Tribology International xx (xxxx) xxxx–xxxx

Contents lists available at ScienceDirect

Tribology International
journal homepage: www.elsevier.com/locate/triboint

Investigations on the fretting fatigue failure mechanism of bolted joints in

high strength steel subjected to different levels of pre-tension
⁎
Carlos Jiménez-Peñaa, , Reza H. Talemib, Barbara Rossic, Dimitri Debruynea
a
KU Leuven, Department of Materials Engineering, Kasteelpark Arenberg 44, 3001 Leuven, Belgium
b
ArcelorMittal Global R & D Gent-OCAS N.V., Pres. J.F. Kennedylaan 3, 9060 Zelzate, Belgium
c
KU Leuven, Department of Civil Engineering, Kasteelpark Arenberg 40, 3001 Leuven, Belgium

A R T I C L E I N F O A BS T RAC T

Keywords: Bolted connections are widely used as an effective joining method with higher fatigue strength than equivalent
Fretting fatigue welded and riveted joints. The fretting fatigue problem caused by components in contact subjected
Pre-tension simultaneously to fluctuating loads and relative small movements is of special importance in the vicinity of
HSS bolted joints. In the present paper, a series of fatigue tests on bolted connections using moderately thick plates
Bolted joint
made of high strength steel S500MC carried out using different pre-tension levels is described. It was shown
Clamping force
that fretting fatigue decreased when a higher pre-tension force is used. In addition, a 3D finite element model
was built to assess the stress gradient in the connection and several multiaxial criteria were used to determine
the location of the initial crack. These numerical results were correlated to the experimental results and showed
acceptable predictions.

1. Introduction weld defects – which is critical in mobile structures subjected to

Reducing the weight of the structural components has always been
a target in structural design, to allow for a decrease of materials
consumptions provided that proper mechanical strength is maintained.
This has a particular importance in vehicles and heavy duty machinery,
such as trucks, excavators and agricultural machinery, where lighter
structures result in higher load carrying capacity and less fuel
consumption. It has hence become a current trend to employ high
strength steels (HSS) in the components of these structures, leading to
a reduction of their weight [1,2].
HSS are an increasingly used generation of steels exhibiting
improved properties compared to conventional steel grades. Their
enhanced mechanical properties (e.g. yield strength varying from
460 N/mm2 to 700 N/mm2 and higher) originate from their complex
microstructure obtained thanks to specific thermo-mechanical treat-
ments (e.g. quenched and tempered). The use of HSS with yield
strength up to 460 MPa in construction and mechanical applications
has increased during the last decades thanks to its structural advan-
tages [3–10]. On the contrary, the use of very high strength grades,
with yield strengths above 690 MPa, in civil and mechanical engineer-
ing structures is relatively uncommon especially because of restrictive
design rules and welding problems [11,12]. The susceptibility of HSS to
alternating stresses – indeed requires special post weld treatments to
be carried out [13]. In order to avoid this problem, bolted connections
are seen as a good alternative to welded equivalents. The fatigue
strength of HSS bolted connections and especially their fretting fatigue
behaviour remains however scarcely investigated in the scientific
literature.
Bolted connections are used as an effective and versatile joining
technique in a variety of engineering applications. Design rules for
bolted connections are widely available in European standards [14,15]
together with extensions to include higher steel grades [16]. The fatigue
strength curves and detail categories in [17] are mainly based on
fatigue tests carried out on bolted and welded carbon steels details with
nominal yield stress ranging from 235 to 400 MPa i.e. mainly S235 and
S355. And, even if the part 1–9 may cover higher structural steel
grades given in EN 1993-1-12 (S235–S700), the phenomena governing
the fatigue resistance of HSS welded joints are currently not perfectly
known. In [18], Puthi el at. investigated the effect of bolt spacings and
edge distances in bolted connections made of S460 and underlined that
the current guidelines may lead to overly conservative design dimen-
sions. Moze et al. [19,20] came up to similar results for S690 double
shear plane bolted connections. Shi et al. [21] tested bolted connec-
tions made of Q460 and concluded that the design guidelines from [17]

⁎
Corresponding author.
E-mail addresses: carlos.jimenezpena@kuleuven.be (C. Jiménez-Peña), reza.hojjatitalemi@arcelormittal.com (R. H. Talemi), barbara.rossi@kuleuven.be (B. Rossi),
dimitri.debruyne@kuleuven.be (D. Debruyne).

http://dx.doi.org/10.1016/j.triboint.2016.11.014
Received 1 July 2016; Received in revised form 2 November 2016; Accepted 4 November 2016
Available online xxxx
0301-679X/ © 2016 Elsevier Ltd. All rights reserved.

Please cite this article as: Jiménez-Peña, C., Tribology International (2016), http://dx.doi.org/10.1016/j.triboint.2016.11.014
C. Jiménez-Peña et al.
CLAMPING LOAD
WASHERS
FIXED PLATE
NUT
60
25.5
Φ17
51
25.5
Fig. 1. (a) Assembly view of the SLBJ specimen (b) Geometry of the dog-bone HSS specimen.
provide unsafe bearing capacity and cannot appropriately predict the
failure modes. Despite the fact that the existing literature on fatigue of
bolted connections in HSS is currently limited, Hämäläinen [22] has
reported the susceptibility to fretting fatigue of bolted connections
made of S960QC and the possible challenges linked to it. Fretting
fatigue damage is reported as one of the principal failure mechanisms
in bolted connections [23]. In slip-resistant joints, small oscillatory
relative movements cause fretting between surfaces in contact. This
phenomenon leads to very high local stresses and wear, which can lead
to premature crack nucleation.
Several factors affect the behaviour of the bolted joint and its
resistance to fatigue. The pre-tension force (also known as preload or
clamping force), which is applied to the bolt and compresses the joined
components against each other, is one of the most significant para-
meters. It creates frictional forces among the mated surfaces to carry
shear loads. Many reference studies [24–28] were achieved on the
effect of the preload on the fatigue life of bolted connections, for a
broad range of materials and joint configurations. However, contra-
dictory results have been reported. For example, Valtinat et al. [27]
concluded that the pre-tension of the bolt has a major positive
influence, as the subsequent high pressure produces a more favourable
stress state around the hole in carbon steel members. As for Minguez in
[28], he indicates that the increase of the preload force in single lap
joints made of aluminium alloys had no significant repercussion on the
fatigue life, whereas in double lap joints, the higher the preload the
higher the fatigue resistance. Moreover, Benhamena [29] reported the
negative effect of the preload force on the fatigue life of bolted
connections made of steel and aluminium. The great majority of these
investigations focus on aluminium and titanium alloys due to their
importance in the aerospace industry, whereas little research has, so
far, been addressed to the effect of the preload force in connections
made of HSS. Overall, there exists currently a lack of knowledge on the
fatigue strength of bolted connection made of HSS and a confusion on
the effect of the preload force on the fatigue strength of bolted
connections.
As the contact interface in a bolted joint is concealed between the
Tribology International xx (xxxx) xxxx–xxxx
M16
BOLT
CRITICAL AREA AXIAL LOAD
PLATE
317.33
184.17
R51
STEEL PLATE
102
THICKNESS= 6 mm
plates, Finite Element Analysis (FEA) allows to study many more
topics, such as the distribution of slip and frictional shear stress under
different loading conditions, than tests performed in a laboratory. It
has been shown that the FEA modelling approach is a suitable
methodology to investigate the behaviour of bolted connections under
different contact conditions [30–33]. Several authors have used this
approach to carry out numerical studies including a thorough repre-
sentation of the contact problems in bolted connections [24–26]. In
[34], Esmaeili used the data extracted from FEA to evaluate the
suitability of different multiaxial criteria for the prediction of fatigue
strength in bolted connections. Although their research did not focus
on fretting fatigue, they demonstrated that some of the studied multi-
axial criteria were appropriate for estimating the fatigue life.
This research aims to extend the existing knowledge about the
failure mechanisms of HSS bolted connections submitted to fretting
fatigue and to evaluate the effect of the preload force on fretting fatigue
primarily variables, such as the slip amplitude or damage at the contact
interface, among others. For this purpose, this paper is organised as
follows: Preceded by a description of the experimental set-up and of the
materials under investigation in Section 2, the Finite element model
(FEM) of one single bolted lap joint is thoroughly described in Section
3. The results of the study are presented in Section 4 where the
experimental observations are shown along with the results from the
numerical model. In Section 5, the results are analysed and conclusions
for this study are provided.
2. Experimental study
2.1. Material, geometry of the specimens and preload forces
The experimental specimen is a single-lap bolted joint (SLBJ)
consisting of two identical plates made of S500MC, a 10.9 M16 bolt
and its respective nut (EN 14399-4) [35] and two washers (EN 14399-
6) [36] placed on both sides. The geometry of the plates is depicted in
Fig. 1. The end of the plates has a dog bone like shape in order to
ensure a smooth stress flow. The edge distances have been chosen in
2
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

Table 1
Chemical [38] and mechanical characteristics of HSS S500MC.

Chemical composition [wt%]

C Si Mn P S Al Nb V Ti Mo B

≤0.120 ≤0.500 ≤1.700 ≤0.025 ≤0.015 0.015 ≤0.090 ≤0.200 ≤0.150 – –

Mechanical properties

Elastic modulus [GPa] Yield stress [MPa] Ultimate tensile stress [MPa] Poisson’s coefficient Elongation [%]

219.0 501.4 669.4 0.29 16.68

accordance with the Eurocode standard (EN 1993-1-8) for bolted joints loss of clamping force. The strain was monitored for the cyclic tests
[14]. The chemical and mechanical characteristics of the S500MC steel with higher stress amplitudes, where more loss of pretension is
are provided in Table 1. The mechanical properties of the tested expected. The sampling rate of the strain gauge was 0.5 Hz.
samples were deduced from previous experiments performed in our
group by Denys et al. [37]. In this study tensile tests on HSS steels were 2.2. Testing set-up
performed and Digital Image Correlation (DIC) coupled with Finite
Element Model Updating (FEMU) were used to identify the material 2.2.1. Fatigue test
properties. The obtained material properties for the S500MC were not The bolted joints were tested with a Zwick HA100 servo-hydraulic
published but are available. The measured mechanical properties fall load frame with a maximal axial load of 100 kN. A side view of a
into the range of those provided by the catalogue [38]. mounted specimen in the testing rig is shown in Fig. 2. It can be noted
The plates were pickled with an 18% HCl solution for 5 min in order that the bolted plates were clamped eccentrically. Both upper and lower
to remove the oxide layer produced during the hot rolling process. The fixture clamps can be adjusted to allow for this eccentricity. The test
plate surfaces were then visually inspected for any abnormal defects. matrix is shown in Table 2. For each nominal stress range and each
They were also wiped with acetone prior to assembling. The plates had pre-tension level, two fatigue tests were performed leading to 24
a thickness of t=6 mm and were cut by abrasive waterjet. Although the fatigue experiments in total. The load ratio was kept constant at
abrasive waterjet cut-edges can show a poor fatigue life performance as R s =0.1 so that the specimen was never subjected to compressive axial
noted by Thomas in [39], for SLBJ the potential crack initiation site is force. The frequency for all the tests was f=25 Hz. In addition, the SLBJ
at the contact interface and not at the edge of the specimen. As it is
shown later on, the expected fretting fatigue crack initiation site was
positioned at a distance from any machined edge. As none of the tested
specimens showed crack propagating from the machined edges, water-
jet cutting proved to be an acceptable cutting process in this study. In
addition, this cutting technique produces no thermal alteration on the
material.
As an axial force is applied to the SLBJ it may occur that a fast
relative slip takes place between both plates. This would cause a violent
collision between the bolt and the plates, which is irrelevant for the
present study. In order to avoid this problem, the plates were fastened
together in a pre-slipped position.
The plates shown in Fig. 1(b) were fastened together by applying a
preload force to the bolt with an electronic torque wrench with an
accuracy of ± 2%. Three different tightening torques were considered
in this study: 200, 300 and 350 N m. The pre-tension load produced by
the tightening torque applied to the bolt can be calculated from Eq. (1)
[40].

Mr
Fp, cd =
km d (1)

where Mr , Fp, cd , k m and d are the tightening torque (N m), the preload
force (kN), the k-class declared by the manufacturer and the nominal
diameter of the bolt (mm), respectively. A typical value of k m=0. 2 was
used, as suggested in [22]. This resulted in three different preload
forces: 62.5, 93.75 and 109.38 kN. These values corresponds to,
respectively, 58%, 88% and 100% of the design preload force specified
in the Eurocode standard (EN 1993-1-8. Eurocode 3) for slip-resistant
connections [14]. Although a degree of uncertainty is attached to the
calculation of the pre-tension, this approach has been used by other
authors [22,41]. Additionally, a special bolt with a strain gauge
attached to its shank has been used to monitor the strain loss during
the cyclic loading. Assuming that the deformation in the shank is
elastic, the measured strains can be linearly related to the clamping
force. Therefore, a loss of strain in the shank will be directly related to a Fig. 2. SLBJ specimen mounted on the servo-hydraulic load frame.

3
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

Table 2 2.3. Digital image correlation measurements

Test matrix for the SBLJ fatigue tests, two fatigue tests per load pre-tension were
performed.
To avoid inaccuracy related to the deflection of the testing machine
Pre-tension Level crosshead, the displacement field occurring in each bolted plate was
measured locally using the digital image correlation (DIC) technique.
200 Nm 300 Nm 350 Nm DIC is an optical method for displacement and strain measurements
that analyses and compares digital images acquired from the surface of
Label Axial load Nominal stress 62.50 kN 93.75 kN 109.38 kN
[kN] amplitude [MPa]
a specimen [47]. Due to rapid new developments in high resolution
digital cameras for static as well as dynamic applications, and computer
A 50 110 A/200 A/300 A/350 technology, the applications for this measurement method broadened.
B 40 88 B/200 B/300 B/350 DIC technique has proven to be a flexible and useful tool for
C 35 77 C/200 C/300 C/350
D 30 66 D/200 D/300 D/350
deformation analysis. It is used to monitor the displacement field,
which is calculated based on the best correlation between the image of
a deformed surface and a reference image of the un-deformed surface.
was also tested under quasi-static loading conditions, up to a load of This calculation is based on the Sum of Squared Differences (SSD)
90 kN at a loading rate of 0.1 kN/s. correlation criterion. The correlation process is only feasible if the
The fractured surfaces of part of the specimens were scanned and surface of the specimen is covered by a random pattern, which is
measured with a Nikon InSight L100 Coordinate Measuring Machine usually applied with aerosol paint.
(CMM). The digital images were taken by two digital cameras, with a
resolution of 2752×2206 pixels, mounted on a tripod positioned
vertically in front of the load frame. Both cameras had a lens with a
focal length of 25 mm. The angle between the two cameras was 14.88°.
2.2.2. Measurement of the coefficient of friction Two halogen lamps were used to improve lightning and contrast. In
An additional test was performed to characterize the coefficient of order to reduce any possible thermal interference caused by the heat,
friction (COF) during the fatigue test. Many authors [42–46] have the lamps were placed behind the cameras. The field of view was
reported a rapid increase of the COF at the initial stages of testing. The approximately 180×145 mm wide. The side of the SLBJ specimen in
initial low COF can be attributed to the presence of surface oxide films which the bolt head is located was completely covered with a speckle
or asperities. The surface topology is modified during contact interac- pattern, excluding the clamping area of the plates. Fig. 3(a) shows the
tions and once these features are removed, the interactions metal- speckled specimen mounted onto the servo-hydraulic load frame. The
metal start promoting an increase of the COF. Consequently, the SLBJ circular markers were attached to the clamps and were used to monitor
has to initially be subjected to a cyclic loading to properly assess the the possible slip between the clamps and the plates in case of false
COF during the fatigue test. gripping. Fig. 3(b) illustrates the mounted specimen, the digital
A SLBJ specimen was assembled using the strain-monitoring bolt cameras and the lamps. The software used to analyse the images was
introduced previously in Section 2.1. The tightening torque in the bolt MatchID Stereo [48]. A summary of the DIC calculation settings is
was 300 N m. The SLBJ was subjected to a dynamic sinusoidal displayed in Table 3.
displacement with an amplitude of 0.3 mm. A total of 1000 cycles with
a frequency f=25 Hz were applied. After the cyclic loading, a constant 3. Modelling
displacement of 0.2 mm/min was applied to the SLBJ specimen, until
gross sliding occurred between the bolted plates. 3.1. Model description
The coefficient of friction µ was obtained by dividing the tangential
force at the onset of gross sliding by the normal force in Eq. (2), as done A finite element model of the SLBJ was built using the commercial
by De Pauw et al. [43]. The onset of gross sliding was manually software Abaqus v6.13. Fig. 4 shows the 3D FEA model along with the
identified from the loading curve. applied loading and the boundary conditions. The model consists of six
parts: the two steel plates, the bolt, the nut and the two washers. Only
FT half of the geometry was modelled due to the symmetry along the XY
μ=
FN onsetgrosssliding (2) plane. C3D8R linear brick elements with reduced integration were
used. As the vicinity of the bolt hole was the critical zone to be
The normal force was obtained by reducing the calculated clamping analysed, the density of the mesh was appropriately refined in this
force in Eq. (1) by the measured strain loss during the cyclic loading. region, as shown in Fig. 4. The minimum mesh size at the contact

CAMERAS

FIELD OF
VIEW

LAMPS

Fig. 3. (a) Specimen with speckled pattern mounted onto the servo-hydraulic load frame together with track patterns attached to the clamps. (b) DIC set-up including two digital
cameras, two halogen lamps and the mounted specimen.

4
C. Jiménez-Peña et al.
Table 3
Parameters used in DIC post-processing analysis.
Parameter Value
Correlation coefficient Zero-Normalized SSD
Interpolation order Bicubic spline
Transformation order Affine
Subset size [pixels] 21
Step size [pixels] 10
Noise handling Gaussian
Kernel size 5 components is maximal. Critical plane approaches attempt to reflect
History Spatial
interface (achieved after a convergence study as elaborated later on)
was 0.3 mm and increased gradually from contact. All parts of the
model had the elastic material characteristics provided in Table 1. The
contact interaction law of the plates, with either the opposite plate or
the washers, were modelled as surface-based contact pairs and using
the Lagrange multiplier method for imposing frictional constraints. A
friction coefficient value of 0.8 was selected based on experimental
evidence, as discussed later on in Section 4.1.5. The rest of the contact
pairs were modelled with tie constraints to reduce the computation
time.
The analysis was executed in three time steps. During the first step,
both plate ends were completely restrained, while the pre-tension was
applied to the bolt. In order to apply the pre-tension load, a pre-tension
surface was defined in the bolt, as shown in Fig. 4. When modelling the
bolt with solid elements, the pre-tension surface is defined as a surface
in the bolt shank that effectively partitions the bolt into two regions. In
continuum elements, the pre-tension surface is defined as a surface
inside the fastener that divides it into two parts. The pre-tension
surface can also be a group of surfaces when a fastener is composed of
several segments. Therefore, in the FEM, the pre-tension load was
modelled through a self-equilibrating concentrated load applied over
the pre-tension surface in the bolt shank, at each the pre-tension nodes
[49]. During the second step, the length of the bolt was fixed to the one
obtained after the first step, so that the bolt may act as a passive
connector element. In the third step, a displacement condition was
gradually applied to the end of the right plate. The displacement
condition is intended to simulate the displacement produced by the
axial force in the experimental test. Displacement control is preferred
over force control in the analysis as it provides greater stability and
computational economy. This step represents a steady-state loading
condition during the fatigue cycle.
Y X
FIXED SIDE
BOLT
SYMMETRY
PLANE
WASHERS
NUT
Fig. 4. FEM of SLBJ along with the applied loading and boundary conditions and detailed view of the mesh at the contact interface.
5
Tribology International xx (xxxx) xxxx–xxxx
3.2. Multiaxial fatigue criteria
During fretting conditions, multiaxial stresses are produced around
the contact interface, often together with non-proportional variation of
the stresses [50]. Several authors [51–54] have proposed the applica-
tion of multiaxial criteria to predict the initiation and location of
fretting fatigue crack at the contact interface. Many of the multiaxial
criteria are based on the critical plane approach, in which the critical
plane is defined as the plane where a combination of stress or strain
the physical nature of fatigue damage (i.e. damage mechanisms) in
their formulation. Non-critical plane approaches however, have the
advantage of their simplicity and shorter computational times. In this
study 4 different approaches have been used to find the crack initiation
site in the SLBJ under fatigue loading. The multiaxial criteria were
implemented by post-processing the FEA results using the python
scripting interface in Abaqus. The numerical results were extracted
from each element/node recursively in all the steps. For the critical
plane criteria, the critical plane was found by rotating the stress tensor
in increments of 5° as in [53].
3.2.1. Smith-Watson-Topper
The damage criterion defined by Smith [55] considers the plane
with maximal strain range throughout one load cycle as the critical
plane. The multiaxial criterion, usually referred as Smith-Watson-
Topper (SWT) is defined as:
∆εI
SWT = σn,max
2 (3)
where ∆εI and σn,max are the maximum normal stress and the maximum
main strain range, respectively. When the loading is not proportional,
as it occurs in fretting fatigue, the application of this parameter is
problematic as the main direction of the strain is varying during the
loading. In this case, the SWT parameter is defined as [56]:
⎛ ∆ε ⎟⎞
SWT = ⎜σn,max
⎝ 2 ⎠max (4)
where σn,max and ∆ε max are the strain range and the maximum normal
stress in the same plane, respectively. The critical plane is thus
determined as the plane in which the product of these two terms is
maximal.
3.2.2. Fatemi-Socie
Fatemi [57] determined the critical plane as the plane with
maximum shear strain range during one loading cycle. Fatemi and
Socie (FS) criterion is expressed as:
F /2
PRETENSION
F /2 SECTION
STEEL PLATES
DISPLACEMENT
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

Fig. 5. Normalized tangential stress distribution along the normalized centreline of the specimen for different mesh sizes.

PLATE [Washer side] BOLTED PLATE [Washer side]

ASSEMBLY

Slipping Area

Sticking Area

Non-contact
Area
Non-contact
Area

(a) (b) (c)

Fig. 6. Sample C/350: (a) Damaged Plate-Washer interface, (b) section view of the pre-slipped SLBJ and (c) fretting fatigue surface in Plate-Plate interface.

0.8 mm

Crack path

Mixed mode

A-A

MULTIPLE CRACK INITIATION SITES 5 mm

FATIGUE CRACK

A-A
TENSILE FRACTURE
A-A

Fig. 7. Sample A/300: (a) Top view and 3D scan of crack surface and (b) Measured crack propagation path.

∆γmax ⎛ σ ⎞ shorter, as there is no need to search for its value among all possible
FS = ⎜1 + k n,max ⎟
2 ⎝ σy ⎠ planes. The Crossland criterion is defined as:
(5)
I1,max ⎛ 3t ⎞
where ∆γmax is the maximum shear strain range, k is a constant fitted Crossland = J2,a + ⎜ − 3⎟
from uniaxial and torsion test data and σn,max is the maximum normal 3 ⎝b ⎠ (6)
stress in the plane where ∆γmax is produced. The value of the constant k where J2,a is the amplitude of the second invariant of the deviatoric
will be assumed to be unit for this research. stress tensor, I1,max is the maximum of the first invariant of the stress
tensor, t is the shear fatigue limit and b is the bending fatigue limit. The
term containing the material properties can be grouped into one
3.2.3. Crossland 3t
material constant k = ( b − 3 ), which will be assumed to be unit for
The criterion proposed by Crossland [58] is not a critical plane
this research as in [34].
method as it based on the values of the invariant components of the
stress tensor, which are independent from the tensor orientation. The
implementation of this criterion in the FEM analysis is more straight- 3.2.4. Ruiz
forward than the previous methods and the computational time is Ruiz et al. [59] proposed a parameter to estimate the location of the

6
C. Jiménez-Peña et al.
Fig. 8. (a) Average crack distance to hole edge at different stress amplitudes. (b) Average crack distance at different preload torques.
Fig. 9. Fatigue test results of SLBJ at three different pre-tensions along with plain
fatigue test results.
initial crack which is only applicable to fretting fatigue conditions. The
parameter is a measure of the frictional work and the additional effect
of the tangential stress at the contact interface. The fretting fatigue
damage (FFD) parameter is defined as:
FFD = σx τδ
where σx , τ and δ are the tangential stress, the frictional shear stress
and the relative displacement at the contact interface respectively.
3.3. Mesh convergence study
The mesh must be fine enough at the contact interface, where high
stress gradients are expected, to provide accurate results.
Consequently, several analyses with decreasing mesh size in steps of
0.05 mm were performed until convergence in the stress state was
achieved. The maximum tangential stress (σxx) was measured during
successive mesh refinement steps and convergence was established
when the difference between two consecutive values was lower than
2%. Similar convergence values were suggested in [60,61]. The
maximum tangential stress was monitored along the centreline of the
specimen in the loading direction. The results from the mesh conver-
gence study show that the optimal mesh size at the contact interface
was 0.3×0.3 mm (as depicted in Fig. 5).
Tribology International xx (xxxx) xxxx–xxxx
4. Results and discussions
4.1. Experimental results
4.1.1. Surface damage
The faying surface of the plates was inspected after the cyclic
loading. Some damage was observed in the contact area between the
plates and the washers due to the application of the clamping force, as
it can be observed in Fig. 6(a). However, there was no clear indication
of fretting damage or excessive wear. Fig. 6(c) provides the schematic
view of the fretting fatigue failure mechanism in SLBJ which was
observed for all the samples in the plate-plate interface. The surface of
the plates displayed two distinguished concentric regions of sticking
and slipping around the hole, with fretting scars at the slip region.
Fretting oxidized particles were found around the slipping region and
accumulated in the fretting scars. Next to the sticking area there was a
semi-circular area in which contact has not occurred between plates.
This is caused by the pre-slipped assembling of the SLBJ, previously
discussed in Section 2.1. Fig. 6(b) displays a section view of the pre-
slipped SLBJ where the non-contact regions can be noticed.
Fig. 7(a) shows the fractured surface of the broken sample together
with the scanned 3D surface. Multiple crack initiation sites were visible
along the fracture surface and were distributed along both sides of the
(7) centreline of the specimen. The cracks propagated until the effective
section considerably diminished. At that point, the applied tensile
stress produced a ductile fracture at the edges of the plate, where large
plastic deformations and necking could be observed. Crack propagation
paths were also measured. Fig. 7(b) displays a characteristic crack
propagation path observed in the sample A/300. The cracks were found
to propagate initially in mixed mode upon a distance of approximately
half of the plate thickness. Then, they continued to propagate in
opening mode (or mode I). The average stick region area for the three
different pre-tensions was measured and an increase of approximately
30% of the sticking area is observed when the pre-tension load is
augmented from 63.5 kN to 93.75 kN. The value of the sticking area
seems to reach a maximum as further increase in the pre-tension force
does not significantly modify the result. The average distance from the
crack to the hole edge at different stress amplitudes is shown in
Fig. 8(a). The results indicate a general trend of decreasing distance of
the crack to the hole edge with larger stress amplitudes. This trend is
less evident for the samples submitted to a torque of 350 N m due to a
considerable deviation in the values at a stress amplitude of 88 MPa.
The average crack distance to the hole edge for the three level of pre-
tension is represented in Fig. 8(b). An increase of the average crack
distance with increasing preload torque can be observed.
4.1.2. Fretting fatigue
The results for the fatigue test sequence at different pre-tensions
7
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

Fig. 10. Fretted surfaces of tested samples at five levels of stress amplitude and under three different bolt preload forces.

Fig. 11. (a) Static axial load versus displacement for SLBJ, (b) comparison between FEA model and SLBJ experiment.

are displayed in a log-linear ΔS-N diagram in Fig. 9. All 24 specimens
exhibited fretting fatigue damage and 23 of them failed due to fretting
fatigue cracks initiation and propagation. The fatigue tests were
stopped after the total fracture of one of the plates in the connection.
Cracks were observed in the plate-plate interface of both bolted plates.
The upper plate (closer to the bolt head) fractured 50% of the times
while the lower plate (closer to the nut), 38% of the times. In some of
the samples (12%) fracture was observed for both plates simulta-
neously. This occurred always at the maximal stress amplitude (A
series). One of the specimens tested at the low nominal stress level and
with the higher pre-tension level (E/350) did not fail before 5 million
cycles. From the experimental results, it can be concluded that an
increase in the preload force increases the fretting fatigue life of the
SLBJ. For a stress amplitude of 66 MPa, an increment of 75% in the

8
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

fretting fatigue scars and that the crack initiation site was always
located at the intersection between the sticking and slipping regions
and at the side of the free edge. As mentioned above, when the bolt
preload force increases, so is the sticking region and, consequently, the
distance between the crack initiation site and the edge of the hole.

4.1.3. DIC
The measurements from DIC technique provide the displacement
field over the surface of both plates during loading. The displacement at
the edge of the plates next to the clamping area was extracted from the
displacement field and plotted versus the axial load, as shown in
Fig. 11(a). It can be noted that, initially, the behaviour is linear, but
that, after a defined axial load, a sudden increment in the displacement
caused by the relative slip between the mating plates occurs.
Afterwards, the bolt starts to bear on the plate and significant
elongation of the hole and of the plate is observed. The axial load
value at which slipping occurs mainly depends on the preload force and
Fig. 12. Clamping force in the bolt during fatigue test at Δσ/2=110 MPa.
the coefficient of friction. The slope of the numerical axial force versus
displacement curve was compared with the experimental loading curve
(see Fig. 11(b)). A good correlation between both slopes in the initial
elastic region was reached.

4.1.4. Strain monitoring

The results from the clamping force monitoring test at a stress
amplitude of 110 MPa are shown in Fig. 12. The behaviour of the
clamping force during the cyclic loading was similar for the three
pretension levels. Initially there is was sudden decrease in the clamping
force that got stabilized after approximately 10,000 cycles. The clamp-
ing force during most part of the fatigue test remained at around 97%
of the original clamping load. Near the end of the test the clamping
force started to decrease again until the specimen was fractured. Since
the clamping force did not considerably diminish during the tests, the
normal force was considered as the originally calculated in Eq. (1).
Fig. 13. Axial force and clamping force during COF quantification test.

4.1.5. Coefficient of friction

preload force resulted in an increase of 400% of the fatigue life.
Fig. 13 shows the axial load and the preload percentage during the
Moreover, the obtained lifetimes were compared with the plain fatigue
cyclic test time. The axial load values were extracted each 100 cycles. It
results for flat specimens made of S500MC. As expected, the total
can be noted that during the first 300 cycles there was an increase in
lifetime of SLBJ under fretting fatigue loading condition is consider-
the axial load amplitude. The preload force on the other hand, abruptly
ably less than the plain fatigue lifetime.
decreased at the beginning of the cyclic test and then stabilized at
The fretted surfaces of the tested plates are shown in Fig. 10 for one
around 200 cycles. After the cyclic loading, the specimen was statically
of the replicates. It can be observed that all the samples presented
loaded and gross sliding took place at 50.32 kN. The normal force at

Fig. 14. FEM results in the contact surface of the plate for Δσ =77 MPa: (a) contact pressure, (b) tangential stress, (c) frictional shear stress and (d) contact status.

9
C. Jiménez-Peña et al. Tribology International xx (xxxx) xxxx–xxxx

Fig. 15. Results along the centreline of the specimen for (a) contact pressure, (b) frictional shear stress and (c) tangential stress, subjected to three tightening torques.

Fig. 16. Multiaxial parameter results for (a) SWT, (b) FFD, (c) Crossland and (d) Fatemi-Socie, together with tested specimen at Fcl =97.75 kN at Δσ=77 MPa.

the point of gross sliding was 65.34 kN. The resulting coefficient of
friction after 1000 cycles was determined to be 0.77.
Prior to this test, the SLBJ specimen was statically tested until gross
sliding occurred in order to obtain the initial coefficient of friction. This
caused a decrease in the preload force which can also be noted in
Fig. 13. The initial coefficient of friction was found to be 0.28.
4.2. Numerical results
4.2.1. Stress and contact variables
The numerical results were evaluated at the contact interface of the
plates, in the surroundings of the hole. In Fig. 14(a) the contact
pressure distribution is shown. The maximum contact pressure value
takes place at the hole edge and decreases when going farther from the
hole. Fig. 14(b) and (c) show the distribution of tangential and
frictional shear stresses at the contact interface between both plates.
As it can be noted a high stress gradient is located at a distance from
the hole near the contact edge (i.e. the side where the axial stress is
applied), which appears to be the potential location of initial crack in
experimental results, as it will be discussed later on. Fig. 14(d) reveals
the same characteristic sticking and slipping circular regions observed
in the tested samples. The numerical results were also evaluated at the

10
C. Jiménez-Peña et al.
Fig. 17. Multiaxial parameter results along the centreline of the specimen for three levels of pre-tension: (a) FFD, (b) SWT, (c) Crossland and (d) Fatemi-Socie.
Fig. 18. Measured crack initiation location vs. Predicted crack initiation location.
washer-plate interface. The analysis revealed that the contact between
the washer and the plates was fully sticking and the presence of high
peaks of contact pressure and frictional shear stress around the washer
sides.
A comparison of the numerical results was performed for different
pre-tension forces and at different stress amplitudes. The results were
extracted at the centreline of the specimen for the sake of conciseness.
11
Tribology International xx (xxxx) xxxx–xxxx
Fig. 19. Predicted crack locations vs. k constant for Fatemi-Socie and Crossland
parameters.
Fig. 15(a) shows the distribution of the contact pressure for the three
pre-tension levels at stress amplitude of 77 MPa. The contact pressure
over the plate surface is clearly larger in the case of higher preload force
resulting in a larger sticking region. The tangential stress distribution is
displayed in Fig. 15(c). Again the preload force produces greater
frictional tangential stress and the peak values are located in the
interface between the sticking and slipping region. On the contrary, the
shear frictional stress, depicted in Fig. 15(b), remains the same
independently of the preload force. In conclusion, the higher the
preload force the larger the sticking region, hence the maximal
frictional shear and tangential stress regions occur further away from
the edge of the hole.
4.2.2. Multiaxial criteria
Fig. 16 displays the (a) SWT, (b) FFD, (c) Crossland and (d) Fatemi-
Socie criteria computed at the plate-plate contact interface. The
C. Jiménez-Peña et al.
Fig. 20. Multiaxial parameter results for (a) SWT, (b) FFD, (c) Crossland, and (d) Fatemi-Socie, together with tested specimen at Fcl =97.75 kN at Δσ=77 MPa.
maximum value of these parameters indicates the crack initiation
location. All of the multiaxial criteria predict the crack initiation at a
similar location, i.e., at a certain distance from the hole edge, around
the centreline of the specimen. The same crack initiation region is
observed in the tested specimen, shown in Fig. 16. The behaviour of the
predicted initial crack location is similar for the four criteria, as higher
stress amplitudes produce initial crack locations closer to the edge of
the hole. In addition, the higher stress amplitudes result in larger
values of the criteria parameters, which indicates more damage caused
by fretting fatigue and lower fatigue life, as it has been observed
experimentally. In Fig. 17(a), the FFD parameter is represented for
different pre-tensions at a stress amplitude of 35 kN. The lower value of
the pre-tension load produces a higher FFD parameter, which is in line
with the results from the fatigue test, i.e., increasing pre-tension load
increases fatigue life. The SWT, the Crossland and the Fatemi-Socie
parameters are plotted in Fig. 17(b)–(d) for three pre-tension loads. In
this case, the value of these parameters increases with a higher pre-
tension load. This would indicate that higher pre-tension loads produce
lower fatigue life, which is not in agreement with the observed
experimental results.
A quantitative comparison between the actual location of the crack
initiation and the predicted values obtained from the multiaxial criteria
is shown in Fig. 18. The predicted crack initiation locations by the
Fatemi-Socie, the Crossland and the FFD criteria are close to the crack
locations observed experimentally. The values predicted by the SWT
criterion tend to overestimate the distance to the crack initiation. This
behaviour is shared by the other criteria at the highest preload level.
A sensitivity study was performed to investigate the influence of the
k constant on the predicted crack location by the Fatemi and the
Crossland multiaxial criteria. Values of k were varied between 0 and 1
while the predicted locations was monitored. The results of the
sensitivity study are presented in Fig. 19. The predicted location by
the Fatemi-Socie parameter slightly varies for low k values (0–0.1) but
remains unchanged for k values higher than 0.1. The predicted location
by the Crossland criterion remains unchanged for the whole range of k
values. Consequently it can be affirmed that the predicted crack
locations are reasonably insensitive to the k constant. This is however
not the case for the magnitude of the calculated parameters which
varies accordingly.
The multiaxial criteria were also computed for the washer-plate
interface. Fig. 20 displays the computed multiaxial parameters for a
clamping force of 300 N m and a stress amplitude of 77 MPa. The SWT
(Fig. 20(a)) and the Crossland (Fig. 20(c)) criteria predict the damage
at both edges of the washer contact, at the load side of the plate.
Specifically the SWT criterion is higher along the outer side of the
12
Tribology International xx (xxxx) xxxx–xxxx
washer while the Crossland criterion is higher along the inner side.
Both FFD and Fatemi-Socie predict the damage along the outer side of
the washer side. However, the calculated multiaxial parameters are
significantly lower than those calculated at the plate-plate interface.
This would indicate that fracture is more likely to happen at the plate-
plate contact region, which is in line with the experimental results from
the fatigue test.
5. Conclusions
In this study, a single-lap bolted joint (SLBJ) experimental set-up
was used to investigate the effect of the pre-tension load on the fretting
fatigue failure mechanism of bolted connections made of high strength
steel (HSS). The material used in the SLBJ was HSS 500MC. The
experimental results reveal that an increase in the pre-tension force
significantly improves the fatigue life of the connection. Fretting
damage was observed in all the tested specimens and fretting fatigue
failure was the dominant failure mechanism. A representative 3D-FEA
model was built to study four fretting fatigue criteria, in the contact
interface. The numerical and experimental results show that the pre-
tension affects the size of the sticking region and the location of the
stress concentration. In addition, several multiaxial criteria were
applied to predict the potential location of the initial crack. A minutious
comparison of the numerical results and the experimentally obtained
initial crack region enabled to assess the accuracy of the proposed
criteria. It is shown that only the Fretting Fatigue Damage parameter is
able to appropiately reflect the beneficial effect of the increased pre-
tension load on the fretting fatigue life.
References
[1] Y. Li, Z. Lin, A. Jiang, G. Chen, Use of high strength steel sheet for lightweight and
crashworthy car body, Mater Des 24 (2003) 177–182.
[2] Y. Zhang, X. Lai, P. Zhu, W. Wang, Lightweight design of automobile component
using high strength steel based on dent resistance, Mater Des 27 (2006) 64–68.
[3] Galambos TV, Hajjar JF, Earls CJ, Gross JL. Required properties of high-
performance steels. Building and Fire Research Laboratory. National Institute of
Standards and Technology, Gaithersburg, Maryland 20899; 1997.
[4] Schröter IF, Schütz D-IW. State of art in the production and use of high-strength
heavy plates for hydropower applications; 2005.
[5] Schröter F. Trends of using high-strength steel for heavy steel structures. In:
Giejowski MA, Kozowski A, lczka L, Zióko J, editors (Hrsg): Progress in Steel,
Composite and Aluminium Structures, S.292–293.
[6] Collin P, Johansson B. Bridges in high strength steel. In: Proceedings of IABSE
Symposium Report; 2006. 92. p. 1–9.
[7] M. Veljkovic, B. Johansson, Design of hybrid steel girders, J Constr Steel Res 60
(2004) 535–547.
[8] E. Gogou, Use of high strength steel grades for economical bridge design, TU Delft,
Delft University of Technology, Delft, 2012.
C. Jiménez-Peña et al.
[9] G.K. Ahiale, Y.-J. Oh, Microstructure and fatigue performance of butt-welded joints
in advanced high-strength steels, Mater Sci Eng: A 597 (2014) 342–348.
[10] G. Shi, F. Hu, Y. Shi, Recent research advances of high strength steel structures and
codification of design specification in China, Int J Steel Struct 14 (2014) 873–887.
[11] Afzali M, Lefebvre F, van Es SHJ, Kolstein MH, Pijpers RJM, Bijlaard FSK. Fatigue
design 2013. In: International Conference Proceedings TIG-dressing of High
Strength Steel Butt Welded Connections – Part 1: Weld Toe Geometry and Local
Hardness. Procedia Engineering. 66. ; 2013. p. 216–25.
[12] U. Kuhlmann, H.-P. Günther, C. Rasche, High-strength steel fillet welded connec-
tions, Steel Constr 1 (2008) 77–84.
[13] RFCS project FATWELDHSS, Improving the fatigue life of high strength steel
welded structures by post weld treatments and specific filler material; 2010.
[14] EN 1993-1-8. Eurocode 3: design of steel structures – Part 1–8: Design of joints.
Brussels: European Committee for Standardisation; 2005.
[15] DASt-Richtlinie/Deutscher Ausschuß für Stahlbau. Band 103: Nationales
Anwendungsdokument (NAD), Richtlinie zur Anwendung von DIN V ENV 1993,
Eurocode 3.
[16] EN 1993-1-12. Eurocode 3: design of steel structures – Part 1–12: additional rules
for the extension of EN 1993 up to steel grades S 700. Brussels: European
Committee for Standardisation; 2007.
[17] EN 1993-1-8. Eurocode 3: design of steel structures – Part 1–8: Fatigue. Brussels:
European Committee for Standardisation; 2005.
[18] R. Puthli, O. Fleischer, Investigations on bolted connections for high strength steel
members, J Constr Steel Res 57 (2001) 313–326.
[19] P. Može, D. Beg, Investigation of high strength steel connections with several bolts
in double shear, J Constr Steel Res 67 (2011) 333–347.
[20] P. Može, D. Beg, High strength steel tension splices with one or two bolts, J Constr
Steel Res 66 (2010) 1000–1010.
[21] Shi Y, Pan B, Shi G, Wang Y. Investigation on bolted connection of high strength
steel tension splices under static shear. In: Proceedings of the 6th international
symposium on steel structures, Korea; 2011. p. 1012–8.
[22] O.-P. Hämäläinen, T. Björk, Fretting fatigue phenomenon in bolted high-strength
steel plate connections, Steel Constr 8 (2015) 174–178.
[23] Maleki Hadi Rezghi, Abazadeh B. Fretting, Fatigue behavior of bolted single lap
joints of aluminum alloys, Int J Mech Aerosp Ind Mechatron Manuf Eng 6 (2012)
1509–1511.
[24] T.N. Chakherlou, M.J. Razavi, A.B. Aghdam, On the variation of clamping force in
bolted double lap joints subjected to longitudinal loading: a numerical and
experimental investigation, Strain 48 (2011) 21–29.
[25] F. Esmaeili, T.N. Chakherlou, M. Zehsaz, S. Hasanifard, Investigating the effect of
clamping force on the fatigue life of bolted plates using volumetric approach, J
Mech Sci Technol 27 (2013) 3657–3664.
[26] F. Esmaeili, T.N. Chakherlou, M. Zehsaz, Investigation of bolt clamping force on the
fatigue life of double lap simple bolted and hybrid (bolted/bonded) joints via
experimental and numerical analysis, Eng Fail Anal 45 (2014) 406–420.
[27] Valtinat G, Hadrych I, Huhn H. Strengthening of riveted and bolted steel
constructions under fatigue loading by preloaded fasteners-experimental and
theoretical investigations. In: Proceedings of the international conference on
connections in steel structures IV, AISC and ECCS, Roanoke, USA; 2000.
[28] J.M. Mínguez, J. Vogwell, Effect of torque tightening on the fatigue strength of
bolted joints, Eng Fail Anal 13 (2006) 1410–1421.
[29] A. Benhamena, A. Talha, N. Benseddiq, A. Amrouche, G. Mesmacque,
M. Benguediab, Effect of clamping force on fretting fatigue behaviour of bolted
assemblies: case of couple steel–aluminium, Mater Sci Eng: A 527 (2010)
6413–6421.
[30] R. Hojjati-Talemi, M.A. Wahab, P.D. Baets, Numerical investigation into the effect
of contact geometry on fretting fatigue crack propagation lifetime, Tribol Trans 55
(2012) 365–375.
[31] R. Hojjati-Talemi, M.A. Wahab, Fretting fatigue crack initiation lifetime predictor
tool: using damage mechanics approach, Tribol Int 60 (2013) 176–186.
[32] R. Hojjati-Talemi, M.A. Wahab, E. Giner, M. Sabsabi, Numerical estimation of
fretting fatigue lifetime using damage and fracture mechanics, Tribol Lett 52
(2013) 11–25.
[33] R. Hojjati-Talemi, M. Abdel Wahab, J. De Pauw, P. De Baets, Prediction of fretting
fatigue crack initiation and propagation lifetime for cylindrical contact configura-
13
Tribology International xx (xxxx) xxxx–xxxx
tion, Tribol Int 76 (2014) 73–91.
[34] F. Esmaeili, T.N. Chakherlou, M. Zehsaz, Prediction of fatigue life in aircraft double
lap bolted joints using several multiaxial fatigue criteria, Mater Des 59 (2014)
430–438.
[35] EN 14399-1:2005. High-strength structural bolting assemblies for preloading –
Part 4: System HV – Hexagon bolt and nut assemblies. Brussels: European
Committee for Standardisation; 2006.
[36] EN 14399-1:2005. High-strength structural bolting assemblies for preloading –
Part 6: Plain chamfered washers. Brussels: European Committee for
Standardisation; 2006.
[37] Denys K, Coppieters S, Seefeldt M, Debruyne D. Identification of plastic material
behaviour of thick high strength steels through FEMU techniques; 2015.
[38] Amstrong – High strength low alloy steels, Arcelormittal Catalogue. Available at
〈http://industry.arcelormittal.com/catalogue/A20/EN〉. [accessed June 2016].
[39] D.J. Thomas, Characteristics of abrasive waterjet cut-edges and the affect on
formability and fatigue performance of high strength steels, J Manuf Process 11
(2009) 97–105.
[40] EN 1090-2:2008. Execution of steel structures and aluminium structures – Part 2:
Technical requirements for steel structures, London: BSI; 2008.
[41] A. Ferjaoui, T. Yue, M. Abdel Wahab, R. Hojjati-Talemi, Prediction of fretting
fatigue crack initiation in double lap bolted joint using continuum damage
mechanics, Int J Fatigue 73 (2015) 66–76.
[42] I.R. McColl, J. Ding, S.B. Leen, Finite element simulation and experimental
validation of fretting wear, Wear 256 (2004) 1114–1127.
[43] J. De Pauw, W. De Waele, R. Hojjati-Talemi, P. De Baets, On the use of digital
image correlation for slip measurement during coupon scale fretting fatigue
experiments, Int J Solids Struct 51 (2014) 3058–3066.
[44] D.A. Hills, Mechanics of fretting fatigue, Wear 175 (1994) 107–113.
[45] K. Nishioka, K. Hirakawa, Fundamental investigations of fretting fatigue: Part 3,
some phenomena and mechanisms of surface cracks, Bull JSME 12 (1969)
397–407.
[46] K. Endo, H. Goto, T. Fukunaga, Behaviors of frictional force in fretting fatigue, Bull
JSME 17 (1974) 647–654.
[47] M.A. Sutton, J.J. Orteu, H. Schreier, Image correlation for shape, motion and
deformation measurements: basic concepts, theory and applications, Springer
Science & Business Media, New York, NY 10013, USA, 2009.
[48] MatchID software. 〈http://www.matchidmbc.be〉.
[49] ABAQUS. ABAQUS documentation, Dassault Systèmes, Providence, RI, USA; 2011.
[50] J. Domínguez, Cyclic variations in friction forces and contact stresses during
fretting fatigue, Wear 218 (1998) 43–53.
[51] M. Nesládek, M. Španiel, J. Jurenka, J. Růžička, J. Kuželka, Fretting fatigue –
experimental and numerical approaches, Int J Fatigue 44 (2012) 61–73.
[52] C. Navarro, S. Muñoz, J. Domínguez, On the use of multiaxial fatigue criteria for
fretting fatigue life assessment, Int J Fatigue 30 (2008) 32–44.
[53] W.S. Sum, E.J. Williams, S.B. Leen, Finite element, critical-plane, fatigue life
prediction of simple and complex contact configurations, Int J Fatigue 27 (2005)
403–416.
[54] M.R. Hirsch, R.W. Neu, Fretting damage in thin sheets: analysis of an experimental
configuration, Tribol Int 44 (2011) 1503–1510.
[55] K. Smith, T. Topper, P. Watson, A stress-strain function for the fatigue of metals
(stress-strain function for metal fatigue including mean stress effect), J Mater 5
(1970) 767–778.
[56] D. Socie, Multiaxial fatigue damage models, J Eng Mater Technol 109 (1987)
293–298.
[57] A. Fatemi, D.F. Socie, A critical plane approach to multiaxial fatigue damage
including out‐of‐phase loading, Fatigue Fract Eng Mater Struct 11 (1988) 149–165.
[58] Crossland B. Effect of large hydrostatic pressures on the torsional fatigue strength
of an alloy steel. In: Proceedings of International Conference on Fatigue of Metals:
Institution of Mechanical Engineers London; 1956.
[59] C. Ruiz, P.H.B. Boddington, K.C. Chen, An investigation of fatigue and fretting in a
dovetail joint, Exp Mech 24 (1984) 208–217.
[60] R.H. Talemi, M.A. Wahab, Finite element analysis of localized plasticity in Al 2024-
T3 subjected to fretting fatigue, Tribol Trans 55 (2012) 805–814.
[61] H.S. Kim, S. Mall, Investigation into three-dimensional effects of finite contact
width on fretting fatigue, Finite Elem Anal Des 41 (2005) 1140–1159.