International Journal of Adhesion & Adhesives 107 (2021) 102845

Contents lists available at ScienceDirect

International Journal of Adhesion and Adhesives

journal homepage: http://www.elsevier.com/locate/ijadhadh

The effects of adherend thickness on the fatigue strength of adhesively

bonded single-lap joints
Resul Sahin, Salih Akpinar *
Dept. of Mechanical Eng., Erzurum Technical University, 25050 Erzurum, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: Different adhesive bonding methods have been prominently used in aerospace applications since the 1980s
Adhesive bonding because adhesively bonded joints are regarded as an alternative to traditional bonding methods. However, the
Joint design effects of fatigue on adhesively bonded joints – one of the most significant issues in the aerospace industry – are
Fracture failure
still currently being investigated. This study experimentally investigated the lifetimes of single-lap joints (SLJs)
Fatigue testing
Static force
of five different adherend thicknesses (2, 3, 4, 5 and 6 mm) – obtained using AA2024-T3 aluminum alloy and
Strength DP460 structural adhesive – under varying tensile fatigue loads. Accordingly, before performing fatigue tests,
static tensile strength tests were carried out on the SLJs to determine the maximum loads to be used for tensile
fatigue tests and also to evaluate their quasi-static performances. The fatigue tests were performed at a loading
ratio (R) of 0.1 and a frequency of 20 Hz. The static tensile strength of the joints was found to increase by
approximately 22% with increasing adherend thickness. However, this increase was not the same as the increase
in adherend thickness. In addition, increased adherend thickness was observed to change the maximum load
applied to the joints over infinite cycles (taken to be 106 cycles). This was because the change in adherend
thickness changed the bending moment formed in the joint and the flexural rigidity of the adherend.

1. Introduction literature [2–5].

Karachalios et al. [2] investigated the mechanical properties of ad­
Adhesive materials have been used in many areas and applications in hesive thickness, adherend thickness, and overlap length in SLJs under
recent years due to a wealth of superior properties; accordingly, exten­ bending loads. It was observed that overlap length had no effect on the
sive research has been done on them. Adhesives are generally used when bending behavior, while adherend thickness had a strong effect. As the
two materials cannot be bonded by traditional bonding methods (such load transfer occurred locally at the ends of the overlap region in SLJs, it
as riveting and screwing) or in order to bond two different adherend was observed that the damage was formed by the peeling mechanism.
materials. Although adhesively bonded joints have quite good shear Gültekin et al. [3] conducted a two-part study to investigate the ef­
properties, their performance is very weak against peel stresses. In fect of adherend thickness experimentally and numerically on the me­
adhesively bonded joints peel stresses form at the edges of the overlap chanical properties of a joint in adhesively bonded SLJs. In the first part,
area due to eccentric loading; thereby causing damage. Different five different thickness values were used, and the lower and upper
methods, e.g., scarf lap joint types, use of composite patches in the joint, adherend thicknesses were the same. In the second part, the lower
use of curvature patches in the joint, changing the lap joint geometry adherend thickness was constant and five different thickness values
and use of adhesive fillet in the joint adhesive fillet exist to reduce these were used for the upper adherend. The results indicated that the failure
peel stresses; and many studies have been conducted on this subject. In loads of the joint changed depending on the thickness of the adherend,
conclusion, reducing the effect of peel stresses is critical for adhesive and the numerical stress analyses were consistent with the experimental
bonding design [1]. study.
In adhesively bonded joints, the adherend thickness has a significant Zaroug et al. [4], experimentally and numerically investigated the
effect on the failure load of a joint. The effect of the adherend thickness mechanical properties of the joints bonded with bolts, adhesives and
on the failure load of joints has been statically examined in the published mixed methods using both the same and different adherends. They used

* Corresponding author.
E-mail address: [email protected] (S. Akpinar).

https://doi.org/10.1016/j.ijadhadh.2021.102845

Available online 13 March 2021

0143-7496/© 2021 Elsevier Ltd. All rights reserved.
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Table 1 study, as the average stress increased, the fatigue life estimation of the
The material properties of the adherend and adhesive used in this study [12]. model was more successful, which was compatible with the experi­
AA2024-T3 DP 460 mental results. It is also stated that the damage parameter is a function of
four fatigue variables such as e maximum principal strain, fatigue cycles,
E (MPa) 72400 ± 530 2077 ± 47
ν 0.33 0.38 fatigue threshold strain and the load ratio.
σt (MPa) 482 ± 12 44.6 ± 1.2 It was seen, based on the literature review, that adherend thickness
εt (mm/mm) 0.1587 0.0428 significantly affects the mechanical properties of adhesively bonded
E: Young’s modulus; ν: Poisson’s ratio; σt: Ultimate tensile strength; εt: Ultimate joints. Considering the operating conditions of the adhesively bonded
tensile strain. joints, they are also subject to dynamic loads rather than static loads.
Therefore, the present study investigated the effect of bending moments
and flexural rigidity on the fatigue load formed due to the eccentric
Table 2 loading of SLJs. In the study, adhesively bonded single-lap joints were
The experimental parameters for adhesively-bonded single lap joints. produced using DP460 liquid structural epoxy as the adhesive and
Joint Type Adherend Thickness-t (mm) Test Type AA2024-T3 aluminum alloy as the adherend. Fatigue tests were per­
formed at a loading ratio (R) of 0.1 and a frequency of 20 Hz. Fatigue
SLJ-S-2 2 Tensile-Static
SLJ-S-3 3 Tensile-Static tests were applied at seven different static tensile loads and the resulting
SLJ-S-4 4 Tensile-Static P–N curves and failure modes of the joints were examined. These failure
SLJ-S-5 5 Tensile-Static modes were discussed based on observation of failure location and na­
SLJ-S-6 6 Tensile-Static
ture of surfaces.
SLJ-F-2 2 Tensile-Fatigue
SLJ-F-3 3 Tensile-Fatigue
SLJ-F-4 4 Tensile-Fatigue 2. Experimental program
SLJ-F-5 5 Tensile-Fatigue
SLJ-F-6 6 Tensile-Fatigue 2.1. Material and specimens

In this study, AA2024-T3 aluminum alloy was used as the adherend,

and DP460 structural adhesive with two components (epoxy and hard­
ener) and produced by the 3 M company (St. Paul, MN, USA) was used as
the adhesive. Mechanical properties of adhesive (DP460) and the
adherend given in Table 1 were taken from a study carried out by Demir
[12].
This study investigated the static tensile and fatigue behavior of the
adhesively bonded joints with five different adherend thicknesses. A
Fig. 1. Singe lap joint geometry used in the experimental investigation.
change of adhesive thickness, overlap length and width of the adhesively
bonded joints changed its mechanical properties. Therefore, in order to
two different adherends (AL7075 and AL6061) and three different
accurately compare the mechanical properties of the joint types with
thicknesses (2, 4 and 6 mm). When the adherend thickness was 2 mm
five different adherend thicknesses used in this study, the adhesive
and the bonding type was mixed (bolt and adhesive), the failure in the
thickness, length and width of both joint types were kept the same. The
joint was observed to occur in the adherend material. They reported that
experimental and geometric parameters of these specimens are given in
when the thickness increased, the failure occurred completely in the
Table 2 and Fig. 1.
bolt. They also indicated that the increase in the adherend thickness
For the static tensile fatigue tests, the adherend specimens were cut
increased the failure load of the joint and decreased the amount of
using a metal saw from aluminum plates of 100 mm × 25 mm at five
displacement.
different thicknesses (Fig. 1). Three aluminum specimens were used for
As in the case of all traditional joints (such as bolts, rivets, welding,
each static test parameter, and five specimens were used for each
and solder), adhesively bonded joints are frequently subjected to dy­
loading ratio in the fatigue tests; Fig. 1 shows the dimensions of all
namic loads rather than just static loads due to environmental in­
aluminum specimens used. Then, surface preparation treatment was
fluences. These joints are subject to regular or irregular cyclic loads;
applied to the adherends because the performance of adhesively bonded
thus, it is very important to investigate their fatigue behaviors. When
joints depends on surface preparation treatments. For this, 240 grit sil­
published literature is reviewed, it was observed that the fatigue per­
icon carbide sandpaper was applied to clean the burrs, oil, grease, and
formance of the adhesive bonded joints was examined [6–11].
dirt that generated on the glued adherends due to cutting them at the
Khoramishad et al. [6] conducted an experimental and numerical
desired dimensions. Then, the scratch lines formed on the adherend due
study on the effect of the loading ratio (R) applied to SLJs subjected to
to this sanding process were removed by sanding again with 1000 grit
dynamic loading. They used a loading ratio of 0.1 and 0.5. They reported
silicon carbide sandpaper. After the sanding process was completed, the
that when the R value increased, the fatigue performance of the joint
specimens were washed with distilled water, kept in acetone for about
increased as adherend material hardened (i.e. the stiffness of the
15 min, and dried in an oven at 60 ◦ C for 30 min.
adherend increased). In addition, they indicated that the numerical
A special bonding mold, shown in Fig. 2a, was used to produce the
model used in the study was in parallel with the experimental results and
SLJs. The desired adhesive thickness, overlap length and joint thickness
the initiation and propagation of the failure was accurately predicted.
was achieved with the mold and some auxiliary apparatus. While pro­
The fatigue life of adhesively bonded joints mainly depends on the
ducing the joints, the adherend – with surface treatment – was placed on
fatigue load and the loading ratio. Katnam et al. [7] used a fatigue
the mold surface, and then small metal pieces (0.18 mm thick hard steel)
failure model to obtain the failure behavior and fatigue life estimations
at the desired adhesive thickness were placed onto the free surface of the
of joints experimentally and numerically. Their experimental results
adherend approximately 20 mm from the overlap region. After this
indicated that the fatigue strength increased with an increased loading
procedure, other metal pieces (aluminum pieces with adherend thick­
ratio. According to the failure model created within the scope of that
ness) at the desired adherend thickness were combined and placed in

2
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Fig. 2. a) The mold used in producing joints b) Joint production scheme.

Fig. 3. a) Hot press used to produce joints b) The joint specimens after curing.

order to adjust the overlap length (Fig. 2b). Rigid metal blocks were different load levels based on the average static strength, and the
placed on the adhesive joints to apply uniform pressure during curing specimens with a fatigue life higher than 106 cycles were considered as
(Fig. 2b). the run-out. For the P–N curves, a total of 180 specimens (7 or 8 load
A high precision hot press device was used to meet the curing con­ levels × 5 specimens × 5 joint types) were produced by taking five of the
ditions of the DP460 structural adhesive, and curing was carried out at specimens produced for each load level. The number of cycles was
70 ◦ C for 120 min under a pressure of 0.10 MPa (Fig. 3a) [13]. Surface recorded for the joint fatigue tests. After the tests, the failure surfaces of
preparation and curing conditions are the same for all joint samples. the specimens were examined.
After hardening, the adhesive protruding from the overlap area of the
joint was removed from the surface of the joint with a cutting tool. After
2.2. Test procedure
the bonding process, the adhesive layer thicknesses of single-lap joint
samples were measured with precision digital calipers and it was found
The static tensile tests were conducted using an Instron 5982 (USA)
that the adhesive thickness was 0.18 ± 0.01 mm (Fig. 3b). Then,
universal testing machine with a load cell of 100 kN at 21 ◦ C, 22%
depending on the thickness of the adherend, metallic patches of 25 * 25
humidity and 1 mm/min pulling speed. As shown in Fig. 4, the boundary
mm were affixed to both ends of the joint for axially place the sample in
conditions and the applied force were the same for each type of spec­
the jaws of the test device.
imen (SLJ).
Three specimens of each joint type (fifteen in total) were produced to
The fatigue tests were performed with a load cell of 25 kN at a 20 Hz
determine static tensile failure load of each joint. In addition, to form the
frequency using a computer controlled Instron 8874 (USA) universal
P–N curve (a plot of the magnitude of an alternating stress versus the
fatigue device shown in Fig. 4. The fatigue test procedures recom­
number of cycles to failure), each joint type was tested at seven or eight
mended by the ASTM D3166 standard and used by some researchers

3
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Fig. 4. Boundary conditions and loading used in the experiments; a) static tensile test, b) fatigue tensile test.

Fig. 5. Numerical analyses model of joint; a) mesh density, b) boundary conditions.

were obtained by applying tensile fatigue loads at certain ratios of the

Table 3
tensile failure load of these joints.
Properties of the DP460 adhesive for CZM modeling [18,19].
σmax
n (MPa) GIC (N/mm) τmax
t (MPa) GIIC (N/mm) 3. Numerical program
32.6 2.56 28.5 11.71
Numerical analysis was done by using ANSYS 18 package software
[17] and the joints used in the study were modeled in 3D. Numerical
were used in the study [14–16]. Accordingly, a 20 Hz sinusoidal analysis in joints was carried out using a nonlinear finite element
waveform was used for all the fatigue tests performed at a loading ratio method, taking into account the nonlinear material behavior of the
(R) of 0.1 and at a constant load amplitude. To form the P–N curve, each adherend (AA2024-T3). One fundamental factor that affects the me­
joint type was tested at seven or eight different load levels based on the chanical strength of adhesively bonded joints is the peel stresses (cause
average static strength, and 106 cycles was determined to be the fatigue damage to the joint) that form at free edges of the overlap region.
lifetime. Another factor is the shear stresses that occur in the middle of the
In this present study, firstly, the static tensile loads of the SLJs of overlap region. For this reason, adhesive behavior is modeled with
different adherend thicknesses were determined, and then S–N plots cohesive zone model (CZM), which is the most suitable method for

4
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

critical point (ideal mesh density), it has been found that multiple mesh
uses do not have any effect for stress analysis. Therefore, to reduce the
time required for numerical solutions, smaller meshs were used in areas
critical for stress in the joint (Fig. 5).
Some parameters of DP460 adhesive must be known in order to
analyze with CZM. These values Kanar et al. [18] and Kazaz et al. [19]
were obtained by performing the Mode I and Mode II experiments and
these values were used in this study. GIC (the tensile critical strain energy
release rate) values were obtained with Double Cantilever Beam joints,
while GIIC (the shear critical strain energy release rate) values were
obtained with Thick Adherend Shear Test joints. These values were used
in this study (Table 3).

4. Results and discussion

4.1. Tensile static test results for the joints

Three SLJ specimens with five different adherend thicknesses (i.e. 2,

Fig. 6. Average failure loads obtained for the single lap joints subject to static 3, 4, 5, and 6 mm) were tested, and the average failure loads obtained
tensile load. from the tensile tests are shown in Fig. 6, in which the failure loads in
Fig. 6, the increase in the adherend thickness increased the failure load
of the joint. The experimental data obtained from the comparison of all
joints with the 2 mm-thick joint (SLJ-S-2) indicated that a 3-mm
thickness (SLJ-S-3) increased the failure load of the joint by approxi­
mately 6%, a 4-mm thickness (SLJ-S- 4) by 15%, a 5-mm thickness (SLJ-
S-5) by 22%, and a 6-mm thickness (SLJ-S-6) by 10% (Fig. 6).
When the load-displacement curves of the five different SLJ types
given in Fig. 7 are examined, the change of the adherend thickness in the
joints with the same overlap length significantly affects both the failure
load and the displacement capacity of the joint. When the thickness of
the adherend is increased from 2 mm to 6 mm, the displacement of the
joint decreases, while its failure load increases. Moreover, in all the
joints subject to static tensile load, although it shows non-linear
behavior up to the first 500 N load, it shows a linear behavior after
500 N until the failure load (Fig. 7).
However, to explain the experimental results, it is necessary to know
the damage mechanism of the adhesively bonded SLJs. When adhesively
bonded single-lap joints were subjected to tensile loading, the lines of
force in the adherends did not overlap with the adhesive line and there
was a distance “y” between two lines (Fig. 8). This “y” distance formed a
bending moment (M = F × y) in the joint. The bending moment value
Fig. 7. Relationship between failure load and displacement for SSLsamples varied depending on the thickness of the adherend material: it increased
with different adherend thicknesses. with increasing distance “y” between the lines of force in both adherend
materials. This negatively affected the load carrying capacity of the
joint. On the other hand, the increase in the adherend material thickness
increased the flexural rigidity of the joint (modulus of elasticity ×
moment of inertia or EI) depending on the moment of inertia which in
turn increased the strength of the joint against the bending moment
formed. The effect of this moment that occurred due to eccentric loading
caused fractures at the edges of the overlap region of the joint, and these
fractures suddenly propagated towards the center of the overlap region,
leading to damage (Fig. 8) [12]. In a study performed by Demir et al.
[12], support patches were added at a varying distances from the
overlap region to decrease the bending effect occurring in adhesively
Fig. 8. Moment due to eccentric loading in SLJs subject to tensile loading [12].
bonded single-lap joints. According to the results of this study, when the
failure load obtained from the experiments was examined, the use of
support patches in single-lap joints increased the failure load of the joint
bonding joints.
between approximately 16% and 94%.
The dimensions, loading and boundary conditions of the joints used
According to the damage mechanism given above for SLJs, the dis­
in numerical analysis are the same as those used in the experimental
tance “y” was smaller in the joint with a thickness of 2 mm than in the
study. In numerical analysis, the joint is modeled using a Solid 186
other joints. This decreased the bending moment caused by eccentric
element type with 20 nodes and three degrees of freedom. Many mesh
loading. On the other hand, low adherend thickness (2 mm) decreased
densities have been tried while designing this connection model. After a
the flexural stiffness of the joint (EI) and increased the transverse

5
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Fig. 9. Surface failures for joint types subjected to static tensile loads.

Table 4
Numerical analysis and experimental average failure loads and displacements of a SLJ.
TYPE Average Experimental Failure Numerical Analysis Failure FNA/ Average Experimental Displacement Numerical Analysis Displacement DNA/
Load FEXP (N) Load FNA (N) FEXP DEXP (mm) DNA (mm) DEXP

SLJ-S- 8283±143 8100 0.98 0.434±0.02 0.415 0.96

2
SLJ-S- 8814 ±202
8650 0.98 0.407 ±0.02
0.393 0.97
3
SLJ-S- 9556 ±162
9835 1.03 0.374 ±0.01
0.365 0.98
4
SLJ-S- 10139±186 10620 1.05 0.375±0.02 0.362 0.96
5
SLJ-S- 9142±195 9340 1.02 0.298±0.01 0.284 0.95
6

numerical study performed by Gültekin et al. [3], supports this result.

It is very important to determine the type of failure to completely
understand the mechanical properties of adhesively bonded joints. The
types of failure in adhesively bonded SLJs given in Fig. 9 were deter­
mined according to the failure types defined in ISO 10365 [20]. Fig. 8
shows that special cohesive failure (the case where the adhesive layer
thickness is not the same in the upper and lower adherend) occurred in
the all joint types (SLJ-S-2, SLJ-S-3,SLJ-S-4, SLJ-S-5 and SLJ-S-6).
Fig. 10. Critical stress line of the joint samples. In the numerical analysis (NA) of the joints, one end of the joint was
held constant and the other end was given some displacement in the y
deflection in the joint during loading. This had a significant effect on the direction. In the FEA for SLJ-S-2, SLJ-S-3, SLJ-S-4, SLJ-S-5 and SLJ-S-6,
failure to the joint. When the adherend thickness was increased from 2 the applied displacement in the y direction (tensile direction) was
mm to 3, 4 and 5 mm, the transverse deflection decreased due to the applied in 0,025 mm displacement steps. During displacement, the
increase in joint stiffness, which in turn increased the failure load of the peeling stress and energy (σ max
n , GIC) causing damage occur in the z-di­
joint (Table 3). However, an increase in thickness increased the bending rection at the edges of the overlap region of the joint. The results ob­
moment depending on “y”, so the increased rate of flexural rigidity and tained from numerical analysis and experimental studies are shown in
the increased bending moment of the joint balanced each other out, Table 4. In order to predict the numerical analysis failure load given in
increasing the joint’s load-carrying capacity. An experimental and Table 4, the parameters (σ max max
n , GIC, τt , GIIC) of the adhesive used in

Fig. 11. Comparison of critical lines obtained from joint types, a) peel stress (σz), b) shear stress (τyz).

6
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Table 5
Relationship between rigidity, moment, load and number of cycles for which the
joints were subjected to fatigue loading.
Joint Flexural Bending Force- Average Applied
Type rigidity of Moment M N number of fatigue
adherend-EI = y.F cycles load/static
(kN⋅mm2) (kN⋅mm) strength

SLJ-F- 1206 2.18F 3000 73214 0.36

2a
SLJ-F- 1206 2.18F 2500 116125 0.30
2b
SLJ-F- 1206 2.18F 2000 195654 0.24
2c
SLJ-F- 1206 2.18F 1750 347174 0.21
2d
SLJ-F- 1206 2.18F 1500 656727 0.18
2e
SLJ-F- 1206 2.18F 1375 897643 0.17
2f
SLJ-F- 1206 2.18F 1250 1000000 0.15
2g
SLJ-F- 4072 3.18F 3000 45515 0.34 Fig. 12. Comparison of P–N curves of joints with different adherend
3a thicknesses.
SLJ-F- 4072 3.18F 2500 84812 0.28
3b
CZM given in Table 3 was used, with the assumption that the failure
SLJ-F- 4072 3.18F 2000 134416 0.23
3c occurred when the parameters calculated at any point of the adhesive
SLJ-F- 4072 3.18F 1750 271567 0.20 layer reached the parameters (σ max max
n , GIC, τt , GIIC) of the adhesive.
3d When the results obtained from the experiments and the results obtained
SLJ-F- 4072 3.18F 1500 456377 0.17
from numerical analysis were compared, it was seen that they were quite
3e
SLJ-F- 4072 3.18F 1375 527063 0.16 compatible with each other. This phenomenon shows the accuracy of the
3f use of CZM in numerical analysis of joints. Therefore, it is important to
SLJ-F- 4072 3.18F 1250 787441 0.14 obtain displacement data with video extensometer in terms of compat­
3g ibility of experimental and numerical analysis.
SLJ-F- 4072 3.18F 1175 1000000 0.13
Considering the stresses formed in the adhesive layer, it is extremely
3h
SLJ-F- 9653 4.18F 3000 61017 0.31 important to find the critical line where the stresses will be taken in
4a order to analyze the stresses occurring in the adhesive layer more
SLJ-F- 9653 4.18F 2500 146124 0.26 accurately. The results of the numerical analysis and experimental study
4b
show that critical loci for failure are the interface between adhesive
SLJ-F- 9653 4.18F 2000 221128 0.20
4c
layer and upper adherend and the interface between adhesive layer and
SLJ-F- 9653 4.18F 1750 367170 0.18 lower adherend, and it was observed that the AB line shown in Fig. 10 is
4d critical in terms of stresses. Stresses were taken over this line.
SLJ-F- 9653 4.18F 1625 523854 0.17 For the five configurations of the SLJ joint type (SLJ-S-2, SLJ-S-3,
4e
SLJ-S-4, SLJ-S-5 and SLJ-S-6), stress distributions along the midline of
SLJ-F- 9653 4.18F 1500 871953 0.16
4f the adhesive layer (line A-B) were studied under a tensile load of 4000 N
SLJ-F- 9653 4.18F 1400 1000000 0.15 (approximately half of the minimum failure load obtained from the
4g experiments conducted on five joint samples). The comparative graphs
SLJ-F- 18854 5.18F 3000 33572 0.30
of stress distributions peel and shear (σz, τyz), which are more effective
5a
SLJ-F- 18854 5.18F 2500 72306 0.25
than other stresses on failure, are given in Fig. 11.
5b When the peel stress (σy) distributions given in Fig. 11a are exam­
SLJ-F- 18854 5.18F 2000 220997 0.20 ined, the stresses occurring at both ends of the overlap area (points A
5c and B) reduce significantly when the thickness of the bonded material is
SLJ-F- 18854 5.18F 1750 546435 0.17
increased from 2 mm to 3 mm, 4 mm and 5 mm, However, the peel
5d
SLJ-F- 18854 5.18F 1675 752478 0.16 stresses increase significantly when the thickness of the bonded material
5e is increased from 5 mm to 6 mm (Fig. 11a). Considering the relationship
SLJ-F- 18854 5.18F 1600 911374 0.16 between reducing the stresses on the adhesive layer and increasing the
5f
failure load of the joint, it is understood that the stress distributions
SLJ-F- 18854 5.18F 1550 1000000 0.15
5g
obtained from the numerical analysis are compatible with the experi­
SLJ-F- 32580 6.18F 3000 118736 0.33 mental failure loads.
6a In SLJ joints subjected to tensile load, peeling stresses in the middle
SLJ-F- 32580 6.18F 2500 164553 0.27 parts of the overlap region (between about 1.5 mm and 11 mm) are
6b
compressive character. However, in joints with a adherend thickness of
SLJ-F- 32580 6.18F 2250 214862 0.25
6c 2 mm, the value of these compressive character stresses decreases in the
SLJ-F- 32580 6.18F 2000 404381 0.22 middle parts of the overlap zone, and a uniform stress distribution does
6d not occur. As the adherend thickness increases, these compressive
SLJ-F- 32580 6.18F 1900 648614 0.21 character stresses that occur in the middle parts of the overlap region
6e
SLJ-F- 32580 6.18F 1800 832672 0.20
turn into a uniform stress distribution (Fig. 11a).
6f In addition, when examining the shear stress (τyz) distributions given
SLJ-F- 32580 6.18F 1750 1000000 0.19 in Fig. 11b, the increase in thickness of the bonded material decreased
6g

7
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Fig. 13. Failure surfaces of adhesively bonded joints subjected to tensile fatigue loading.

the absolute value of the maximum shear stress at points A and B. effect varied by the adherend thickness. The experimental data pre­
sented in Table 3 should be interpreted according to the damage
mechanism given above for SLJs and the observations made during tests.
4.2. Mechanical properties of adhesively bonded SLJs subjected to tensile The experimental data provided in Table 5 indicated that when
fatigue loading approximately 30% of the static failure load (3000 N) was applied to the
adhesively bonded joints, the number of cycles decreased by 61% as the
In this section, in the light of the experimental results obtained in adherend thickness was increased from 2 to 3 mm. However, increasing
section 4.1, the effect of eccentric loading on fatigue of the adhesive the adherend thickness from 3 to 4 mm increased the number of cycles
bonded SLJ types subjected to tensile loading was experimentally by 34%, while increasing it from 4 to 5 mm decreased the number of
investigated. cycles by 45%. When the adherend thickness was increased from 5 to 6
A different case was seen in the observations made during tensile mm, the number of cycles increased by 254%. This inconsistency, which
fatigue tests: the 20 cycles per second (20 Hz) applied during tensile is related to the adherend thickness, was attributed to the relationship
fatigue tests on the SLJs created a lateral vibration effect in the joint due between the bending moment and flexural rigidity in the joint, as well as
to the bending moment resulting from eccentric loading. This vibration

8
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

the larger lateral vibration effect in the joint caused by the large fatigue lower or upper adherend). According to the previous studies [21], this is
load applied (approximately 30% of the static failure load). related to the crack growth phenomenon. In all joint types, the adhesion
Comparison of the loads applied for 106 cycles – considered to be failure occurred in cycles close to the fatigue strength limit (an infinite
infinite life – showed that when the adherend thickness was increased number of cycles: 106). After the tensile fatigue tests, no plastic defor­
from 2 mm (SLJ-F-2) to 3 mm (SLJ-F-3), the applied load decreased by mation occurred in the adherends. Because the loads applied in fatigue
6% (Table 5). When the adherend thickness was increased from 2 to 4 tests are the loads below the yield strength of the adherends.
mm (SLJ-F-4), the applied load increased by 12.5%, and when it was
increased to 5 mm (SLJ-F-5), the load increased by 24%. For a thickness 5. Conclusions
of 6 mm (SLJ-F-6), the load was increased by 40%. This result can be
explained by the bending moment in the joint and the flexural rigidity of This study experimentally examined the static and fatigue strengths
the adherend. Based on an adherend thickness of 2 mm, the flexural of single-lap joints (SLJs) with five different thicknesses subjected to
rigidity and bending moment values of the joints were examined, which tensile test. Accordingly, the following conclusions were made:
indicated that for an adherend thickness of 3 mm, the flexural rigidity
increased by 237% and the bending moment by 46%; for a thickness of 4 • In the adhesively bonded SLJs subjected to tensile loading, the fail­
mm, the flexural rigidity increased by 700% and the bending moment by ure load of the joint increased by approximately 6%–22% with
92%; for thickness of 5 mm, the flexural rigidity increased by 1463% and increased adherend thickness. According to the failure mechanism of
the bending moment by 138%, and for an adherend thickness of 6 mm, SLJs, the increase in the adherend thickness increased the bending
the flexural rigidity increased by 2601% and the bending moment by moment depending on the “y” value (the distance between the two
183%. Comparison of the joints with 2 mm and 3 mm thicknesses lines of force), and this moment had a negative effect in terms of joint
showed that the effect of the bending moment was more pronounced failure. However, as the increase in the adherend thickness increased
than the flexural rigidity in the joint. However, the increase in the the flexural rigidity of the joints, it decreased the transverse deflec­
flexural rigidity of the adherend was higher than the increase in the tion (due to the moment) during loading. This had a positive effect
bending moment, which increased the fatigue performance of the joint. on the failure load.
Also, all data of the five specimens obtained for each load level are given • Special cohesion failure (the case where the adhesive layer thickness
in the appendix. is not the same in the upper and lower adherend) occurred in joint
Two important conclusions can be made from the P–N curves of the types with an adherend thickness of 2, 3, 4, 5 and 6 mm.
joints given in Fig. 12: The first was that the P–N curves of the adhesively • When 30% of the static failure load was applied to the adhesively
bonded joints were nonlinear, which is attributed to the high deform­ bonded SLJs for fatigue loading, increasing the adherend thickness
ability of the adhesive in adhesively bonded joints. When the materials from 2 mm to 6 mm changed (either increased or decreased) the
with low deformability (i.e. bulk metallic materials such as steel) are number of cycles for the joint. This inconsistency, which occurred
subject to fatigue loading, their P–N curves generally become linear. due to the adherend thickness, was attributed to the relationship
However, when the adhesive joints with high deformability (i.e. the between the bending moment in the joint and the flexural rigidity, as
composite joint obtained by combining the adhesive with aluminum) well as the higher lateral vibration effect on the joint caused by
are subjected to fatigue loading, the P–N curves generally become applying higher fatigue loads.
nonlinear. The nonlinear P–N curves obtained in this study prove the • In addition, comparison of the loads applied for 106 cycles indicated
accuracy of the data obtained from the experiments. that when the adherend thickness was increased from 2 mm to 3 mm,
The second important result was that the effect of the bending applied fatigue load decreased by approximately 6%. On the other
moment formed due to eccentric loading in SLJs was greater as a result hand, when the adherend thickness was increased to 4 mm, applied
of the increase in the maximum applied load. In the observations made load increased by 12%; when it was increased to 5 mm, applied load
during the experiments, when a maximum tensile fatigue load of 3000 N increased by 24%; and when it was increased to 6 mm, applied load
(fatigue load ratio R = 0.1) was applied to the joint, there was a lateral increased by 40%. This result can be explained by the bending
vibration due to eccentric loading. However, this vibration was observed moment in the joint and the flexural rigidity of the adherend. The
to decrease when the applied load level was decreased. In addition, rate of increase in the flexural rigidity due to increased adherend
increasing the adherend thickness increased the flexural rigidity of the thickness was higher than the rate of increase in the bending
adherend, which increased the ability to absorb the bending moment in moment, which led to an increase in the fatigue performance of the
the joint. As can be seen in Fig. 12, the increase in the adherend thick­ joint.
ness significantly increased the fatigue strength limit of the joint. • During the tests of SLJs, it was observed that when the maximum
The failure type should be determined correctly in order to tensile fatigue load of 300/3000 N (R = 0.1) was applied to the
completely understand the mechanical properties of adhesively bonded joints, lateral vibration occurred due to eccentric loading. However,
joints. The failure types formed in the SLJs, which are presented in this vibration was observed to decrease when the applied force level
Fig. 13, were determined according to the failure types defined in ISO was decreased. In addition, increased adherend thickness caused an
10365. As can be seen from the failure surfaces given in Fig. 13, inter­ increase in the flexural rigidity of the adherend, thereby increasing
face/adhesive failure - which occurred between the adhesive layer and the ability of the joint to absorb the bending moment. This signifi­
the interface of the adherend - occurred in all joint types subjected to cantly increased the fatigue strength limit of the joint.
high load levels and low cycles.
However, when the maximum load level decreased and the number Acknowledgement
of cycles increased (e.g. SLJ-F-2f, SLJ-F-3e, SLJ-F-3g, SLJ-F-4f, SLJ-F-5f,
SLJ-F-6e and SLJ-F-6f), the failure type changed from interface/adhe­ This research has been supported by Erzurum Technical University
sive to adhesion in all joint types (failure occurred between the adhesive Scientific Research Project (Project No: 2019/16). The authors would
layer and the adherend interface, but the adhesive layer remains on the like to thank Erzurum Technical University for funding this project.

9
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

Appendix

Force, number of cycles and average number of cycles for which the joints were subjected to fatigue loading.

Joint Type Force-N Number of cycle Average number of cycles Joint Type Force-N Number of cycle Average number of cycles

SLJ-F-2a 3000 72857 73214 SLJ-F-3a 3000 61369 45515

76445 36390
68968 34426
87851 64340
59949 31050
SLJ-F-2b 2500 102841 116125 SLJ-F-3b 2500 114118 84812
118653 59243
125571 108037
136406 73020
97154 69642
SLJ-F-2c 2000 207370 159654 SLJ-F-3c 2000 129593 134416
186385 117903
208729 157779
220201 96880
155585 169925
SLJ-F-2d 1750 301402 347174 SLJ-F-3d 1750 243101 271567
353193 326612
372563 227364
308874 196813
399838 363945
SLJ-F-2e 1500 683779 656727 SLJ-F-3e 1500 523525 456377
624038 420154
601170 392050
666614 562821
708034 383335
SLJ-F-2f 1375 901342 897643 SLJ-F-3f 1375 546323 527063
915454 476808
904358 496258
844216 644335
922845 47591
SLJ-F-2g 1250 1000000 1000000 SLJ-F-3g 1250 849631 787441
1000000 720710
1000000 858286
1000000 708971
1000000 799607
SLJ-F-3h 1175 1000000 1000000
1000000
1000000
1000000
1000000
SLJ-F-4a 3000 64176 61017 SLJ-F-5a 3000 41369 33572
49959 37861
65098 47267
70925 19608
54927 21755
SLJ-F-4b 2500 180703 146124 SLJ-F-5b 2500 97215 72306
148317 85239
95318 68067
171871 65292
134411 45717
SLJ-F-4c 2000 239871 221128 SLJ-F-5c 2000 261143 220997
195495 290814
179684 184569
202359 178655
288231 189804
SLJ-F-4d 1750 364024 367170 SLJ-F-5d 1750 489499 546435
319442 551700
400747 477554
425051 557133
326586 656289
SLJ-F-4e 1625 646145 523584 SLJ-F-5e 1675 859821 752478
625347 741055
402945 723756
498186 652816
446647 784942
(continued on next page)

10
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

(continued )
Joint Type Force-N Number of cycle Average number of cycles Joint Type Force-N Number of cycle Average number of cycles

SLJ-F-4f 1500 967585 871953 SLJ-F-5f 1600 967585 911374

8721575 872157
934249 934249
744898 744898
840876 840876
SLJ-F-4g 1400 1000000 1000000 SLJ-F-5g 1550 1000000 1000000
1000000 1000000
1000000 1000000
1000000 1000000
1000000 1000000
SLJ-F-6a 3000 141403 118736
76398
98219
153662
123998
SLJ-F-6b 2500 182851 164553
117059
149667
174913
198275
SLJ-F-6c 2250 168918 214862
236805
175561
243128
249898
SLJ-F-6d 2000 379806 404381
363048
477164
462026
339861
SLJ-F-6e 1900 586974 648614
572494
616292
761391
705919
SLJ-F-6f 1800 827276 832672
816262
793760
796831
929231
SLJ-F-6g 1750 1000000 1000000
1000000
1000000
1000000
1000000

11
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

P–N curves of joints with different adherend thicknesses; a) 2 mm, b) 3 mm, c) 4 mm, d) 5 mm, e) 6 mm.

12
R. Sahin and S. Akpinar International Journal of Adhesion and Adhesives 107 (2021) 102845

References [11] Azari S, Ameli A, Papini M, Spelt JK. Adherend thickness influence on fatigue
behavior and fatigue failure prediction of adhesively bonded joints. Composites
Part A 2013;48:181–91.
[1] da Silva LFM, Adams RD. Techniques to reduce the peel stresses in adhesive joints
[12] Demir K, Bayramoglu S, Akpinar S. The fracture load analysis of different support
with composites. Int J Adhesion Adhes 2007;27:227–35.
patches in adhesively bonded single-lap joints. Theor Appl Fract Mech 2020;108:
[2] Karachalios EF, Adams RD, da Silva LFM. The behaviour of single lap joints under
102613.
bending loading. J Adhes Sci Technol 2013;27:1811–27.
[13] Bayramoglu S, Demir K, Akpinar S. Investigation of internal step and metal part
[3] Gültekin K, Akpinar S, Özel A. The effect of moment and flexural rigidity of
reinforcement on joint strength in the adhesively bonded joint: experimental and
adherend on the strength of adhesively bonded single lap joints. J Adhes 2015;91:
numerical analysis. Theor Appl Fract Mech 2020;108:102613.
637–50.
[14] Mactabi R, Rosca LD, Hoa SV. Monitoring the integrity of adhesive joints during
[4] Zaroug EM, Kadioglu F, Demiral M, Saad D. Experimental and numerical
fatigue loading using carbon nanotubes. Compos Sci Technol 2013;78:1–9.
investigation into strength of bolted, bonded and hybrid single lap joints: effects of
[15] Khashaba UA, Najjar IMR. Adhesive layer analysis for scarf bonded joint in CFRE
adherend material type and thickness. Int J Adhesion Adhes 2018;87:130–41.
composites modified with MWCNTs under tensile and fatigue loads. Compos Struct
[5] Khoramishad H, Akhavan-Safar A, Ayatollahi MR, da Silva LFM. Predicting static
2018;184:411–27.
strength in adhesively bonded single lap joints using a critical distance based
[16] ASTM D3166-99. Standard test method for fatigue properties of adhesives in shear
method: substrate thickness and overlap length effects. Proc IME J Mater Des Appl
by tension loading. 2012.
2016;231:237–46.
[17] ANSYS. The general purpose finite element software. Version 18.1. Houston, Texas:
[6] Khoramishad H, Crocombe AD, Katnam KB, Ashcoft IA. A generalised damage
Swanson Analysis Systems, Inc.; 2017.
model for constant amplitude fatigue loading of adhesively bonded joints. Int J
[18] Kanar B, Akpinar S, Akpinar IA, Akbulut H, Ozel A. The fracture behaviour of
Adhesion Adhes 2010;30:513–21.
nanostructure added adhesives under ambient temperature and thermal cyclic
[7] Katnam KB, Crocombe AD, Khoramishad H, Ashcroft IA. Load ratio effect on the
conditions. Theor Appl Fract Mech 2018;97:120–30.
fatigue behaviour of adhesively bonded joints: an enhanced damage model.
[19] Kazaz I, Akpinar S, Ozel A. The effects of thermal cycle and nanostructure
J Adhes 2010;86:257–72.
reinforcement on the shear load in adhesively bonded joints. Mech Adv Mater
[8] Azari S, Jhin G, Papini M, Spelt JK. Fatigue threshold and crack growth rate of
Struct 2018. https://doi.org/10.1080/15376494.2018.1522560.
adhesively bonded joints as a function of load/displacement ratio. Compos Appl Sci
[20] ISO 10365. Adhesives-Designation of main failure patterns Adhesives-Designation
Manuf 2014;57:59–66.
of main failure patterns. 1992.
[9] Mariam M, Afendi M, Abdul Majid MS, Ridzuan MJM, Gibson AG. Tensile and
[21] Kim MH, Hong HS. An adaptation of mixed-mode I + II continuum damage model
fatigue properties of single lap joints of aluminium alloy glass fibre reinforced
for prediction of fracture characteristics in adhesively bonded joint. Int J Adhesion
composites fabricated with different joining methods. Composites structure 2018;
Adhes 2018;80:87–103.
200:647–58.
[10] Gavgali E, Sahin R, Akpinar S. An investigation of the fatigue performance of
adhesively bonded step-lap joints: an experimental and numerical analysis. Int J
Adhesion Adhes 2021;104:102736.

13