1 ¼ 0:35m1 g ð13Þ
where the dimensions of wheelset shown in Fig. 20 are used. F1 in the figure is here specified as zero since no brake disks or
powering devices are considered. The first term in Eq. (12) means the vertical force P1,V magnified by the vertical vibration and
the second term means the increase of vertical force P1,H by the balance of the rotary moment and the horizontal force. There-
fore, the bending moment by the vertical force MX,V and the moment by the horizontal force MX,H are expressed as follows.
m1 g
MX;V ¼ P 1;H ðb  sÞ ¼ 1:25ðb  sÞ ð14Þ
2
h1
MX;H ¼ P 1;V ðb  sÞ þ RY 1 ¼ 0:0875 ðb  sÞm1 g þ 0:35Rm1 g ð15Þ
b
If the dynamic load factor FDL is specified as the ratio of MX,V and MX,st, FDL is expressed as follows by substituting Eq. (9).
MX;V
F DL ¼ ¼ 1:25 ð16Þ
M X;st
When the dimensions are also taken as shown in Table 5, MX,H becomes the range of (0.187 m)  m1g.

6.4. Comparison of the dynamic load factor and the bending moment between JIS and European standard

Table 6 summarizes the dynamic load factor estimated in above sections. It is recognized that the dynamic load factor
estimated from JIS is larger than that from European standard. On the other hand, the bending moment at the center of wheel
seat by the horizontal force estimated from European standard is in the range of that from JIS. Therefore, it is necessary to
compare the total bending moment by the vertical and horizontal forces.
The total bending moment between rolling planes, MX,JIS, at the wheel seat according to JIS (System 1 SA),substituting the
axle dimensions given in Table 5 into Eq. (6), is expressed as follows:
MX;JIS ¼ ð1:2—1:3Þ  f0:15ð1 þ aV Þ þ 0:64aL gm1 g ð17Þ
On the other hand, the total bending moment between rolling planes, MX,EN, at the wheel seat according to European stan-
dard, substituting the axle dimensions given in Table 5 into Eq. (11), is expressed as follows:
MX;EN ¼ 0:375m1 g ð18Þ
Fig. 21 shows the relationships between the bending moment MX,JIS (Eq. (17)), MX,EN (Eq. (18)) and the train velocity. As
shown in the figure, MX,JIS increases as the velocity increases; however MX,EN is constant. It is recognized that the bending
moment calculated from both standard takes the same value at V = 150–180 km/h. This result indicates that the JIS estimates
the higher stresses over the velocity of 150–180 km/h and the lower stresses below the velocity of 150–180 km/h than those
estimated from the European standard.

6.5. Calculation of the safety factor for Japanese axles based on European standard

The bending stress rb can be obtained from the following equation:


32M D
rb ¼ 4 X 04
ð19Þ
p D d

Table 6
Dynamic load factors and bending moments by horizontal force.

Standard
JIS E4501 EN13104
Material Induction hardened carbon steel EA1 N carbon steel EA4T alloy steel
Dynamic load factor, FDL 1.85–2.53a 1.25b
Bending moment at wheel seat by vertical force 0.278m1g to 0.380m1ga 0.188m1gb
Bending moment at wheel seat by horizontal force 0.115m1g to 0.200m1ga 0.187m1gb
a
Estimated from JIS [1] at train velocity being 200–350 km/h (System 1 SA) for powered axles with parallel cardan driving devices, of which dimensions
given in Table 5.
b
Estimated from European standard [2] for powered axles of which dimensions given in Table 5.
824 T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825

0.50
0.48
0.46 1.3
0.44 m=
JI
S 1.2

MX/m1g (m)
0.42
m=
0.40
0.38 European standard
0.36
0.34
V=150-180km/h
0.32
0.30
100 150 200 250 300
Velocity V (km/h)
Fig. 21. Comparison of MX from European standard and JIS.

where D is the wheel seat diameter of axle and d0 is the bore diameter of hollow axle. The safety factor for induction hard-
ened axle Sf,IH can be calculated by the following equation:

147 MPa
Sf ; IH ¼ ð20Þ
rb
where the value of ‘‘147 MPa” in Eq. (20) is the fatigue strength for the induction hardened axle specified in JIS.
The safety factors for Japanese axles based on European standard are tentatively calculated from Eqs. (18)–(20). Then the
axle load is taken as W ton, and the weight of the wheelset is assumed as 1000 kg, m1g in Eq. (10) becomes 9.807(W – 1) kN.
The calculation results of the safety factor are shown in Fig. 22 and in Table 7. These values are the same as those from JIS at

5
4.5 Solid axle,
calculated
4 Axle diameter:
in Table 7
209mm
Safety factor S f,IH

3.5
3
2.5
2
1.5 Hollow axle,
1 Axle diameter: 194mm
0.5 Bore diameter: 60mm
0
0 5 10 15 20
Axle load W (ton)
Fig. 22. Safety factors obtained from European standard for geometries of Shinkansen axles.

Table 7
Comparison of safety factor for Japanese and European axle.

Type
Japanese axle European axle
Material Induction hardened carbon steel EAIN carbon steel EA4T low alloyed steel
Axle load, W (ton) 16 11 – –
Axle diameter, D (mm) 209 194 – –
Bore diameter, d’ (mm) 0 60 – –
Safety factor 2.39a 2.58a 1.5b 1.66b
a
Calculated from bending moment based on European standard and the fatigue strength (allowable stress) specified in JIS.
b
Specified in European standard.
T. Makino et al. / Engineering Fracture Mechanics 78 (2011) 810–825 825

V = 150–180 km/h. As shown in the table, the safety factors of Japanese axles are 2.39 and 2.58, larger than those specified in
European standard.

7. Concluding remarks

To maintain the safety record of railway systems, much effort has been paid to the improvement of the axle manufactur-
ing processes, design, testing and maintenance. As stated, the safety of the railway has been ensured by maintenance, such as
regular inspection for fatigue cracks at critical parts.
Fracture mechanics was applied to the evaluation of the crack propagation behavior of the induction hardened axle.
Induction hardening axle indicated high threshold stress intensity factor. This fact suggested the high performance of induc-
tion hardened axle against fatigue damage.
Axle fatigue design method in Japan and Europe was reviewed in the paper. It can be seen that a complete difference ex-
ists in the concepts about the critical part of axles for fatigue. However, both standards applied almost the same bending
moments at the velocity of 150–180 km/h. This fact suggests that the JIS estimates higher stress beyond the velocity of
150–180 km/h or lower stress below 150–180 km/h than European standard. Moreover the safety factors of Shinkansen
axles calculated by European standard become larger than those specified in European standard. This result suggests that
Japanese axles are normally more conservatively designed than European axles.
The effect of variable amplitude stresses on the fatigue life in the long life range is not fully understood [4–6]. The fatigue
design method can be, therefore, improved. Comprehensive investigation in this area is still necessary.

References

[1] Japanese Industrial Standard, E4501. Railway rolling stock-design method for strength of axles; 1995 [in Japanese].
[2] European Standard EN-13104. Railway applications-wheelsets and bogies-powered axles-design method, 13104; 2001.
[3] Hirakawa K. High speed railway axles. J Jpn Soc Mech Eng 1998;101:146–8 [in Japanese].
[4] Makino T. Fretting fatigue strength of press-fitted railway axles, Doctor Thesis Kyushu, University; 1999 [in Japanese].
[5] Yamamoto M, Makino T, Hirakawa K. The effect of variable load on fatigue properties of press-fitted axles. In: Proceedings of 11th international
wheelset congress, vol. 1; 1995. p. 29–34.
[6] Makino T, Yamamoto M, Hirakawa K. The fretting fatigue crack Initiation behavior at press-fitted axle assembly with variable stress loading (effects of
stress cycle ratio, number of load levels and axle size). In: Proceedings of 12th international wheelset congress; 1998. p. 147–52.
[7] Hirakawa K, Toyama K, Kubota M. The analysis and prevention of failure in railway axles. Int J Fatigue 1998;20:135–44.
[8] Hirakawa K, Kubota M. On the fatigue design method for high-speed railway axles. Proc Instn Mech Engrs, Part F 2001;215:73–82.
[9] Isomura S, Yomoda K. Manufacturing history of axles for Shinkansen. In: Proceedings of 11th international wheelset congress, vol. 2; 1995. p. 51–4.
[10] Ishizuka H. Non destructive testing of the bogie frame and axles. In: 28th Fatigue course ‘‘The basics and application of fatigue and fracture”. The Japan
Society of Material Science; 2008. p. 48–60 [in Japanese].
[11] Ishizuka H. Thirty years of axle maintenance for Japanese bullet train-Shinkansen. In: Proceedings of 11th international wheelset congress, vol. 2;
1995. p. 43–6.
[12] Yohso J. Development of automatic ultrasonic testing equipment for general and bogie inspection of Shinkansen hollow axle. In: Proceedings of 11th
international wheelset congress, vol. 2; 1995. p. 47–52.
[13] Ishizuka H. Probability of improvement in routine inspection work of Shinkansen vehicle axles. QR of RTRI 1999;40:70–3.
[14] Tanaka S, Komatsu H, Nishioka K, Hirakawa K. Fatigue strength of press-fitted axles. Proc JSME 1976;760(13):264–6 [in Japanese].
[15] Makino T, Yamamoto M, Hirakawa K. Fracture mechanics approach to the fretting fatigue strength of axle assemblies. Fretting fatigue: current
technology and practices, vol. 1367. ASTM STP; 2000. p. 509–22.
[16] Madia M, Beretta S, Zerbst U. An investigation on the influence of rotary bending and press fitting on stress intensity factors and fatigue crack growth
in railway axles. Eng Fract Mech 2008;75:1906–20.
[17] Erdogan F, Sih GC. J Basic Eng Trans ASME 1963;1:519–27.
[18] Murakami Y et al., editors. Stress intensity factors handbook, vol. 2. Berlin: Pergamon Press; 1987. p. 657–58.

You might also like