XML Template (2014) [5.12.

2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Original Article

Proc IMechE Part C:

J Mechanical Engineering Science
Thermographic method for very high 0(0) 1–11
! IMechE 2014
cycle fatigue design in transportation Reprints and permissions:
[Link]/[Link]

engineering DOI: 10.1177/0954406214562463

[Link]

V Crupi, G Epasto, E Guglielmino and G Risitano

Abstract
With the increasing progress of the technological development in the transport industry, the required fatigue life has
increased, so it is very important to determine a safe fatigue strength for 109 cycles. Nowadays, the very high cycle
fatigue constitutes one of the main fatigue design criteria for applications in transport industry. In this paper, the infrared
thermography and an energetic approach were applied to investigate a tool steel in very high cycle fatigue regime. The
traditional energetic approach was developed in order to extend it in very high cycle fatigue regime and to predict the
S-N curves. The failure mechanism of the investigated steel was evaluated by means of scanning electron and optical
microscopies in order to assess if the nature of microstructure and the metallurgical defects, in terms of inclusions and
pores, can influence the crack initiation.

Keywords
Very high cycle fatigue, transport engineering, thermographic analysis, energetic approach, microstructure

Date received: 2 July 2014; accepted: 29 October 2014

possible to carry out tests on components with ultra-

Introduction sonic test systems, because they require specimens
With the increasing progress of the technological with a special geometry. In 1999, Bathias2 experimen-
development in the transport industry, the required tally proved that there is no infinite life in metallic
fatigue life has increased. Nowadays, many compo- materials and in 2007 Sonsino3 showed that a con-
nents and structures in the field of transportation tinuous decrease of fatigue strength occurs in the
engineering (ship, aircraft, automotive, and railway) range of large cycle numbers. The most noticeable
are subjected to fatigue loading with a number of conclusion was that fatigue failure can occur beyond
cycles higher than 107: the fatigue life of a car 107 cycles at stress values lower than conventional
engine ranges around 108 cycles and the big diesel fatigue limit, so it is more appropriate to use the
engines for ships or the high-speed trains work up term fatigue strength at a certain number of cycles
to 109 cycles, the ship structural members are designed than ‘‘fatigue limit’’. Based on the fatigue testing in
to survive at least 20 years, trains and offshore struc- literature, the fatigue life can be subdivided in the
tures have to withstand more than 1010 cycles. following regions: low cycle fatigue (LCF, Nf ¼ 104–
The S-N curves, obtained by the traditional fatigue 105 cycles), high cycle fatigue (HCF, 105 < Nf < 107
tests, are generally represented by a hyperbole with a cycles), and very high cycle fatigue (VHCF, Nf > 107
horizontal asymptote between 106 and 107 cycles. cycles). Assuming that the fatigue life of engineering
They are generally limited to 107 cycles, after which
the standard considers an infinite fatigue life. Department of Electronic Engineering, Chemistry and Industrial
A review of the current studies in the area of the Engineering, University of Messina, Contrada di Dio, Sant’Agata,
very high cycle fatigue (VHCF) is reported by Sakai.1 Messina, Italy
Starting from early 1980s, many researches pub-
Corresponding author:
lished important scientific works based on the results
E Guglielmino, Department of Electronic Engineering, Chemistry and
of ultrasonic fatigue testing, which allows to reach Industrial Engineering, University of Messina, Contrada di Dio 98166,
gigacycles within a suitable testing time (half a day) Sant’Agata, Messina, Italy.
because of its high frequency (20 kHz). It is not Email: eguglie@[Link]

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

2 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

components and structures can range above 108 not always in the interior of the specimen, since the
cycles, it is very important to determine a safe fatigue probability of finding a defect at the surface depends
strength for 109 cycles. This requirement is applicable on the number of defects for a given volume of the
for ship, aircraft, automotive, and railway structures. specimen. Moreover, the cause of the initiation site
Nowadays, the VHCF constitutes one of the main can be not only inclusions, but also the pores. Sohar
fatigue design criteria for applications in transport et al.9 carried out ultrasonic fatigue testing on AISI
industry. D2 type wrought cold work tool steel, finding that
Several experiments show that the effect of fre- crack growth behavior can follow two mechanism,
quency is not critical at 109 cycles. It is found that giving two different crack surface appearance: the
the results are similar in ultrasonic fatigue and in con- so-called fisheye (internal crack initiation) or half-of
ventional fatigue as long as time-dependant mechan- fish-eye (near-surface crack initiation), depending to
isms, corrosion or creep, are not involved.4 In Sakai,1 the presence of primary carbides (clusters).Moreover
the effects of frequency and heat dissipation were ana- the temperature evolution during the fatigue tests in
lyzed and the results show no noticeable frequency HCF10–14 and LCF15 regimes has been investigated by
effect on the fatigue behavior. Furuya5 demonstrated several researchers, but there are only few studies in
that the results, obtained by ultrasonic fatigue testing VHCF regime. Xue et al.16 investigated fatigue
conducted on high-strength steel, were in good agree- damage progression of three kinds of alloy in the
ment with those obtained by conventional servohy- VHCF regime by thermographic analysis. All the spe-
draulic testing on specimens with an almost equal cimens were tested at R ¼ 1 and environment tem-
risk volume, estimated as the region subjected to a perature without cooling air using ultrasonic fatigue
stress level higher than 90% of maximum stress. machine (20 kHz) in order to evaluate the effects of
Bathias et al.6 obtained a large number of S-N temperature increment produced by internal damping
curves for steels and iron loaded in tension–compres- due to cycling at a very high frequency. Blanche
sion or in tension–tension where crack initiation et al.17 proposed a heat diffusion model to estimate
appears up to 109 cycles and sometimes after 109 dissipated energy during VHCF tests at high loading
cycles. They observed that the investigated steels frequency (20 kHz) and low stress (i.e. far below the
have different S-N curves with higher slope in the yield stress). Wang et al.18 measured the temperature
gigacycle regime for the steels with higher UTS, for during high frequency tests by a thin thermocouple
which the difference of the fatigue strength at 106 and attached to the specimen surface at the center section.
109 cycles ranges from 50 to 200 MPa. They identify two zones in the temperature versus
The fatigue crack initiation in the gigacycle regime cycles’ curves during the VHCF test:
seems to occur essentially inside the specimen1 and
not at the surface as it generally occurs in the LCF 1. Stage I, where a steep rise in temperature occurs
and HCF regime. This means that the effect of envir- up to 106 cycles, attaining a maximum tempera-
onment is quite small in the gigacycle regime as the ture which depends on the vibration amplitude;
initiation of short cracks is inside the specimen and 2. Stage II, where the curve saturates and is almost
the surface plays a minor role especially if it is horizontal. The second zone was interpreted as an
smooth. It seems that cyclic plastic strain in the adaptation of the material to interior friction. The
plane stress condition becomes very small in the giga- authors explain that the dissipation of a part of
cycle regime. From a probabilistic point of view, it is the ultrasonic energy is due to interior crack
obvious that the presence of a defect is higher at the nucleation.
interior than on the surface, so a competition can
occur between the internal defects or large grain size Moreover, they observed that once the maximum
and surface damage and defects. The fatigue strength temperature is reached, there is no decrease in the
of high strength steels in VHCF regime is sensitive to temperature range, particularly for a low number of
small defect and inclusions, located inside the speci- cycles to failure where the initiation starts on the sur-
mens. The effect of internal hydrogen trapped by non- face and when the initiation is in the interior of the
metallic inclusions on HCF was first indicated by specimen, for cycles greater than 107, the maximum
Murakami et al.7 They observed that around the temperature could correspond to crack nucleation, as
internal inclusion, from which the failure starts, an is the case in gigacycle fatigue.
optical dark area (ODA) can be seen by an optical In the present paper, the infrared (IR) thermog-
microscope. The size of ODA, which cannot be raphy and an energetic approach were applied to
found on the fracture surface of specimens failed investigate a chromium-vanadium alloyed cold work
at a small number of cycles, increases with the tool steel (DIN EN 115CrV3) in VHCF range.
fatigue life. Moreover, the failure mechanism was evaluated by
The rule of a crack initiation located inside the means of scanning electron and optical microscopy.
specimen in VHCF regime is not rigid, but has several Special attention was given to the presentation of dif-
exceptions. According to Bayraktar et al.,8 in some ferent types of mechanisms operating at the level of
automotive metallic alloys, the crack initiation site is 106 cycles (high fatigue) and 108 cycles (gigacycle

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Crupi et al. 3

fatigue). The aim of the failure analysis was to assess computer control. The test is automatically stopped
if the nature of microstructure and the metallurgical when the frequency falls down to 19.5 kHz.
defects, in terms of inclusions and pores, can influence The tests were carried out on specimens, made of a
the crack initiation. chromium-vanadium alloyed cold work tool steel
(DIN EN 115CrV3). The chemical composition of
the investigated specimens was derived from X-ray
Material and methods
fluorescence analysis and the results are shown in
The fatigue tests were performed without cooling at Table 1.
R ¼ 1 and f ¼ 20 kHz by a piezoelectric fatigue
machine (Figure 1). The vibration of the specimen is
Table 1. Chemical Composition of the tested specimens by
included with a piezo-ceramic transducer which gen- means of XRF analysis.
erates acoustical waves to the specimen through a
power concentrator (horn) in order to obtain more Grade DIN
important displacement and an amplification of the 1.2210 115 CrV 3
stress. The specimen geometry is represented in
Figure 2. The dynamic displacement amplitude of Si Mn S Cr V Fe
the specimen extremity is controlled in order to keep 0.22% 0.34% 0.027% 0.69% 0.09% balance
constant the stress, during the test, through the

Figure 1. Experimental set-up.

Figure 2. Specimen geometry for ultrasonic testing (unit: mm).

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

4 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

During the fatigue tests, the radiometric surface scale. The fatigue life diagram can be divided into
temperature was measured by two IR cameras. The four regions and reveals two knees (black dashed
first one (model FLIR Systems SC 8400) is equipped line). A multistage fatigue life diagram with four dif-
with a e 1280  1024 pixels InSb focal plane array ferent regions was suggested by Mughrabi,19 who
cooled detector working in the MWIR (1.5–5.1 mm) distinguished two classes of materials: ductile single-
spectral band (NETD < 25 mK at 30  C). A 100 mm phase metallic materials (type I) without internal
focal length lens (FOV 11  9 ) was used. The defects and type II materials, e.g. high-strength
thermographic images were acquired at 200 fps by steels containing internal defects (nonmetallic inclu-
FLIR ResearchIR software, in sub-windows reso- sions or pores), which acts as (internal) crack initi-
lution, at 416  484 pixels. Figure 1 shows the experi- ation sites.
mental set-up. The second one (model FLIR Systems An interesting theoretical approach to describe the
A40M) is an uncooled long wave infrared (LWIR) S-N curve also in gigacycle regime was proposed by
focal plane array camera with a resolution of Naimark et al.20 This model is based on the kinetics of
320  240 pixels, a measurement accuracy of 2 C defect accumulation in metals under fatigue loading
and a frame rate during the acquisition of thermal and is able to explain the duality of the S-N curve, due
increment of 1 frame per minute. to different failure mechanisms: the surface nucleation
Fractographies were carried out by means of a of the crack and the crack nucleation in the bulk of
scanning electron microscope (SEM – JEOL the specimen.
JSM5900 LV) and an optical stereomicroscope (SM According to the multistage fatigue life dia-
– Leica M165C). Moreover, a 3D computed tomog- gram,19–21 Figure 3 shows four regions:
raphy (CT) system, based on a variable focal-spot size
technology, was used to analyze the specimens. This . Region I (HCF region) where fatigue life are finite;
unit is equipped with an X-ray source having max- . Region II, which corresponds to the conventional
imum voltage and current of 225 kV and 7.1 mA, fatigue limit in HCF regime, Region I and Region
respectively, depending on the focal-spot size that II represent the conventional S-N curve;
can be chosen among these values: 250 mm, 300 mm, . Region III (VHCF region), where fatigue life are
500 mm, and 800 mm. The detector system is a flat long but finite;
panel with a resolution of 1920  1536 pixels. . Region IV with infinite fatigue life. So far, the
existence of region IV has been disputed:21 accord-
ing to Nishijima and Kanazawa22 a real fatigue
Results and discussion limit for crack initiation at inclusions exists,
according to Sakai1 it does not.
Energetic approach
The experimental results, obtained by the fatigue During HCF test of common engineering metals
tests, are shown in Figure 3 in a semi-logarithmic and welded joints,10–14 the temperature evolution of

Figure 3. Fatigue life diagram.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Crupi et al. 5

the specimen undergoes three separate stages Crupi25 developed a theoretical model able to
(Figure 4): an initial rise (Stage I), slow and steady describe the temperature evolution during the Stage
increase (Stage II), and a drastic increase prior to fail- I and Stage II of the fatigue life
ure (Stage III). The same trend was observed in LCF  N

regime by Crupi et al.15 This temperature evolution is
Td ¼
Tas  1  e  ð1Þ
closely related to the internal microstructural changes,
as demonstrated in Crupi,23 and to crack initiation where  is a constant; if N is  then
Td is 0.63
Tas
and propagation.24 and if N is 4 then
Td is 0.98
Tas. The value of 4,
In the present study, the temperature evolution in applied successfully to conventional steel under HCF
VHCF regime was analyzed using an IR thermog- loading in Crupi25 was considered also in the investi-
raphy. The highest dissipation is close to the center gated VHCF tests. The
Td  N data were interpo-
section of the specimen, where the stress intensity is lated by means of an exponential function according
also highest. However, this result was not systematic. to equation (1) and the convergence was achieved, as
Sometimes a high dissipation zone was observed fur- demonstrated in Figure 5.
ther away from the specimen center. Energetic approaches, based on IR analysis, has
The
Td – N, obtained applying the IR thermog- been applied by several researchers26–29 in HCF
raphy in VHCF regime, shows similar trend with the regime to obtain the S-N curve and to assess the resi-
three stages (Figure 5). dual fatigue life.

Figure 4. Theoretical
Td – N curve.

Figure 5.
Td – N curve.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

6 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

The basic assumption of the so-called ‘‘energy fatigue life (Figure 3), demonstrated that  is no
approach’’26 is that the fatigue failure takes place longer constant for fatigue life higher than HCF
when the absorbed energy reaches a certain threshold zone (about 2  106, as shown in Figure 3), but
value EC characteristic for each structural detail. The increases by an order of magnitude with the increment
limit energy EC is proportional to the integral  of the of the number of cycles. According to equation (3),

Td–N curve there is an high increment of  in the VHCF regime,
where the number of cycles to failure Nf is higher than
Z Nf 107. The traditional energetic approach is based on
EC /  ¼
Td ðNÞdN ð2Þ the assumption that the integral  is correlated to
0
the slope of the S-N curve, but, if the investigated
where Nf is the number of cycles to failure. When a fatigue life ranges from 105 to 1010 cycles, it could
single constant-amplitude stress range level
 is be obtained a multiscale fatigue life diagram with dif-
applied, since the number of cycles of the stages 1 ferent regions and slopes of the S-N curves (Figure 3),
and 3 is negligible compared with that of stage 2, which correspond to different values of the integral .
the integral  can be practically assumed to be The increment of the integral  has also a physical
reason. The failure mechanism in VHCF regime, as
 
TAS Nf ð3Þ reported in literature, is different respect the one in
HCF regime, so it is required an higher amount of
It has been demonstrated that the energy param- energy to produce the failure. Thus, three distinct
eter  is a material constant in HCF regime because it average values of  were evaluated according to equa-
is the same when load histories at different levels are tion (3):
applied. This ascertainment, even if the involved par-
ameters are not the same, substantially confirms the . 1 between 105 and 106 cycles (Region I of
well-known Miner’s rule. Moreover the traditional Figure 3);
approach of the thermographic method10 is based . 12 between 106 and 107 cycles (Region II of
on the assumption that there is no temperature incre- Figure 3);
ment for stress range lower than the fatigue limit. . 2 over 107 cycles (Region III of Figure 3).
Really, there is a very low asymptote surface tempera-
ture
Tas (red line in Figure 4), so the integral  has a In Figure 6, the red points corresponds to the pre-
finite value according to equation (3). dicted values of fatigue life, obtained according to
The traditional energetic approach was developed equation (2) by the ratio between the average values
by the authors in order to extend it in VHCF regime of  and the corresponding average values of mea-
and to predict the S-N curves. The assessment of the sured asymptotic surface temperature
Tas. Two dis-
integral  for the tests, carried out in a wide range of tinct slopes of the S-N curve have been identified

Figure 6. Experimental and predicted S-N curves.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Crupi et al. 7

interpolating in the fatigue life diagram, shown in Other authors studied the influence that spheroidite
Figure 6, the predicted values relative to 1 and and metallurgical instability due to the raising tem-
12 and relative to 12 and 2, respectively. The perature in ultrasonic fatigue test31,32 have on the
S-N curves predicted by the energetic approach, S-N curve.
shown in Figure 6 (red dash lines) are quite close to Fracture surfaces of the investigated steel were
the experimental S-N curves. observed after testing by SEM and SM. The stages
of crack initiation, stable crack propagation, unstable
crack propagation, and final failure are well defined.
Fracture analysis
After the fatigue tests, not all specimens were broken
in two parts, even though the tests were automatically
interrupted by the control unit of the fatigue test
equipment for an excessive deformation of the
sample. The authors used to apply a higher stress
level for a very small number of cycles with the aim
to separate the failed specimens in order to analyze
the fracture surfaces. The procedure consists of
increasing the applied stress of 40 MPa at R ¼ 1 in
order to separate the failed specimens after about 100
cycles.
This procedure does not alter the broken surface,
as showed in the CT analyses of the specimen imme-
diately after the fatigue test that seems to be undam-
aged (Figure 7). CT slice of the failed specimen shows
the more evident mechanical damage of the analyzed
section (smaller damage or cracks are not visible at
this resolution). The same region is shown with the
stereomicroscope (SM) after the split-up, in which in
green is marked the same region (Figure 8).
For the analyzed steel, thermal treatment consists
of annealing by heating between 750 C and 800 C Figure 8. Fractography for the same specimen of Figure 7;
and holding for several hours followed by slow cool- the specimen was separated by applying a high stress level.
ing producing a spheroidized structure (Figure 9).
Some hardness tests were carried out on some spe-
cimens after the fatigue tests (Table 2) and these
values can be referred to an annealing process. As
spheroidite is a very soft phase,30 the measured hard-
ness could be derived only from a microstructure
transformation occurred during the fatigue test.

Figure 9. Micrography of spheroidize annealed 115CrV3.

Nital etch, 1000X.

Table 2. Average Rockwell C hardness

for some tested specimens.

Specimen 1 2 3
Figure 7. CT analysis of the specimen after the fatigue test HRC 25.1 22.8 22.2
(
 ¼ 518 MPa, Nf ¼ 2.65107, R ¼ 1).

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

8 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Figure 10. Fracture surface with subsurface crack initiation site for a specimen,
 ¼ 528 MPa, Nf ¼ 2.63  108, R ¼ 1.
(a) Overview, (b) detail of crack initiation site with hole, (c) pores, (d) ductile fatigue striations, (e) nonmetallic particle, (f) interface
between unstable crack propagation and final failure, (g) brittle failure surface, and (h) ductile failure surface with microcracks.

Figure 11. Fracture surface with subsurface crack initiation site for a specimen (
 ¼ 588 MPa, Nf ¼ 4.11  105, R ¼ 1).
(a) Overview, (b) detail of cracks on the surface, (c) non-metallic inclusion and its EDS spectrum, (d) large crack, and (e) interface
between unstable crack propagation and final failure.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Crupi et al. 9

Inclusions are not the key parameters in this case. If

we can consider that a new phase may play a similar
role as an inclusion, its presence or discontinuity in
microstructure35 at the initiation site is crucial for
determining fatigue life.
The new transformed microstructure of the ana-
lyzed steel can be bainite. The possible presence of
bainite is confirmed as by the calculated hardness as
by the thermal history recorded during the test.
Bainite that forms at temperatures below approxi-
mately 350 C or 400 C is called lower bainite,36
whose presence can be the explanation of no ODA
on the fracture surface.37,38
Fractography of a specimen (
 ¼ 588 MPa,
Nf ¼ 4.11  105, R ¼ 1) shows an intergranular fail-
ure with opened cracks (Figure 11(b), (d), and (e)) due
to high strain. Onto the fracture surface, nonmetallic
inclusions can be found and they are not the site in
which initiation of cracks occurred. EDS analyses of
Figure 11(c) show that nonmetallic particles are sili-
cates and that in the intergranular fracture region
there is no trace of them (Figure 11(b)). An extensive
surface defect should be the site in which crack
initiated. The same behavior was observed in the frac-
tography of the specimen (
 ¼ 610 MPa,
Nf ¼ 1.79  105, R ¼ 1, Figure 12).

Conclusions
Nowadays, the VHCF constitutes one of the main
fatigue design criteria for applications in transport
industry. In this paper, the IR thermography and an
energetic approach were applied to investigate a tool
steel in VHCF regime. The traditional energetic
approach was developed in order to extend it in
VHCF regime and to predict the S-N curves. The fail-
Figure 12. Optical fractography for a specimen ure mechanism of the investigated steel was evaluated
(
 ¼ 610 MPa, Nf ¼ 1.79  105, R ¼ 1) after the fatigue test; by means of scanning electron and optical micros-
the cracks initiated onto the external surface. copy. The failure analysis allowed the authors to
assess that the microstructure of the analyzed steel
influences the crack initiation and its behavior in the
VHCF regime. Cross reading of the results, coming
As shown in Figure 10, in the VHCF regime of the from hardness test, SEM, and ST analyses, confirms
initiation sites the crack started sub-superficially at an that at the test frequency (20 kHz), the heating of the
average depth of approximately 100 mm from the specimen, induced by fatigue, can initiate the micro-
external face. The crack initiation was noninclusion structural phases transformations.
induced. Figure 10(a) shows the fracture surface
morphology for a specimen (
 ¼ 528 MPa,
Nf ¼ 2.63  108, R ¼ 1). Where the fracture surface Conflict of interest
became rough it means that the crack propagated rap- None declared.
idly and unstably, as shown in Figure 10(g) and (f),
giving a surface that consists of small facets, seem- Funding
ingly caused by cleavage. As shown in Figure 10(b), This research received no specific grant from any funding
the crack initiated from a brighter site near the sur- agency in the public, commercial, or not-for-profit sectors.
face, which should be a new fatigue-induced
phase.33,34 Thus, in this case, the failure initiation References
mechanism strongly depends on the microstructure 1. Sakai T. Review and prospects for current studies on
and, as reported in Bayraktar et al.,8 Paris’s model very high cycle fatigue of metallic materials for machine
or Murakami’s formulas cannot be applied. structural use. J Solid Mech Mater Eng 2009; 3: 425–439.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

10 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

2. Bathias C. There is no infinite fatigue life in metallic 20. Naimark OB, Plekhov OA, Betekhtin VI, et al. Kinetics
materials. Fatigue Fract Eng Mater Struct 1999; 22: of defect accumulation and duality of the Weller curve
559–565. in gigacycle fatigue of metals. Tech Phys 2014; 59:
3. Sonsino CM. Course of SN-curves especially in the 398–401.
high-cycle fatigue regime with regard to component 21. Pyttel B, Schwerdt D and Berger C. Very high cycle fati-
design and safety. Int J Fatigue 2007; 29: 2246–2258. gue – Is there a fatigue limit? Int J Fatigue 2011; 33: 49–58.
4. Stanzl-Tschegg S. Fracture mechanisms and fracture 22. Nishijima S and Kanazawa K. Stepwise S-N curve and
mechanics at ultrasonic frequencies. Fatigue Fract Eng fish-eye failure in gigacycle fatigue. Fatigue Fract Eng
Mater Struct 1999; 22: 567–579. Mater Struct 1999; 22: 601–607.
5. Furuya Y. Specimen size effects on gigacycle fatigue 23. Fan J, Guo X, Wu C, et al. Using infrared thermog-
properties of high-strength steel under ultrasonic fati- raphy in effect evaluation of heat treatments on mar-
gue testing. Scripta Mater 2008; 58: 1014–1017. tensitic steel. Exp Techniques 2014. DOI: 10.1111/
6. Bathias C, Drouillac L and Le Francois P. How and ext.12019.
why the fatigue S-N curve does not approach a hori- 24. Plekhov OA, Palin-Luc T, Saintier N, et al. Fatigue
zontal asymptote. Int J Fatigue 2001; 23: S143–S151. crack initiation and growth in a 35CrMo4 steel investi-
7. Murakami Y, Nomoto T and Ueda T. Factors influen- gated by infrared thermography. Fatigue Fract Eng
cing the mechanism of superlong fatigue failure in Mater Struct 2005; 28: 169–178.
steels. Fatigue Fract Eng Mater Struct 1999; 22: 25. Crupi V. An unifying approach to assess the structural
581–590. strength. Int J Fatigue 2008; 30: 1150–1159.
8. Bayraktar E, Garcias IM and Bathias C. Failure mech- 26. Fargione G, Geraci A, La Rosa G, et al. Rapid deter-
anisms of automotive metallic alloys in very high cycle mination of the fatigue curve by the thermographic
fatigue range. Int J Fatigue 2006; 28: 1590–1602. method. Int J Fatigue 2002; 24: 11–19.
9. Sohar CR, Betzar-Kotas A, Gierl C, et al. 27. Amiri M and Khonsari MM. Life prediction of metals
Fractographic evaluation of gigacycle fatigue crack undergoing fatigue load based on temperature evolu-
nucleation and propagation of a high Cr alloyed cold tion. Mat Sci Eng A-Struct 2010; 527: 1555–1559.
work tool steel. Int J Fatigue 2008; 30: 2191–2199. 28. Williams P, Liakat M, Khonsari MM, et al. A thermo-
10. La Rosa G and Risitano A. Thermographic method- graphic method for remaining fatigue life prediction of
ology for rapid determination of the fatigue limit of welded joints. Mater Des 2013; 51: 916–923.
materials and mechanical components. Int J Fatigue 29. Risitano A and Risitano G. Cumulative damage evalu-
2000; 22: 65–73. ation in multiple cycle fatigue tests taking into account
11. Amiri M and Khonsari MM. Rapid determination of energy parameters. Int J Fatigue 2013; 48: 214–222.
fatigue failure based on temperature evolution: Fully 30. ASM Handbook. Metals handbook: Heat treatment. 9th
reversed bending load. Int J Fatigue 2010; 32: 382–389. ed. Materials Park, OH: ASM International, 1981.
12. Meneghetti G, Ricotta M and Atzori B. A synthesis of 31. Bathias C. Coupling effect of plasticity, thermal dissi-
the push-pull fatigue behaviour of plain and notched pation and metallurgical stability in ultrasonic fatigue.
stainless steel specimens by using the specific heat Int J Fatigue 2014; 60: 18–22.
loss. Fatigue Fract Eng Mater Struct 2013; 36: 32. Berger C and Christ H. Fifth International Conference
1306–1322. on Very High Cycle Fatigue – Proceedings, 28–30 June
13. Curà F, Curti G and Sesana R. A new iteration method 2011, Berlin, Germany.
for the thermographic determination of fatigue limit in 33. Liu YB, Li SX, Yang ZG, et al. Thermal effect of bai-
steels. Int J Fatigue 2005; 27: 453–459. nite/martensite duplex-phase steel under ultrasonic fati-
14. Fan JL, Guo XL, Wu CW, et al. Stress assessment and gue testing. J Mater Sci 2010; 45: 2553–2557.
fatigue behavior evaluation of components with defects 34. Müller-Bollenhagen C, Zimmermann M and Christ HJ.
based on the finite element method and lock-in therm- Very high cycle fatigue behaviour of austenitic stainless
ography. Proc IMechE, Part C: J Mechanical steel and the effect of strain-induced martensite. Int J
Engineering Science. DOI: 10.1177/0954406214541432. Fatigue 2010; 32: 936–942.
15. Crupi V, Chiofalo G and Guglielmino E. Infrared 35. Zhu ML, Xuan FZ and Chen J. Influence of micro-
investigations for the analysis of low cycle fatigue pro- structure and microdefects on long-term fatigue behav-
cesses in carbon steels. Proc IMechE, Part C: J ior of a Cr-Mo-V steel. Mat Sci Eng A 2012; 546: 90–96.
Mechanical Engineering Science 2011; 225(4): 833–842. 36. ASM Handbook. Metallography and microstructures.
16. Xue H, Wagner D, Ranc N, et al. Thermographic ana- Materials Park, OH: ASM International, 2004.
lysis in ultrasonic fatigue tests. Fatigue Fract Eng Mater 37. Murakami Y, Nomoto T, Ueda T, et al. On the mech-
Struct 2006; 29: 573–580. anism of fatigue failure in the superlong life regime
17. Blanche A, Chrysochoos A, Ranc N, et al. Dissipation (N > 107 cycles). Part I: Influence of hydrogen trapped
assessments during dynamic very high cycle fatigue by inclusions. Fatigue Fract Eng Mater Struct 2000; 23:
tests. Exp Mech 2014. DOI: 10.1007/s11340-014-9857-3. 893–902.
18. Wang QY, Berard JY, Dubarre A, et al. Gigacycle fati- 38. Murakami Y, Toriyama T, Tsubota K, et al. What hap-
gue of ferrous alloys. Fatigue Fract Eng Mater Struct pens to the fatigue limit of bearing steel without non-
1999; 22: 667–672. metallic inclusions?: Fatigue strength of electron beam
19. Mughrabi H. On ‘multi-stage’ fatigue life diagrams and remelted super clean bearing steel. In: JJC Hoo and
the relevant life-controlling mechanisms in ultrahigh- WB Green (eds) Bearing steels: Into the 21st century.
cycle fatigue. Fatigue Fract Eng Mater Struct 2002; Philadelphia, PA: American Society for Testing and
25: 755–764. Materials, 1998, pp.87–105. ASTM STP 1327.

Downloaded from [Link] at University of Huddersfield on December 29, 2014

XML Template (2014) [5.12.2014–3:51pm] [1–11]
//[Link]/cenpro/ApplicationFiles/Journals/SAGE/3B2/PICJ/Vol00000/140278/APPFile/SG-PICJ140278.3d (PIC) [PREPRINTER stage]

Crupi et al. 11

 stress range (Mpa)

Appendix
 0 fatigue limit (MPa)

Tas asymptotic temperature increment
Notation during fatigue test (K)
EC energy to failure per unit volume (J/m3)
Td surface temperature increment during
f frequency (Hz) fatigue test (K)
N number of cycles
 thermal increment to failure per unit
Nas asymptotic number of cycles
volume (K/m3)
Nf number of cycles to failure
R stress ratio

Downloaded from [Link] at University of Huddersfield on December 29, 2014