International Journal of Fatigue 118 (2019) 262–270

Contents lists available at ScienceDirect

International Journal of Fatigue

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

Fatigue crack growth in wheelset axles under bending and torsional loading T
⁎
R. Hannemann , P. Köster, M. Sander
Institute of Structural Mechanics (StM), University of Rostock, Albert-Einstein-Str. 2, 18059 Rostock, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: In the past, unexpected failures in wheelset axles lead to catastrophic consequences. To avoid those failures
Rotating bending inspection intervals must be defined. The definition of the inspection intervals is based on fracture mechanical
Mixed mode loading approaches. In the present work, the railway axle steel 34CrNiMo6 is investigated. For this material fatigue crack
Wheelset axles growth data have been experimentally determined for a range of different negative and positive stress ratios.
Constant amplitude loading
Furthermore, the FORMAN/METTU parameters for different probabilities of survival are identified. In wheelset
2010 MSC:: axles different geometry factors influence the crack growth behaviour. Therefore, different shouldered solid
00-01 shafts with transition radii and a press fit seat are designed and tested under rotating bending. Moreover, the
99-00
load situation also has an influence on the crack growth. The main bending load is overlapped with torsional
loading at special wheel-rail-situations. With the developed mixed mode test rig, the influence of different load
situations on wheelset axles are examined.

1. Introduction been developed. This test rig is able to subject specimen with different
diameters and length to constant and variable amplitude loading.
Railway axles are safety-critical components. The lifetime of The majority of the disastrous fatigue failures in driving axles are
wheelset axles is commonly 30 years or more. Unexpected failures can not only related to the rotating bending but to the superposition of
cause a derailment with possibly disastrous consequences. In order to cyclic bending with torsion [14]. These mixed-mode loading conditions
ensure a safe operation, railway axles are designed with a safe life occur for example at cornering of trains or at deceleration and accel-
approach in combination with regular non-destructive inspections in eration processes. Numerous investigations in the literature show
defined intervals. Those measures shall ensure that the flaw or the crack clearly that the test conditions concerning the crack growth at mixed-
does not exceed a critical size. The relevant inspection intervals are mode loadings differ widely and lead to contradictory results and
commonly determined in the current practice on the basis of experience statements [15]. Furthermore, on the one hand, many experimental and
[1,2]. To determine inspection intervals, a reliable computed remaining numerical tests exist concerning the superposition of mode I and II or III
lifetime with fracture mechanical concepts can be used. Usually, the at thin-walled tube specimens e.g. [16–19]. On the other hand, a
crack propagation models are defined for thin sheet metal structures, variety of tests merely analyse crack initiation e.g. [14,20–24]. In this
e.g. from the aerospace industry. However, the models for remaining paper, the focus is on solid shafts. The purpose is to most accurately
lifetime calculations of thick-walled shaft structures, e. g. wheelset represent the crack propagation conditions of large cracks in wheelset-
axles are mainly investigated using mode I crack growth concepts axles concerning the superposition of all three crack-modes. Therefore,
[3–8]. During the operation, railway axles are exposed to rotating another special test rig was developed and first experimental results
bending with variable amplitudes. In the literature, different experi- could be achieved.
mental investigations on shafts under rotating bending can be found
(e.g. [2,9–13]). 2. Material
This paper presents the investigations on the influence of different
design details of wheelset axles, i.e. transition radii or press fits, on the In this paper the quenched and tempered high-strength steel
crack propagation behaviour. Furthermore, scaled specimens are used 34CrNiMo6 is exemplary investigated for railway axle steels. The ma-
in the investigations, with which a statistical analysis is possible in terial properties are listed in Table 1. Fracture mechanical parameters
contrast to the common component tests. In order to reproduce the are determined for positive and negative R-ratios. Therefore, SEN-spe-
main loading of railway axles, a 4-point rotating bending test rig has cimen (l = 230 mm, w = 50 mm and t = 10 mm) are used for negative

⁎
Corresponding author.
E-mail address: [email protected] (R. Hannemann).

https://doi.org/10.1016/j.ijfatigue.2018.07.038
Received 30 November 2017; Received in revised form 16 July 2018; Accepted 27 July 2018
Available online 29 July 2018
0142-1123/ © 2018 Elsevier Ltd. All rights reserved.
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Nomenclature U electrical potential drop

U0 initial electrical potential drop
A deepest point of a surface crack Y geometry factor, normalized stress intensity factor
A0 coefficient in the Newman crack opening function a crack depth, crack length
AB elongation at failure a0 intrinsic crack length
B surface point of a surface crack a/c aspect ratio
CFM material-dependent coefficient of the NASGRO equation c crack length
Cth parameter in Newman’s empirical function for describing d minimal diameter of the shaft specimen
the R-dependence of the threshold value da/dN crack growth rate
D maximum diameter of the shaft specimen l specimen length
D∗ diameter of the shaft specimen at the crack position nFM, p , q material-dependent exponent of the NASGRO equation
E Young’s modulus t specimen thickness
KIC fracture toughness for mode I w specimen width
KI,max maximum stress intensity factor for mode I αCF constraint factor
ΔK cyclic stress intensity factor αK stress concentration factor
ΔKI cyclic stress intensity factor for mode I γ crack opening function
ΔKI,th threshold value for mode I σ maximum principal stress in the transition radius of the
ΔKth threshold value shaft specimen
ΔKth,1 threshold value for R → 1 σF flow stress
Mb bending moment σmax maximum applied stress
N number of load cycle σN nominal stress in the minimal diameter section of the shaft
PS probability of survival specimen
R stress ratio (R = σmin/ σmax = Kmin/ Kmax ) ζ nominal radial interference of press fit
Rm tensile strength

R-ratios of R = [−0.5, −0.7, −1.0, 0]. The experiments were completed probabilities of survival are given in Table 2.
with crack growth results obtained by Richard and Sander [25] for the Moreover, the threshold values ΔKth for the different stress ratios
positive R-ratio 0.1, which were determined by standardized CT-spe- were determined in dependence of the probability of survival. The
cimens (w = 72 mm and t = 10 mm). For each stress ratio, three to four threshold value of fatigue crack growth is defined as the value, at which
experiments were performed. The statistical analysis of the crack pro- the crack growth rate asymptotically approaches zero. According to
pagation data was automated by using a MATLAB program, which was ASTM E647-11 [29] a regression line must be determined using five
developed at the Institute of Structural Mechanics [26]. In the first step data points in a double logarithmic diagram. This regression line has to
for each stress ratio, quantile curves were calculated for different be extrapolated to a fatigue crack growth rate of e.g. 10−7 mm/cycle.
probabilities of survival (PS) . Therefore, the crack propagation data But, Döker [30] has shown that the “true” threshold value is given at
were divided into intervals and for each interval a polynomial regres- lower fatigue crack growth rates so that another method has been ap-
sion function and a confidence interval for a pretended confidence plied. With the so-called DLR method all data points between
probability were calculated. The transition between the domains was da/dN = 10−8 mm/cycle and da/dN = 10−7 mm/cycle were plotted in a
smoothed and equidistantly spaced values were calculated using in- graph with linear scales. Subsequently, a regression line was de-
terpolation functions. For the statistical analysis of the crack propaga- termined using the least squares method and linearly extrapolated to
tion data, the probabilities of survival PS = [5%, 50% and 95%] were da/dN = 0 . A comparison of the experimental determined and by the
used. In the next step, the parameters of the FORMAN/METTU equation empirical NEWMAN equation [28]
[27] (also known as “NASGRO equation” [28])
(1 + C + ·R)
ΔKI,th p ⎧ ⎛ 1 − R ⎞ th
da
= CFM·⎡ ⎛
1−γ
⎞·ΔKI ⎤
nFM
·
(
1− ΔKI ) a
⎪ ⎝1−γ ⎠
⎪ (1 − A0 )(1 − R)·Cth+
R⩾0
KI,max q
dN ⎣ ⎝ 1−R ⎠ ⎦ (1− KIC ) (1)
ΔKI,th = ΔKth,1· · −
a + a0 ⎨ ⎛ 1 − R ⎞(1 + Cth·R)
⎪ ⎝1−γ ⎠
were fitted for each quantile curve. The automated adaption of the ⎪ (1 − A )(Cth+ − Cth−·R) R < 0
⎩ 0 (2)
fitting coefficients on the quantile curves was divided into two steps.
First, the threshold values were calculated for every stress ratio and
+ −
therewith the coefficients ΔK1, Cth and Cth were determined. In the
second step of the adaption, the coefficients CFM, nFM, p , and q were
calculated by use of a search algorithm with nested intervals and the
least square method. Fig. 1 shows the crack propagation data for the
stress ratio R = −0.5 and the fitted quantile curves for the chosen
probabilities of survival. The determined FORMAN/METTU curves for
all examined R-ratios are presented in Fig. 2 for a probability of survival
of 50%. The corresponding fitting parameters for all analysed

Table 1
Material properties for the high-strength steel 34CrNiMo6.
E [GPa] AB [%] Rm [MPa]
Fig. 1. Crack propagation data of R = −0.5 and the fitted quantile curves;
210 9 1200
Material: 34CrNiMo6.

263
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 2. Experimental data with FORMAN/METTU curves (PS = 50%) for positive
and negative R-ratios; Material: 34CrNiMo6 [36]. Fig. 3. Comparison of experimentally determined and with the NEWMAN’s
equation calculated threshold values.
calculated threshold values are presented in Fig. 3 for different stress
ratios. clips and the potential drop is measured with two soldered cables.
It can be noted that the analytically calculated threshold values are
in good agreement with the experimentally determined threshold va- 3.2. Specimen geometry
lues.
In order to investigate the influence of different structural details of
wheelset axles on the crack growth behaviour, two specimen types were
3. Investigations on rotating bending
designed with respect to the standard for design methods of railway
applications (DIN EN 13103 and DIN EN 13104 [31,32]). The specimen
In this chapter, the developed rotating bending test rig and scaled
type 1 represents the transition radius design detail of a wheelset axle
specimen are presented. Furthermore, the influence of different struc-
and specimen type 2 represents the press fit (Fig. 6). For specimen type
tural details of wheelset axles on the crack growth behaviour is shown.
1, the transition radius has been varied, i.e. different stress concentra-
tion factors were realized (Table 3). The stress concentration factor was
3.1. Experimental setup calculated by
σ 32·Mb
The developed rotating bending test rig is pictured in Fig. 4. The αK = with σN =
σN π·d3 (3)
inner view of the test rig is shown in Fig. 5. In order to achieve defined
rotating bending loads with constant and variable amplitudes, the test where σ is the maximum principal stress in the transition radius, Mb is
rig is mounted in a servohydraulic testing system (5) and a servo drive the applied bending moment on the specimen, and σN is the nominal
(1) was chosen as a rotational drive. With these components, an auto- stress in the minimal diameter section of the respective specimen with
mated testing system has been developed. The installation of the spe- the diameter d (Fig. 6).
cimen (6) in the test rig is done by a clamping system (7) consisting of For specimen type 2, the preassigned transition radii were un-
clamping collets (7.1) and clamping nuts (7.2). These are usually used changed, but a press fit has been realized with a clamping set.
in milling machines for tool clamping. With this system, a high preci- The initial notch for the crack growth is always located in the sec-
sion and resistance is guaranteed and an adaptation to different spe- tion planes with maximum bending stresses and the diameter D∗. The
cimen diameters is possible. Inside the two shaft housings (3), the drive insertion was done by using a picosecond laser. The crack depth of the
shaft (12) is connected with the servo drive over a metal bellows cou- initial notch is a = 0.5 mm (Table 3).
pling (10). This coupling allows an axial, lateral and angular dis-
placement. The drive shaft (12) is supported by two spherical roller 3.3. Crack shape development under rotating bending
bearings (15). Furthermore, in order to avoid axial stresses, a shifting of
the supporting shaft (9) and the two spherical roller bearings (15) is The investigation of the influence of different structural details of
enabled. To protect the inner components against dust and dirt, both railway axles on the development of the crack shape is performed by
housings (3) are equipped with a housing cover (13) and a seal ring rotating bending tests with constant amplitude. In order to visualize the
(14). The application of the load is done by the servo hydraulic actuator crack shape development, overloads were introduced in the constant
(5), which is connected with a bridge (8). Moreover, the housings (3) amplitude loading to produce arrest marks on the fracture surface.
are supported by a frame (2) and two ball bearings (4) for each housing Furthermore, these tests were also used for the calibration of the po-
to allow a bending of the specimen. tential drop technique on the different specimen types. Therefore, each
The measurement of the crack depth is done by the potential drop arrest mark with a crack depth a is mapped with a potential drop U
technique (Fig. 4b). Therefore, the test is interrupted for every crack (Fig. 7). Herein the potential drop U is normalized with the potential
depth measurement. The direct current is introduced with two alligator drop U0 of the initial notch and the crack depth a is normalized with the

Table 2
FORMAN/METTU parameters for the high-strength steel 34CrNiMo6.
PS (%) CFM n p q ΔKth,1 KIC + −
Cth αCF σmax
Cth
σF
[MPa m0,5] [MPa m0,5]

5 2.52·10−5 0.95 3.18 0.50 1.23 150 4.07 0.12 1.9 0.3
50 9.38·10−7 1.89 2.39 0.43 1.14 145 3.89 0.05 1.9 0.3
95 4.26·10−6 1.18 1.70 1.08 0.95 140 4.09 0.04 1.9 0.3

264
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 4. Set-up for rotating bending tests [36].

diameter D∗ of the specimen at the crack position. It can be observed factors (Fig. 9a). Furthermore, at higher stress concentration factors
that the small change in the transition radius does not have a major and a superimposed press-fit load, the crack front tilts in lower crack
effect on the potential drop. Therefore, the calibration curve is fitted for depth regions (Fig. 9b and c).
all specimens of type 1 with different transition radii. For a good fitting
of the calibration curve on the experimental data, the curve is divided 3.4. Experimental and analytical investigation on the residual lifetime
into two areas. The fist area represents the crack growth between
0 < a/ D∗ < 0.2 . This area is fitted by a third-order polynomial function: On the basis of the calibration curve (Fig. 7, Eqs. (4) and (5)), ro-
a U
3
U
2
U tating bending tests with constant amplitude loading were performed.
= 2597.1 ⎛ ⎞ −8486.5 ⎛ ⎞ + 9255.7 ⎛ ⎞−3365.9
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
The bending load was chosen so that the cyclic stress intensity factor for
D∗ U
⎝ ⎠
0 U
⎝ ⎠
0 U
⎝ 0⎠ (4)
the initial crack size in a round bar (αK = 1.0 ) is ΔK = 20 MPa m . This
The second area represents the crack growth between bending load was applied on all tested specimens and was not changed
0.2 < a/ D∗ < 0.64 and is also fitted by a third-order polynomial func- in all constant amplitude tests. The determined residual lifetime curves
tion: for the specimens with transition radii (specimen type 1) are shown in
3 2 Fig. 11 for various experiments.
a U U U
= 1.79 ⎛ ⎞ −6.62 ⎛ ⎞ + 11.05 ⎛ ⎞ + 0.2
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ For comparison reasons, Fig. 11 additionally shows analytical si-
D∗ ⎝ U0 ⎠ ⎝ U0 ⎠ ⎝ U0 ⎠ (5) mulation results with NASGRO [28]. The numerical determination of
Figs. 8 and 9 show the development of the semi-elliptical crack the necessary stress intensity factors for different crack depths and as-
shape for specimens with different transition radii (specimen type 1) pect ratios was done by the MVCCI-method and is presented in [35].
and specimens with a press fit design detail (specimen type 2). On all The stress intensity factor solutions of the appropriate solid shafts are
fracture surfaces, it can be observed that the crack becomes flatter shown in Fig. 10 for positions A and B on the crack front. Furthermore,
during the crack growth under rotating bending. This phenomenon is the FORMAN/METTU parameters of Table 2 (PS = 50%) are used in the
independent of the stress concentration factor and the superimposed analytical calculations; the results show the expected trend. For the
press fit loading. However, in rotating bending experiments on smooth lowest stress concentration factor, the residual lifetime is at its max-
round bars, Carpinteri et al. [33] also observed that the crack front imum. Consequently, as the stress concentration factor increases, the
becomes flatter under rotating bending. Furthermore, in higher crack residual lifetime decreases. The analytically calculated remaining life-
depth regions the crack shape changes in both specimen types from a times are 4% and 13% smaller for stress concentration factors 1.18 and
semi-elliptical to a wavy shape until the final failure shape is reverse 1.32 in comparison to the specimen with αK = 1.10 , respectively. Up to
semi-elliptical. now, the experimental results did not show the trend this clearly.
The effect of the stress concentration on the crack shape develop-
ment is visible in form of a tilting of the crack front, which was also 4. Initial tests focusing on overlapping bending and torsional
noted in experiments on sharp notched round bars by Jacoby [34]. It loading
can be observed that the crack front begins to tilt in higher crack depth
regions for specimens with a transition radius and a highstress con- In practical applications, many shafts are loaded under mixed-mode
centration factor (Fig. 8c). For specimens with a press fit design detail, instead of mode I conditions, which warrant further experiments.
the crack front already begins to tilt at lower stress concentration Therefore, next to the rotating bending test rig, an additional test rig for

Fig. 5. Components of the rotating bending test rig (CAD).

265
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 6. Example for design details of a wheelset axsle.

Table 3 comparison between the two curves (Fig. 14) shows a good match be-
Variation of the specimens. tween the numerical and experimental results.
Design detail Stress concentration factor αK Crack depth a [mm] However, the main focus of the additional test rig is investigating
overlapping of bending and torsional loading. Therefore, next to the
Transition radius 1.10, 1.18, 1.32 0.5 initial tests on pure cyclic bending, experiments for alternating bending
Press fit 1.13, 1.21, 1.37 and torsion were successfully realized. Digital image correlation (DIC)
was used to monitor the crack initiation and propagation at the be-
ginning of the test. In Fig. 15a, the relevant part surrounding the notch
overlapping bending and torsional loading was developed (Fig. 12). It
prepared with the speckle pattern as well as the displacement in the
enables investigations of the influence on the fatigue crack growth due
monitored area including the first parts of the crack propagation on the
to pure cyclic bending and cyclic torsion as well as their superposition
specimen surface can be seen; it also includes the first parts of the crack
with different phase shift angles.
propagation on the specimen surface. A 3D-model of this area rebuilt in
Its implementation required specimens to be designed in the form of
the evaluation software of the DIC is shown in Fig. 15b. The observed
solid hourglass-shaped round bars. The diameter is 65 mm in the sup-
shifting of the crack angle is typical for the change between bending to
ported area and 45 mm in the relevant test area, while the total length
torsional loading.
of the specimens is about 790 mm. In order to ensure crack initiation at
Evidence for the correct measurement of the DIC is given by the
a predefined location, a semi-elliptical starter notch with a depth of
final crack surface of the specimen (Fig. 16a) at the end of the experi-
1 mm was added in the symmetry plane of the hourglass shaped test
ment. As the crack surface is observed at the front and from a diagonal
area using CNC milling.
point-of-view towards the notch (Fig. 16b and c), the same angular
During the adjustment of the test rig and its measuring systems,
crack path as digitally monitored is clearly visible. However, the
experiments for the crack growth under pure bending loading were
measurement of the crack growth into the specimen is not possible
successfully completed. The fracture surfaces of three specimens after
using the DIC. Even the potential drop technique cannot be used re-
individual load sequences of different constant load increments are
liably because of the contact of both fracture surfaces as a result of the
shown in Fig. 13.
torsion.
The load increments of specimen 1 (Fig. 13a) were applied with an
R-ratio of R = −1 until a crack depth of nearly 20 mm. They were then
5. Conclusions
changed to R = 0.1 to prevent further crack closure effects. In contrast,
at specimen 3 (Fig. 13c), the R-ratio was R = −1 until the final fracture,
The crack growth behaviour of semi-elliptical surface cracks under
while the R-ratio at specimen 2 (Fig. 13b) changes continuously during
rotating bending and mixed mode loading has been investigated.
the load sequence. Nevertheless, the crack becomes flatter on all three
Furthermore, the influence of different design details of wheelset axles
fracture surfaces as observed under rotating bending.
on the crack shape development and residual lifetime has been ex-
As with previous experiments, the potential drop technique was
amined. The investigations can be summarized as follows:
used on the last specimen. Together with the appropriate crack depths
deduced from the arrest marks on the fracture surface of the third
(1) For a typical wheelset axle steel (34CrNiMo6), crack growth curves
specimen, a calibration curve has been created. A verification of the
with different negative and positive stress ratios have been
curve was realized with the aid of a FE-Modell in MARC/MENTAT. The

Fig. 7. Calibration curve divided in two areas for the crack depth measurement by the potential drop technique for specimens with a transition radius. Area 1:
0 < a/ D∗ < 0.2 , Area 2: 0.2 < a/ D∗ < 0.64 .

266
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 8. Crack shape development for specimen with transition radius.

Fig. 9. Crack shape development for specimen with press fit design detail.

Fig. 10. Geometry function Y for different stress concentration factors αK and aspect ratios (a/ c ) for the crack front points A and B; design detail - transition radius,
Loadcase – bending [35].

Fig. 11. Crack propagation behaviour for rotating

bending with constant amplitudes for different αK in
comparison to analytical simulation results.

267
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 12. Developed bending-torsion test rig with components: spherical roller bearings (1), lever (2), torsion shaft (3), Omniflex®-coupling (4), barrel roller bearing
(5), specimen (6), constructed adapters (7), half shell adapter (8) and bridge (9) [36,37].

Fig. 13. Crack shape development for specimen under bending loading on the bending-torsion test rig.

Fig. 14. Calibration curve for the crack depth measurement by the potential drop technique for the bending-torsion test rig.

determined. Additionally, the crack propagation data have been in lower crack depth regions. For all examined design details, it was
statistically analysed and FORMAN/METTU parameters for dif- observed that the crack shape becomes flatter during the rotating
ferent probabilities of survival were presented. bending.
(2) The developed rotating bending test rig is able to perform rotating (5) The developed mixed-mode test rig is able to perform tests on pure
bending tests for a wide range of round bar specimens with constant bending, pure torsion as well as their superposition in-phase and
and variable amplitude loading. out-of-phase. A verified calibration curve for the experiments on
(3) Furthermore, the influence of different design details of wheelset cyclic bending could be created using the potential drop technique.
axles on the crack propagation behaviour and residual lifetime have The investigations on additional torsion loading were successfully
been investigated. The transition radius design detail with high completed with the application of the digital image correlation.
stress concentration factors leads to a tilting of the crack front at
higher crack depth. In contrast, the superposition of a press fit load
already tilts the crack front at lower stress concentration factors and

268
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

Fig. 15. Observation of the displacement on the surface surrounding the notch using digital image correlation for crack development under bending and torsional
loading.

railway axles: experimental characterization and computation. Eng Fail Anal

2010;17(3):617–23.
[7] 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(8):1906–20.
[8] Beretta S, Carboni M. Experiments and stochastic model for propagation lifetime of
railway axles. Eng Fract Mech 2006;73(17):2627–41.
[9] Carpinteri A, Ronchei C, Vantadori S. Analysis of cracked and notched round bars
under rotary bending. Mater Perform Charact 2015;4(2):20140025.
[10] Purnowidodo A, Makabe C. The crack growth behavior after overloading on ro-
tating bending fatigue. Eng Fail Anal 2009;16(7):2245–54.
[11] Beretta S, Ghidini A, Lombardo F. Fracture mechanics and scale effects in the fa-
tigue of railway axles. Eng Fract Mech 2005;72(2):195–208.
[12] Beretta S, Carboni M, Cantini S, Ghidini A. Application of fatigue crack growth
algorithms to railway axles and comparison of two steel grades. Proc Inst Mech Eng
Part F: J Rail Rapid Transit 2004;218(4):317–26.
[13] Shin C, Chen P. Fatigue crack propagation testing using subsized rotating bending
specimens. Nucl Eng Des 2004;231(1):13–26.
[14] Fonte Md, Reis L, Romeiro F, Li B, Freitas Md. The effect of steady torsion on fatigue
crack growth in shafts. Int J Fatigue 2006;28(5–6):609–17.
[15] Zerres P, Vormwald M. Review of fatigue crack growth under non-proportional
mixed-mode loading. Int J Fatigue 2014;58:75–83.
[16] Döring R, Hoffmeyer J, Seeger T, Vormwald M. Short fatigue crack growth under
nonproportional multiaxial elastic–plastic strains. Int J Fatigue 2006;28(9):972–82.
[17] Highsmith S. Crack path determination for non-proportional mixed-mode fatigue.
PhD thesis, Georgia Institute of Technology, USA; May 2009.
[18] Yang Y. Linear elastic fracture mechanics-based simulation of fatigue crack growth
under non-proportional mixed-mode loading. PhD thesis, TU Darmstadt, Darmstadt;
2014.
[19] Zerres P, Brüning J, Vormwald M. Fatigue crack growth behavior of fine-grained
steel S460N under proportional and non-proportional loading. Eng Fract Mech
2010;77(11):1822–34.
[20] Fonte Md, Reis L, Freitas Md. Fatigue crack growth under rotating bending loading
on aluminium alloy 7075-t6 and the effect of a steady torsion. Theoret Appl Fract
Fig. 16. Crack shape development under bending and torsional loading. Mech 2015;80:57–64.
[21] Freitas Md, Reis L, Fonte Md, Li B. Effect of steady torsion on fatigue crack initiation
and propagation under rotating bending: multiaxial fatigue and mixed-mode
Acknowledgements cracking. Eng Fract Mech 2011;78(5):826–35.
[22] Yu X, Abel A. Crack surface interference under cyclic mode I and steady mode II
loading – part I: experimental study. Eng Fract Mech 2000;66(6):503–18.
The authors gratefully acknowledge the financial support of this [23] Yu X, Abel A. Crack surface interference under cyclic mode I and steady mode II
research project S047/10029/2013 provided by the Karl-Vossloh loading – part II: modelling analysis. Eng Fract Mech 2000;66(6):519–35.
[24] Karolczuk A, Macha E. Critical planes in multiaxial fatigue of materials, vol. Nr. 298
foundation in Essen, Germany.
of Fortschritt-Berichte VDI. Reihe 18, Mechanik, Bruchmechanik, VDI-Verlag,
Düsseldorf; 2005.
References [25] Richard HA, Sander M. Fatigue crack growth: detect – assess – avoid. Solid me-
chanics and its applications, vol. 227. Switzerland: Springer; 2016.
[26] Lebahn J, Sander M. Determination of statistical secured residual lifetime based on
[1] Gänser H-P. Damage tolerance of railway axles: the issue of transferability revisited. sensitivity analysis and stochastic crack propagation simulation. In: Proceedings of
Int J Fatigue 2016;86:52–7. 13th international conference on fracture, Beijing; 2013.
[2] Mädler K, Geburtig T, Ullrich D. An experimental approach to determining the re- [27] Forman RG, Mettu SR. Behavior of surface and corner cracks subjected to tensile
sidual lifetimes of wheelset axles on a full-scale wheel-rail roller test rig. Int J and bending loads in Ti-6Al-4V alloy. Frac Mech: Twenty-Second Symposium,
Fatigue 2016;86:58–63. ASTM STP 1131, vol. 1; 1992. p. 519–46.
[3] Yamamoto M, Makino K, Ishiduka H. Comparison of crack growth behaviour be- [28] NASGRO. Fracture mechanics and fatigue crack growth analysis software. Version
tween full-scale railway axle and scaled specimen. Int J Fatigue 2016;92:159–65. 6.0 - Reference Manual, NASA Johnson Space Center und Southwest Research
[4] Cervello S. Fatigue properties of railway axles: new results of full-scale specimens Institute; 2009.
from euraxles project. Int J Fatigue 2016;86:2–12. [29] ASTM E647-11 standard test method for measurement of fatigue crack growth
[5] Luke M, Varfolomeev I, Lütkepohl K, Esderts A. Fatigue crack growth in railway rates; December 2011.
axles: assessment concept and validation tests. Eng Fract Mech 2011;78(5):714–30. [30] Döker H. Fatigue crack growth threshold: implications, determination and data
[6] Luke M, Varfolomeev I, Lütkepohl K, Esderts A. Fracture mechanics assessment of evaluation. Int J Fatigue 1997;19(93):145–9.

269
R. Hannemann et al. International Journal of Fatigue 118 (2019) 262–270

[31] DIN EN 13103 Bahnanwendungen - Radsätze und Drehgestelle - Laufradsatzwellen - intensity factor solution for scaled wheelset axles under bending. Proc Struct Integr
Konstruktions- und Berechnungsrichtlinie; Oktober 2012. 2016;2:2527–34.
[32] DIN EN 13104 Bahnanwendungen - Radsätze und Drehgestelle - Treibradsatzwellen [36] Hannemann R, Köster P, Sander M. Investigations on crack propagation in wheelset
- Konstruktionsverfahren; März 2013. axles under rotating bending and mixed mode loading. Proc Struct Integr
[33] Carpinteri A, Brighenti R, Spagnoli A. Surface flaws in cylindrical shafts under ro- 2017;5:861–8.
tary bending. Fatigue Fract Eng Mater Struct 1998;21(9):1027–35. [37] Köster P, Sander M. Development of a test rig for round bars under out-of-phase
[34] Jacoby G. Fractographic methods in fatigue research. Exp Mech 1965;5(3):65–82. bending and torsional loading. in: Proc of ICMFF11, Sevilla.
[35] Hannemann R, Sander M. Effect of specimen geometry and press-fit on the stress

270