Experimental Mechanics (2018) 58:259–267
DOI 10.1007/s11340-017-0340-9
Rapid Determination of Fretting Fatigue Limit
by Infrared Thermography
S. Chhith 1,2 & W. De Waele 1 & P. De Baets 1,2
Received: 28 February 2017 / Accepted: 19 September 2017 / Published online: 4 October 2017
# Society for Experimental Mechanics 2017
Abstract This paper demonstrates the feasibility of infrared
thermography to determine the so-called fretting fatigue limit.
Fretting fatigue tests are performed on aluminum and steel
specimens. The coupled fatigue and tangential loads are se-
quentially increased (block loading) whilst the normal load is
kept constant for all blocks. The temperature data is processed
and analyzed using a Fast Fourier Transform (FFT) algorithm
implemented in the commercial software Matlab. It is demon-
strated that the second harmonic of the temperature signal can
be linked to the specific loading block below which no or
negligible damage is generated in the specimen. The stress
amplitude of this block is considered to be a best estimate of
the fretting fatigue limit. A constant amplitude fretting fatigue
test with this stress amplitude confirmed that the specimen
remains intact at 107 cycles.
Keywords Fatigue limit . Fretting fatigue . Thermography .
Fast Fourier Transform . Second harmonic
Introduction
Dynamic forces occurring in most mechanical systems cause
vibrations and as a result very small reciprocating movements
between the interacting surfaces. This microslip, typically 5 to
* S. Chhith
saosometh.chhith@ugent.be
1
Department of Electrical Energy, Metals, Mechanical Construction
and Systems, Soete Laboratory, Ghent University, Technologiepark
903, 9052 Ghent, Belgium
2
Member of Flanders Make, The Strategic Research Center for the
Manufacturing Industry, Ghent, Belgium
50 μm, will lead to local material removal called fretting, and
cracks might initiate at the border of the stick-slip area. These
cracks can propagate due to fatigue and thus the lifetime of the
component will be degraded. Examples of installations and
machine components that are cyclically loaded and prone to
this type of damage are heat exchangers, bolted or riveted
connections, dovetail connections, … [1–5].
Fretting Fatigue Limit
The fatigue regime is usually classified in three domains: low
cycle fatigue (lifetime: Nf < 104 cycles), high cycle fatigue
(104 cycles < Nf < 107 cycles), and ultra high cycle fatigue
(Nf > 107 cycles) [6, 7]. Low cycle fatigue is characterized by
macroscopic plastic deformation, whilst only localized plastic
deformation occurs in high cycle fatigue [7]. The localized
plastic deformation in high cycle fatigue can be due to cyclic
slip activity at the free surface of a certain grain, to a sharp
cross-sectional change, a tip of a notch, a fretting contact, etc.
In this sense, fretting fatigue is typically considered to be high
cycle fatigue which should show a fatigue limit at 107 cycles
[8] even though failures due to fretting fatigue have been
observed in ultra high cycle regimes [9].
The fatigue limit is defined as the stress amplitude thresh-
old below which a crack has not nucleated or has nucleated
but also arrested. Crack arrest can be divided into two types:
microstructural crack arrest where micro-crack growth is
stopped due to grain boundaries, and mechanical crack arrest
where the stress intensity factor of a micro-crack is below the
stress intensity factor threshold [7]. The fretting fatigue limit
has been studied based on the phenomenon of mechanical
crack arrest. It is considered to be that loading condition for
which the stress intensity factor is lower than the threshold
value for a non-propagating micro-crack, which can be ana-
lytically calculated from the Kitagawa-Takahashi diagram
260
[10, 11]. This threshold is experimentally determined by
running plain fatigue tests on pre-cracked specimens
(micro-crack) [12].
Fatigue Analysis Using Thermography
It is well-known that there is a correlation between the damage
accumulated during the fatigue process and the heat dissipa-
tion [13–16]. Based on the measurement of the temperature
rise of the specimen due to the dissipated energy, the fatigue
limit or even a complete S-N curve of metals can be deter-
mined [17–21]. This thermal methodology was also success-
fully applied to composite materials [22, 23].
Instead of only measuring the temperature rise of the spec-
imen, lock-in thermography allows to separate linear and non-
linear temperature responses of materials, where the former is
thermoelastic response and the latter is linked to the dissipated
energy. An excellent review of four lock-in thermography
algorithms can be found in [24]. The temperature responses
of the first and higher harmonics, or the linear and non-linear
responses, are used to rapidly determine the heat dissipation
and fatigue limit of metals [25]. For example, the so-called E-
and D-mode algorithms developed by CEDIP are capable to
measure directly the linear (thermoelastic) and non-linear tem-
perature response of materials [26, 27]. In another example,
crack initiation in a weld was detected by separating temper-
ature responses based on signal post-processing over each
pixel of the thermal images, and allowed the detection of crack
initiation of a weld [28]. The linear temperature response ob-
tained by thermoelastic stress analysis of the loading signal,
while the non-linear temperature response was calculated as
the difference of the total experimental temperature and the
linear or thermoelastic response.
Fretting Fatigue Analysis Using Thermography
Damage accumulation during fretting fatigue is similar to
plain fatigue except that surface damage due to friction is
combined with the plastic deformation of materials. Infrared
thermography has been shown to be an excellent non-
destructive method for evaluation of fretting damage in closed
surfaces of riveted lap joints [29]. It has furthermore been
demonstrated that analysis of the first and second order har-
monics using least-squares lock-in processing allows to sepa-
rate the plastic deformation from the friction-induced surface
damage in plain fretting conditions (without fatigue load)
[30]. The separation is achieved comparing the temperature
responses at a contact zone of the sample in an experiment,
which take into account both friction and plasticity, and those
of a thermal model which includes only the frictional effect.
This allows to predict which load conditions under plain fret-
ting lead to a crack in the high cycle regime (defined as
106 cycles).
Exp Mech (2018) 58:259–267
The first harmonic of the thermal response was effectively
used in previous work of the authors [31] to determine on-line
crack initiation lifetime during fretting fatigue tests with con-
stant stress amplitude for both an aluminum alloy and a struc-
tural steel. Three stages were observed for the evolution of the
first harmonic response: running-in, stabilization, and micro-
crack propagation. An increase of the first harmonic response,
larger than a filtered signal-to-noise ratio of 2 mK from the
stabilized stage was defined as initiation criterion. Several
stress levels were applied and the crack length corresponding
to the initiation criterion was determined; these crack lengths
varied between 130 to 220 μm. The current study exploits the
feasibility of the second harmonic to predict the fretting fa-
tigue limit of the two metals. The fatigue stress level, and
consequently also the tangential force imposed by the fretting
pad, is increased stepwise in blocks. With thermal images
recorded, temperature responses are analysed in the frequency
domain via the Fast Fourier Transform (FFT) implemented in
the commercial software Matlab. Different physical contribu-
tions are linked to the harmonics at specific frequencies. In
particular, fretting fatigue damage is related to the harmonic at
twice the loading frequency. Monitoring the second harmonic
allows to identify the stress level below which no or negligible
damage is found. This stress level is assumed to be the fretting
fatigue limit.
Experiments
Experimental Set-up and Materials
A fretting fixture has been designed for installation on an ESH
100 kN servo hydraulic load frame, in order to perform fret-
ting fatigue tests [32]. A schematic drawing of this fixture is
shown in Fig. 1. The 100 kN hydraulic cylinder of the load
frame is used to apply dynamic axial stress, σ, on a dog-bone
Fig. 1 Schematic drawing of the fretting fatigue test rig [33]
Exp Mech (2018) 58:259–267
Fig. 2 Photographs of the
experimental set-up
specimen. The normal clamping load, P, is applied to the
specimen by a separate horizontal hydraulic cylinder which
loads the fretting pads mounted on flexural beams. The C-
beam mounted on ertalon blocks can float freely on the base
plate and ensures that the normal load is applied equally at
both fretting pads. The tangential force, Q, between the dog-
bone specimen and the pads is generated by the compliant
springs as a result of the elastic deformation of the dog-bone
specimen. The interaction of these forces on the specimen is
shown on the photographs in Fig. 2 and the schematic diagram
in Fig. 3. The load cell mounted on the C-beam is used to
directly measure the normal load. The tangential force gener-
ated by the compliant springs can be measured in two ways.
First, as the difference between the force measured by the load
cell of the ESH 100 kN and the force below the contact areas
measured by a calibrated strain gauge mounted on the speci-
men surface (see Fig. 2). Secondly, it can be calculated from
the deflection of the springs based on a calibrated strain gauge
attached on one of the springs. The latter methodology has
been used since it does not require the use of a strain gauge
Fig. 3 Schematic drawing showing interacting forces and diagram of
axial force under fretting fatigue
261
on each tested specimen; evidently it requires a calibration of
the spring.
For each test the specimen and pads are made of identical
material; two different metals are used in this study. The first is
an aluminum alloy (Al2024-T3) and the second one is a cold
drawn structural steel (S235JRC). Their strength properties
are summarized in Table 1 and their dimensions are shown
in Fig. 4. The dog-bone specimen’s contact surface is plane,
while the pad’s contact surface is cylindrical. The cylindrical
radius, Rpad, of the aluminum and the steel pads are 50 mm
and 150 mm, respectively.
Sliding Regimes under Fretting
The sliding regime under fretting is divided into two types,
partial slip and gross slip, depending on the sliding amplitude
or the magnitude of the tangential force. Partial slip is consid-
ered to be a mix between sticking and slipping within the
contact area. In gross slip the entire contact areas slip over
each other; this occurs if the sliding amplitude or the magni-
tude of the tangential force is increased. The partial slip regime
is selected for fretting fatigue tests since in that case cracking
is the dominant damage mechanism while for gross slip re-
gime wear is more dominant [34]. To determine the transition
between the two regimes, the approach used in [33] has been
adopted in a previous study [31]. The expected partial slip
condition (σ = 180 MPa, P = 1 kN) for Al2024-T3 is cycli-
cally applied for 103 cycles. Then a quasi-static load of
325 MPa is applied until the pads slip over the specimen, i.e.
the deflection of the compliant springs drop even though the
axial load is increased. The maximum value of the tangential
force is defined as the border between the two regimes. The
Table 1 Strength properties of Al2024-T3 and S235JRC
Materials Young’s Modulus E [GPa] σy [MPa] σult [MPa]
Al2024-T3 73 383 506
S235JRC 210 650 700
262
Fig. 4 Dimensions of the
specimen and the pad
approach is also applied to the structural steel, and Fig. 5
illustrates the two sliding regimes under fretting [31].
Experimental Procedures
Fretting fatigue tests under ambient conditions with cylindri-
cal pads against a plane specimen are performed and moni-
tored by an infrared camera. The specimen and the pads are
painted in black to improve their thermal emissivity. The in-
frared camera model is InfraTec series 8300 with a telephoto
lens of 50 mm focal distance. It is based on a focal plane array
with InSb detectors cooled by a Stirling system. Its spatial
resolution is 640 by 512 pixels and its temporal sensitivity at
25 °C is about 25 mK. The camera is used with a default
standard calibration with a temperature range of 5 to 60 °C
and an integration time of 1050 μs. The camera is installed at a
distance of around 200 mm from and perpendicular to the
specimen surface. This corresponds to a pixel size of the ther-
mal images equal to 65 μm, the lowest pixel size achievable.
The tests are run with fatigue ratio (σmin/σmax) equal to 0, a
tangential load ratio (Qmin/Qmax) equal to −1, and a loading
frequency of 10 Hz. The maximum fatigue load (σmax) is
increased step by step as shown schematically in Fig. 6. This
enables a step-by-step increase of the maximum tangential
load (Qmax) too, since the test rig is a single actuator with a
spring system as shown in Fig. 1. The normal load, P, is kept
Fig. 5 Sliding regimes under fretting for Al2024-T3 and S235JRC [31]
Exp Mech (2018) 58:259–267
constant for all the load steps. Test conditions of each load
block are specified in Tables 2 and 3.
Each load step is run for 5000 cycles to ensure that a sta-
bilized temperature is reached; the total duration of a test is
only 40,000 cycles. All data of the interacting forces are ac-
quired with a sampling rate of 102.4 Hz and logged into the
ESH 100 kN control computer whereas thermal images are
recorded with a frame rate of 100 Hz on a separate laptop.
Since the loading frequency is 10 Hz, 10 data points are re-
corded for each load cycle, which can be considered to be
oversampling and offers good accuracy of data processing.
Analysis of Temperature Response in Fretting
Fatigue Tests
Heat Sources in Fretting Fatigue Tests
Heat generation in fretting fatigue experiments is due to the
thermoelastic effect and the inelastic effect which accounts for
fretting friction, plastic dissipation, and disturbing sources
such as environmental temperature variations and heat con-
duction originating from the testing machine. Without any
inelastic effect, the total temperature response will be deter-
mined by the thermoelastic effect which follows the wave
form of the mechanical load (a sinusoid in this study). The
sinusoidal function of the total temperature response will be
distorted when an inelastic effect occurs, which results in a
non-perfect sinusoidal waveform. Two conditions should be
considered:
& If the total temperature response is periodic, a Fourier
series decomposition allows any periodic function to be
written as a sum of simple oscillating functions having a
frequency equal to the multiples of the fundamental fre-
quency. Hence, the second harmonic and higher har-
monics are the results of the distortion of the thermoelastic
response due to the inelastic effect.
& When the temperature response is not periodic, one can
still decompose this function as a periodic function and a
residue. Then, the periodic function is analyzed as in the
first case. Therefore, the second harmonic and higher
Exp Mech (2018) 58:259–267 263

Fig. 6 Schematic load blocks

during fretting fatigue limit tests

harmonics of the periodic function again account for most the thermoelastic constant, and σkk is the sum of the ampli-
part of the inelastic effect. Of course, the residue function tudes of the principal stresses within one cycle.
can be also used to study the inelastic [28] but the second
αT 0
and higher harmonics allow a more simple analysis using T1 ¼ − σkk ¼ −kσkk ð2Þ
the FFT. ρC p

The second harmonic effect is more difficult to be calculat-

The temperature variation at the specimen surface can be ed but can be linked to fretting fatigue damage (friction in-
best fitted by equation (1) [25, 30, 31]. duced surface damage and plasticity) in the specimens.
T exp ðt Þ ¼ T 0 þ ΔT * f * t þ T 1 sinðωt þ φ1 Þ þ T 2 sinð2ωt þ φ2 Þ
ð1Þ Temperature Decomposition and Damage Detection

Here Texp is the overall experimental temperature,
T1sin(ωt + φ1) is the first harmonic effect (pure thermoelastic
effect), ΔT*f and T2sin(2ωt + φ2) are linear drift and second
harmonic effect respectively, which account for friction, plas-
ticity and other disturbing factors [30, 31]. ΔT is the mean
temperature change which can be obtained from a fitting pro-
cedure [25, 30], and f is the loading frequency. Higher har-
monic effects are neglected. Parameter ω is the angular fre-
quency of the fatigue loading, ϕ1 and ϕ2 are phase shifts of the
controlled sinusoidal force, and t is time.
If the system is adiabatic the thermoelastic effect can easily
be quantified by a very well-known relation, equation (2),
developed by Lord Kelvin [35], where α is coefficient of
thermal expansion, T0 is the mean temperature in a cycle, ρ
is the mass density, Cp is the specific heat capacity at constant
pressure, T1 is the thermoelastic temperature amplitude, k is
Four regions of interests, ROIs, (see Fig. 7) are selected from
the recorded thermal images. The dimensions of each ROI are
selected as 10 by 20 pixels which is approximately equal to a
by 2a, where a is the Hertzian contact semi-width [30, 31].
Such small region of interest allows a better sensitivity for the
localized stress concentration occurring in fretting fatigue con-
ditions [35]. ROI’s R1 and R2, see Fig. 7, are the most impor-
tant since they are located near the contact zones where the
fretting fatigue damage or crack initiation will most likely
occur [31]. ROI’s R3 and R4 are chosen above and below the
contact regions respectively for reference purposes since they
are only subjected to uniaxial fatigue load without fretting
damage. The thermoelastic effect at these two ROIs allows
to monitor the tangential force, which is measured by a cali-
brated strain gauge. The axial force along a specimen subject-
ed to fretting fatigue is discontinuous by a magnitude of 2Q at

Table 2 Test conditions for

fretting fatigue limit of Blocks σmax [MPa] Qmax/P [−] Block f [Hz] R (σmin /σmax) [−] P [kN] Rpad [mm]
Al2024-T3 duration
[cycles]

I 40 0.13 5000 10 0 1 50
II* 60 0.18 5000
III 80 0.24 5000
IV 100 0.29 5000
V 120 0.36 5000
VI 140 0.4 5000
VII 160 0.47 5000
VIII 180 0.5 5000
*
Fretting fatigue limit of Al2024-T3
264 Exp Mech (2018) 58:259–267

Table 3 Test conditions for

fretting fatigue limit of S235JRC Blocks σmax [MPa] Qmax/P [−] Block f [Hz] R (σmin /σmax) [−] P [kN] Rpad [mm]
duration
[cycles]

I 80 0.08 5000 10 0 1 150

II 120 0.12 5000
III* 160 0.17 5000
IV 200 0.22 5000
V 240 0.27 5000
VI 280 0.31 5000
VII 320 0.4 5000
VIII 360 0.45 5000
*
Fretting fatigue limit of S235JRC

the contact points (see Fig. 3). Equilibrium of forces in axial
direction allows the tangential force to be determined as the
difference between the axial forces above F(R3) and below
F(R4) the contact points divided by 2. Combining this with
equation (2) allows to calculate the tangential force, Q, as a
function of the temperature amplitudes at R3 and R4 as
shown in equation (3). Here, A is a cross-sectional area of
the specimen.
F ðR3 Þ− F ðR4 Þ σkk ðR3 Þ−σkk ðR4 Þ A
Q¼ ¼ A ¼ − ðT 1 ðR3 Þ−T 1 ðR4 ÞÞ
2 2 2k
ð3Þ
Average temperature data corresponding to the four regions
of interests are extracted from all recorded thermal images
(400,000 images) via the thermal CAM research software of
InfraTec. The FFT is performed with Matlab for every series
of 1024 (=210) data points (corresponding to 102.4 cycles or
10.24 s) for all the ROIs in order to determine the frequency of
the maximum temperature amplitude (the first harmonic),
which should be in the vicinity of the loading frequency.
Then, the second harmonic frequency is extrapolated so that
the temperature amplitude at the second harmonic frequency
can be determined.

A FFT algorithm is used to determine the second harmonic

temperature amplitude for each load block. The evolution of
this temperature amplitude is used to determine the load block
corresponding to the fretting fatigue limit. This damage detec-
tion technique is believed to improve the method based on the
self-heating of the specimen in which a separation of the dif-
ferent temperature contributions is not implemented [25–27].
This technique is also assumed to be more robust in case the
room temperature is not controlled.
Results of Temperature Decomposition (T1 and T2)
Figures 8 and 9 represent the evolutions of temperature
amplitudes during fretting fatigue tests on Al2024-T3 (see
Table 2) at the first harmonic and second harmonic frequen-
cies, respectively. The first harmonic temperature ampli-
tude, T1, is confirmed to be increasing with the load blocks
due to the basic thermoelastic effect. From Fig. 9, one can
see that the first two load blocks do not generate any dam-
age at all. The temperature levels T2 at the contact areas R1
and R2 are at the same level as these at the remote areas R3
and R4. During the third load block, the temperature level at

0.25
Block VIII

0.2 Block VII

T1-R1
Block VI
T1-R2
0.15 T1-R3 Block V
T1 [K]

T1-R4
Block IV
0.1
Block III
Block II
0.05 Block I

0
0 0.5 1 1.5 2 2.5 3 3.5 4
4
No of Cycles [Cycles] x 10
Fig. 7 Temperature change between the maximum and minimum axial
stress at the beginning of the 7th load block (Al2024-T3) Fig. 8 First harmonic temperature amplitude of Al2024-T3
Exp Mech (2018) 58:259–267 265

0.025 Block VII 0.2

Block VI Block VIII
Block VIII
0.02 Block VII
T2-R1 0.15
Block V T1-R1
T2-R2 Block VI
0.015 T2-R3 T1-R2
T2 [K]

T1-R3 Block V

T1 [K]
T2-R4
0.1 T1-R4
Block IV Block IV
0.01
Block III

0.05 Block II
Block III
0.005 Block I
Block I Block II

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
4
No of Cycles [Cycles] x 10 No of Cycles [Cycles] x 10
4

Fig. 9 Second harmonic temperature amplitude of Al2024-T3 Fig. 11 First harmonic temperature amplitude of S235JRC

the right contact R2 begins to change and rises with third load block. Again fretting fatigue tests with constant
increasing fatigue stress and tangential load level in stress amplitude were performed and results are shown in
the following blocks. However, a decrease in T2 value Fig. 13. The specimen remained intact at 107 cycles for the
is noticed during the last load block. This may be be- fretting fatigue test corresponding to the third load block.
cause the sliding regime shows a transition from the
partial slip regime closer to gross slip conditions (see
Fig. 5). Therefore, from Fig. 9 one can conclude that
the second load block is the fretting fatigue limit of the Conclusions
material since below this block there is no sign of fric-
tional energy loss. The test has been repeated three Fretting fatigue tests with the stepwise increased stress level
times to confirm this observation. Following, fretting (block loading) on Al2024-T3 and S235JRC specimens have
fatigue tests with constant stress amplitude were run to been performed. Each block was run for 5000 cycles, and the
verify the fretting fatigue limit. The results in Fig. 10 temperature evolution was monitored by an IR camera. The
show that during a test with the load condition of the temperature response of all the blocks has been post-processed
second load block, the specimen did not fail at in the frequency domain, using an FFT algorithm implement-
107 cycles. ed in Matlab, to separate the different physical contributions.
The same methodology has been applied to a specimen The thermoelastic effect of the regions of interest above and
made of structural steel grade S235JRC. Test conditions are below the contact zones, R3 and R4, allows to measure the
shown in Table 3. The thermoelastic temperature amplitude, tangential force, Q. On the other hand, comparison of the
T1, is again observed to increase with increasing load level second harmonic response at the contact zones, R1 and R2,
(see Fig. 11). The second harmonic temperature amplitude, to that of the remote regions, R3 and R4, gives a clear indica-
T2, at the left contact R1 deviates from the others after the tion of that loading block which does not generate fretting
fourth load block as illustrated in Fig. 12. This means that fatigue damage. The stress at the transition from that load
the fretting fatigue limit corresponds to the stress level of the block which does not show any second harmonic response
to that load block where the second harmonic response is
visible, is called the fretting fatigue limit. A best estimate of
the fretting fatigue limit of both materials could be determined
with one test lasting a few hours. One additional sample was
0.015

Block VIII

Block VII
0.01 T2-R1
Block VI
T2-R2
T2 [K]

T2-R3 Block V
T2-R4
0.005 Block IV
Block I Block II Block III

0
0.5 1 1.5 2 2.5 3 3.5 4
4
No of Cycles [Cycles] x 10
Fig. 10 Lifetime curve for Al2024-T3 under fretting fatigue with
P = 1kN Fig. 12 Second harmonic temperature amplitude of S235JRC
266
Fig. 13 Lifetime curve for S235JRC under fretting fatigue with P = 1kN
used for validation. The experimental preparation is very sim-
ple and basically non-destructive. At this moment, a rather
wide window between two load blocks is used to estimate
the fretting fatigue limit, increasing the number of load blocks
would allow a more accurate determination of the fretting
fatigue limit.
Acknowledgements This project has been funded with support from
the European Commission, Erasmus Mundus Action 2 EACEA 2012-
2635. This publication reflects the views only of the authors, and the
Commission cannot be held responsible for any use which may be made
of the information contained therein.
References
1. Attia MH (2006) Fretting fatigue and wear damage of struc-
tural components in nuclear power stations — Fitness for
service and life management perspective. Tribol Int 39:
1294–1304. https://doi.org/10.1016/j.triboint.2006.02.052
2. Wagle S, Kato H (2009) Ultrasonic detection of fretting fatigue
damage at bolt joints of aluminum alloy plates. Int J Fatigue 31:
1378–1385. https://doi.org/10.1016/j.ijfatigue.2009.03.017
3. Li YJ, Tao CH, Zhang WF, Jiang T (2008) Fracture analysis on
cage rivets of a cylindrical roller bearing. Eng Fail Anal 15:796–
801. https://doi.org/10.1016/j.engfailanal.2007.05.010
4. Gutkin R, Alfredsson B (2008) Growth of fretting fatigue cracks in
a shrink-fitted joint subjected to rotating bending. Eng Fail Anal 15:
582–596. https://doi.org/10.1016/j.engfailanal.2007.04.003
5. Tang H, Cao D, Yao H et al (2009) Fretting fatigue failure of
an aero engine turbine blade. Eng Fail Anal 16:2004–2008.
https://doi.org/10.1016/j.engfailanal.2008.07.010
6. Wagner D, Ranc N, Bathias C, Paris PC (2009) Fatigue crack ini-
tiation detection by an infrared thermography method. Fatigue Fract
Eng Mater Struct, 12–21. https://doi.org/10.1111/j.1460-2695.
2009.01410.x
7. Schijve J (2009) Fatigue of Structures and Materials. Springer
Science+ Business Media, B. V., Dordrecht
8. Kondo Y, Sakae C, Kubota M, Yanagihara K (2005) Non-
propagating crack behaviour at giga-cycle fretting fatigue limit.
Fatigue Fract Eng Mater Struct 28:501–506. https://doi.org/10.
1111/j.1460-2695.2005.00896.x
Exp Mech (2018) 58:259–267
9. Hirakawa K, Toyama K, Kubota M (1998) The analysis and
prevention of failure in railway axles. Int J Fatigue 20:135–
144. https://doi.org/10.1016/S0142-1123(97)00096-0
10. Araújo JA, Nowell D (1999) Analysis of pad size effects in fretting
fatigue using short crack arrest methodologies. Int J Fatigue 21:
947–956. https://doi.org/10.1016/S0142-1123(99)00077-8
11. Fouvry S, Nowell D, Kubiak K, Hills D a. (2008)
Prediction of fretting crack propagation based on a short
crack methodology. Eng Fract Mech 75:1605–1622.
https://doi.org/10.1016/j.engfracmech.2007.06.011
12. Kubota M, Kataoka S, Takazaki D, Kondo Y (2017) A quantitative
approach to evaluate fretting fatigue limit using a pre-cracked spec-
imen. Tribol Int 108:48–56. https://doi.org/10.1016/j.triboint.2016.
10.017
13. Giancane S, Chrysochoos A, Dattoma V, Wattrisse B (2009)
Deformation and dissipated energies for high cycle fatigue of
2024-T3 aluminium alloy. Theor Appl Fract Mech 52:117–121.
https://doi.org/10.1016/j.tafmec.2009.08.004
14. Boulanger T, Chrysochoos A, Mabru C, Galtier A (2004)
Calorimetric analysis of dissipative and thermoelastic effects asso-
ciated with the fatigue behavior of steels. Int J Fatigue 26:221–229.
https://doi.org/10.1016/S0142-1123(03)00171-3
15. Bubulinca C, Balandraud X, Grédiac M et al (2014)
Characterization of the mechanical dissipation in shape-memory
alloys during stress-induced phase transformation. J Mater Sci
49:701–709. https://doi.org/10.1007/s10853-013-7751-5
16. Benaarbia A, Chrysochoos A, Robert G (2015)
Thermomechanical behavior of PA6.6 composites subjected
to low cycle fatigue. Compos Part B Eng 76:52–64.
https://doi.org/10.1016/j.compositesb.2015.02.011
17. Risitano A, Risitano G (2010) Cumulative damage evaluation of
steel using infrared thermography. Theor Appl Fract Mech 54:82–
90. https://doi.org/10.1016/j.tafmec.2010.10.002
18. Fargione G, Geraci A, La Rosa G et al (2002) Rapid determination
of the fatigue curve by the thermographic method. Int J Fatigue 24:
11–19. https://doi.org/10.1016/S0142-1123(01)00107-4
19. La Rosa G, Risitano A (2000) Thermographic methodology for
rapid determination of the fatigue limit of materials and mechanical
components. Int J Fatigue 22:65–73. https://doi.org/10.1016/
S0142-1123(99)00088-2
20. Luong MP (1998) Fatigue limit evaluation of metals using an
infrared thermographic technique. Mech Mater 28:155–163.
https://doi.org/10.1016/S0167-6636(97)00047-1
21. Micone N, De Waele W (2016) On the Application of Infrared
Thermography and Potential Drop for the Accelerated
Determination of an S-N Curve. Exp Mech 57:143–153.
https://doi.org/10.1007/s11340-016-0194-6
22. Colombo C, Vergani L, Burman M (2012) Static and fatigue char-
acterisation of new basalt fibre reinforced composites. Compos
Struct 94:1165–1174. https://doi.org/10.1016/j.compstruct.2011.
10.007
23. Montesano J, Fawaz Z, Bougherara H (2013) Use of infrared
thermography to investigate the fatigue behavior of a carbon
fiber reinforced polymer composite. Compos Struct 97:76–83.
https://doi.org/10.1016/j.compstruct.2012.09.046
24. Krapez J (1998) Compared performances of four algorithms used
for modulation thermography. Quant Infrared Thermogr 4:148–153
25. Krapez JC, Pacou D, Gardette G (2000) Lock-in thermography and
fatigue limit of metals. Quant Infrared Thermogr 5:277–282
26. Brémond P (2004) IR imaging assesses damage in mechanical
parts: determining the fatigue limit of real structures under real
operating conditions saves time. Photonics Spectra, Pittsfield
27. Bremond P, Potet P (2001) Lock-In Thermography: A tool to ana-
lyse and locate thermo-mechanical mechanisms in materials and
structures. In: Society of Photo-Optical Instrumentation Engineers
(SPIE). pp 560–566
Exp Mech (2018) 58:259–267
28. Ummenhofer T, Medgenberg J (2009) On the use of infrared ther-
mography for the analysis of fatigue damage processes in welded
joints. Int J Fatigue 31:130–137. https://doi.org/10.1016/j.ijfatigue.
2008.04.005
29. Forsyth D, Genest M, Shaver J, Mills T (2007) Evaluation of
nondestructive testing methods for the detection of fretting dam-
age. Int J Fatigue 29:810–821. https://doi.org/10.1016/j.
ijfatigue.2006.09.012
30. Berthel B, Moustafa A-R, Charkaluk E, Fouvry S (2014) Crack
nucleation threshold under fretting loading by a thermal method.
Tribol Int 76:35–44. https://doi.org/10.1016/j.triboint.2013.10.008
31. Chhith S, De Waele W, De Baets P, Van Hecke T (2017) On-line
detection of fretting fatigue crack initiation by lock-in thermography.
Tribol Int 108:150–155. https://doi.org/10.1016/j.triboint.2016.10.019
267
32. De Pauw J, De Baets P, De Waele W (2013) Design of a fretting
fatigue test rig. J Sustain Constr Des 4:6. https://doi.org/10.21825/
scad.v4i2.1074
33. Hojjati-talemi R, Wahab MA, De Pauw J, De Baets P (2014)
Prediction of fretting fatigue crack initiation and propagation life-
time for cylindrical contact configuration. Tribol Int 76:73–91.
https://doi.org/10.1016/j.triboint.2014.02.017
34. Fouvry S, Kapsa P, Vincent L (1996) Quantification of fretting
damage. Wear 200:186–205. https://doi.org/10.1016/S0043-
1648(96)07306-1
35. Szolwinski MP, Harish G, Farris TN, Sakagami T (1999) In-
Situ Measurement of Near-Surface Fretting Contact
Temperatures in an Aluminum Alloy. J Tribol ASME 121:
11–19. https://doi.org/10.1115/1.2833942