International Journal of Adhesion & Adhesives 56 (2015) 24–31

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives

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

Finite element analysis of adhesively bonded composite joints

subjected to impact loadings
R. Hazimeh a,b, G. Challita b,n, K. Khalil b, R. Othman c
a
LUNAM Université, Ecole Centrale de Nantes, Institut de Recherche en Génie Civil et Mécanique, UMR CNRS 6183, BP 92101, 44321 Nantes Cedex 3, France
b
Equipe MMC, Université Libanaise, Faculté de génie, Campus Hadath, Beyrouth, Lebanon
c
Mechanical Engineering Department, Faculty of Engineering, King Abdulaziz University, PO Box 80248, Jeddah 21589, Saudi Arabia

art ic l e i nf o a b s t r a c t

Available online 30 July 2014 The main concern of this paper is to explore the geometrical and material effects on composite double
Keywords: lap joints (DLJ) subjected to dynamic in-plane loadings. Thus, three-dimensional finite element analyses
Double lap joint were carried out at quasi-static and impact velocities. The DLJ alone was used for quasi-static case while
Impact an output bar was added for impact case. Elastic behavior was assumed for both adhesive and
Composites adherends. Average shear stress and stress homogeneity were extracted and compared. It was observed
Stress distribution that the adhesive shear stiffness increases the average shear stress. Moreover, it makes the stress
Finite element stress analysis heterogeneity more important. On the other hand, higher values of the substrates longitudinal stiffness
make the average shear stress higher; whereas, the stress homogeneity in the joint is better achieved for
lower substrates’ shear stiffness.
& 2014 Published by Elsevier Ltd.

1. Introduction and Kim [7], plies delamination were observed at the highest peel
and shear stresses. Challita and Othman [8] simulated the SHPB
Adhesive bonding offers many advantages to mechanical joints tests on double-lap bonded joints with metal substrates and
such as low weight, cost and the ability to join dissimilar concluded that the SHPB bar method gives a good estimation of
materials. It does not cause distortion or weld worms. For this the mean adhesive stress value and not for adhesive average strain
reason, many works have dealt with this subject in the literature. and maximum stress and strain; a unified parameter was proposed
Different aspects were considered such as static, dynamic and to correct the SHPB results. Stress wave propagation in epoxy-steel
impact loadings. Stepped-lap joints under tensile impact were cylinders subjected to impact push-off loads under small strain
analyzed by Sawa and Ichikawa [1] and showed that peak stress rates was analyzed by Liao and Sawa [9] and showed that normal
increases with the increase of adherends young’s modulus. stresses increase with the increase of adherend/adhesive stiffness
Mechanism of damage formation in glass-epoxy composite joints ratio. Liao et al. [10] studied the single-lap joint (SLJ) subjected to
under transverse impact was analyzed by Kim et al. [2]. Vaidya impact tensile medium strain rate, as overlap length increases,
et al. [3] found that peak stress for bidirectional composite joint maximum principal stress decreases while adherends young’s
under transverse impact is 10 times higher than under in-plane modulus and loading rate have the opposite effect.
quasi-static loading. Carlberger and Stigh [4] analyzed impact The aim of this paper is to present a numerical 3D parametric
fracture in aluminum joints under tensile impact and showed that study on the stress distribution inside adhesive layer for composite
an increase of the strain rate can be achieved in the adhesive layer. DLJ under in-plane quasi-static and impact loading cases. Contrarily
Bonded cylinders under shear impact loading were modeled by to Challita and Othman [8], we are dealing here with composite
Sawa et al. [5] to verify experimental split Hokinson pressure bars substrates.
(SHPB) tests. Silberschmidt et al. [6] studied the effect of impact
fatigue on the crack growth of bonded joints. High velocity
transverse impact on composite joints was investigated by Park
2. Method

n
2.1. Sample geometry
Corresponding author.
E-mail addresses: [email protected] (R. Hazimeh),
[email protected] (G. Challita), [email protected] (K. Khalil), In this paper, we were interested in the double-lap adhesive
[email protected] (R. Othman). joints as depicted in Fig. 1. Since peel stresses are reduced in

http://dx.doi.org/10.1016/j.ijadhadh.2014.07.012
0143-7496/& 2014 Published by Elsevier Ltd.
 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31 25

double lap joints comparing to single lap joints, we focused on the plate is moving at almost constant velocity whereas the second
shear stress distribution inside the adhesive layer. A compressive one is fixed. In order to simulate such loading case, a 0.1 mm/min
rightwards load was applied on the central substrate which yields to a steady-state velocity was applied on the inner substrate of the
shear stress state in the adhesive layer. First, a reference model was joint for a total time of 90 s. Therefore, a total displacement of
studied. Subsequently, a parametric study considering the different 0.15 mm was imposed to the inner substrate by the end of the
geometrical and material parameters was carried out while maintain- simulation. This loading case is referred hereafter as the quasi-
ing adherends’ similarity in the material and orientation. static case. A second loading case was considered and is referred
hereafter as the impact case. It simulates the loading to which a
double-lap specimen is submitted during a direct-impact Hopkin-
2.2. Loading cases son bar test. Therefore, a velocity impact pulse was applied on the
inner substrate for a total duration of 20 ms. The impact pulse is
Following Challita and Othman [8], we are interested in shown in Fig. 2. Similarly to the quasi-static case, an almost total
analyzing the accuracy of experimental testing of double-lap joints displacement of 0.15 mm was imposed to the inner substrate at
by using finite-element simulations. Contrarily to Ref. [8], we are the end of the numerical simulation.
dealing with composite substrates and not metal ones.
In the quasi-static range, the double lap joints are commonly tested
by servo-hydraulic machines. The loading rate can be as slow as 2.3. Material properties
0.1 mm/min. At impact loading rates, the split Hokinson pressure bars
system is widely used. Strain rate can be of 104 s
1. The specimen is Both adhesive and substrates were assumed elastic as sug-
sandwiched between two bars, termed respectively input and output gested by Higuchi et al. [17,18] and Sawa et al. [19] who compared
bar; the incident wave splits into two other waves at the specimen- their results to drop weight experiments. Indeed, elastic behavior
“input bar” interface, a reflecting wave through the first input bar and is valid for elastic-brittle adhesives before failure and for ductile
a transmitted wave through the specimen to the output bar. The adhesives before yielding. The results of this study should be
reader is referred to Ref. [8] for further details on the SHPB method. considered in this framework. Isotropy was assumed for the
Yet, Dharan and Hausser [11] introduced the direct-impact technique, adhesive. However, as we were interested in composite laminate
whereas, the input bar is removed and the specimen is directly adherends, orthotropic elasticity was considered for substrates.
impacted by the incident bar. In the case of the reference numerical model, the material
Usually, in a servo-hydraulic mechanical test, the specimen (the properties of Polyether ether ketone (PEEK) reinforced with 60%
double lap joint) is sandwiched between two rigid plates. One volume glass fiber were adopted. Moreover, we assumed that
fibers are unidirectional and oriented parallel to the load with ply
thickness of 0.2 mm. The adhesive is Araldite 2031, black epoxy
system suitable for composite bonding. Material properties for
reference model are shown in Table 1.
The material properties for the substrates are calculated using
the mixing law.

2.4. Numerical model

For the quasi-static loading, the sole specimen was modeled

Fig. 1. Sample geometry with width W. with proper boundary conditions. For the impact case, the speci-
men and the output bar were modeled. Due to symmetry condi-
tions, one-fourth of the system was modeled. The numerical
models are shown in Fig. 3.
The commercial software ABAQUS was used with its implicit
module for quasi-static case and explicit module for impact case.
Tied node-to-surface was used between adhesive and substrates.
The C3D8R 8-node solid element was used; each node has
3 degrees of freedom with reduced integration and hourglass
control.
Since the thickness of the adhesive is a parameter, and we
averaged stresses through the adhesive thickness, an element size
of 0.025 mm through thickness was chosen to build-up any
adhesive layer, which corresponds to 4 elements over adhesive
thickness for the reference model. A 5  100  25 μm3 smallest
element was used at the joint edges. This mesh size was sufficient
Fig. 2. Impact pulse. to achieve convergence as will be shown in Section 2.6.

Table 1
Material properties [12–16].

Young’s modulus (GPa) Poisson’s ratio Rigidity’ s modulus (GPa) Limit stress (MPa)

PEEK Em ¼ 4.1 νm ¼0.4 Gm ¼ 1.3 σ ¼118

Glass E Ef ¼72 νf ¼ 0.22 Gf ¼ 30 σ ¼3300
ν12 ¼ν13 ¼ 0.292 G12 ¼ G13 ¼ 3.05
Composite ply: Vf ¼60% and Vm ¼40% E1 ¼ 44.84 E2 ¼ E3 ¼9.44 σR ¼2730
ν23 ¼ 0.4 G23 ¼ 3.37
Adhesive (Araldite 2031) E¼ 1 ν ¼0.4 G ¼0.35 σ ¼20
26 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31

Fig. 3. Numerical mesh of the model: (a) quasi-static case and (b) impact case.

For the quasi-static case, the composite plies were modeled as

solid composite layup with 5 integration points each, which means
1 element through thickness contains 10 plies. Each simulation
was carried out for 90 s with results written each 0.5 s.
For the impact case, and since composite layup cannot be used
in ABAQUS explicit, each composite ply was modeled as a part
having orthotropic material properties with a specified material
orientation, and tied to each other which developed the composite
laminate. Each simulation was carried out for 40 ms with results
stored each 0.25 ms.

2.5. Average shear stress and homogeneity coefficient

Fig. 4. Convergence curves, edge shear stress in terms of fine-mesh quality.
The objective is to capture the geometrical and material
influence on the shear stress distribution in the adhesive layer. Table 2
For the quasi-static case, stress distributions were extracted at the Error% for different fine-mesh quality configurations.
total time point, which corresponds to the higher displacement–
Number of elements for the 0.5 mm edge 10 20 30 45 60 75 90
stress levels. Meanwhile, stress distributions were extracted at
Error (%) 12.4 5.0 0.5 0.1 0 0 0
their highest point for the impact-case which does not correspond
to the higher displacement value, as will be shown in Section 3.1,
Fig. 7. Stress values were extracted at the mid-plane cutting the were interested in the permanent regime which occurs before the
model into half through the width, and averaged through adhesive maximum shear stress value as will be shown in Fig. 7. Therefore,
thickness. Actually, the average shear stress τav xy should be calcu-
we focused on the homogeneity coefficient ατxz ðtÞ at the instant
lated as: (time reference) when the shear stress equals its maximum value.
Z L Z W Z Ta
1
τav
xz ðtÞ ¼ τxz ðx; y; z; tÞdxdydz ; ð1Þ
L  Ta  W 0 0 0 2.6. Convergence study
where L, Ta and W are defined in Fig. 1 as overlap length, adhesive
thickness and joint width, respectively. Fortunately, the shear In order to ascertain convergence, a primary mesh convergence
stress is almost constant through the joint width. Consequently, study for the entire model was applied. Subsequently, we have
we considered the shear stress at the vertical mid-plane focused on the critical region (near edges) because of the high
(y ¼ W=2). Then the average shear stress was approximated by stress gradients in this part of the model. Namely, a very fine mesh
Z L Z Ta   was applied at the edges of the adhesive layer since stress peaks
1 W and stress gradients occur at these edges. Fig. 4 depicts the shear
τav
xz ðtÞ  τxz x; y ¼ ; z; t dxdz: ð2Þ
L  Ta 0 0 2 stress along the adhesive joint for a number of elements ranging
The maximum average shear stress used for comparison was from 10 to 90. We were interested in the peak stress value. It is
defined as the maximum with respect to time of the average shear therefore shown that convergence was obtained for 45 elements.
stress. It reads Subsequently, the result obtained by 90 elements were considered
as the reference value to calculate the relative errors in terms of
τxzmax ¼ max τav
xz ðtÞ ð3Þ the number of elements at joint edge. Table 2 shows that the
t
relative error is less than 0.1% for a number of elements higher
Additionally to the average shear stress, homogeneity coeffi-
than 45. Since processing time difference was negligible between
cient ατxz ðtÞ was defined to quantify the stress heterogeneity in the
45 and 60 elements, we chose the 60 elements formulation in
joint. This coefficient reads
order to preserve convergence in case if higher peak stress occurs.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u  
u 1 Z L Z T a τxz ðx; y ¼ ðW=2Þ; z; tÞ
τav ðtÞ2
ατxz ðtÞ ¼ t xz
dxdz ð4Þ
L  Ta 0 0 jτavxz ðtÞj
2
3. Results and discussions
The more this coefficient tends to 0, the more the shear stress is
homogeneous in the adhesive layer. 3.1. Reference model
Specimen geometry, material properties and wave propagation
govern the homogeneity of stress distribution [8]. Dynamic effects A reference numerical model was first investigated then only
disappear once the transient regime disappears. In this study, we one parameter was changed each time for comparison purposes.
 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31 27

Table 3 shows material and geometrical parameters for reference observed for both loading cases. It is worth noting that the
model. tendency of the homogeneity coefficient in terms of the adhesive
Figs. 5 and 6 reveal shear stress in the adhesive layer with thickness is in line with the observations made by Challita and
respect to time and length for impact and quasi-static reference Othman [8]. The increase in the adhesive thickness reduces its
models, respectively. As expected, no wave propagation occurs for shear stiffness. Hence, it seems that stiffer adhesives enhance the
quasi-static case, shear stress increases monotonously with dis- average shear stress and the homogeneity coefficient.
placement and time. On the opposite, multiple oscillations appear
for the impact case. Peaks or resonances are due to the multiple 3.3.2. Adherend thickness effect
wave reflections within the joint. Shear stress reaches maximum Adherends’ thickness effects were investigated and results are
around 20 μs which corresponds to the duration of the impact presented in Fig. 10. It is shown that the average shear stress
pulse. Subsequently, the stress decreases as soon as the energy increases with the increase of the adherend thickness. On the
transmits to the output bar. opposite, stress homogeneity coefficient is decreasing for increas-
This difference in tendency between quasi-static and impact cases ing adherend thickness. This last observation is in line with the
is confirmed in Fig. 7. Fig. 7a depicts the average shear stress which results obtained in Ref. [8] with isotropic metallic substrates.
increases linearly in the quasi-static case. For the impact case, the The increase of the adherend thickness induces a decrease in
average shear stress is monotonously increasing during the impulse the adherends’ shear stiffness. On the opposite, it improves the
load (o20 μs). However, the stress increase rate is lower at the first μ- longitudinal stiffness. Therefore, it seems that the average shear
seconds. In this transient period, the left side of the joint is more stress increases with increasing adherends’ longitudinal stiffness.
loaded than the right part, and stress field is highly heterogeneous. On the other hand, the homogeneity coefficient increases with
This is confirmed by Fig. 7b. Namely, homogeneity coefficient is very increasing adherends’ shear stiffness.
high at the beginning of the simulation (first 6 μs); it drops to an
asymptotic value. In the quasi-static, the homogeneity coefficient is 3.3.3. Overlap length
constant and equal to the asymptotic value of the impact. Actually, In a third step, we investigated the influence of the overlap
heterogeneity is caused either by wave propagation within the speci- length. The variation of the average shear stress and stress
men (transient heterogeneity) or by the specimen geometry (struc- homogeneity in terms of the overlap length are depicted in
tural heterogeneity). As observed by Challita and Othman [8], the
transient heterogeneity decreases with time and disappears after
some μ-seconds (here  6 μs). The structural heterogeneity is inde-
pendent of time. For the quasi-static case, heterogeneity is only due to
the specimen geometry, meanwhile, for the impact case, transient and
structural heterogeneities coexist at the beginning of the simulation.
In this parametric study, we focused on the maximum average
shear stress and on the homogeneity coefficient. The average shear
stress gives information on the global evolution of the stress state
within the joint. On the other hand, the homogeneity coefficient
gives valuable information on the evolution of the stress field
heterogeneity. By studying the average shear stress, we will have
an insight into the sensitivity of the average shear stress state
within the joint to the different investigated parameters. The
homogeneity coefficient quantifies the distribution of the stress
around its average value.
Fig. 5. Shear stress along overlap with respect to time, impact reference model.
3.2. Normal stresses inside adherend

In this section, we were interested in the no failure assumption for

the substrates. This study is limited to the reference model. Namely,
normal stresses were extracted and compared with limit normal
stresses for outer composite substrate. Normal stresses in all the plies 100
are plotted in Fig. 8. Being the fact that the limit stress for composite is 80 80-100
τxz (x,t) (MPa)

2730 MPa, and the peak normal stress is 540 MPa, one can assume 60 60-80
that no failure occurs inside the composite adherend. 40
87 40-60
20
51 20-40
0
3.3. Effect of geometrical parameters -20 0.0 14 0-20
0.4
4.0 4
9.0 -20-0
3.3.1. Adhesive thickness effect 13.5 0
The influence of the adhesive thickness was investigated by 14.0

varying the adhesive thickness from 0.05 mm to 0.25 mm. Fig. 9

shows that both average shear stress and homogeneity coefficient Fig. 6. Shear stress along overlap with respect to time, quasi-static
decrease with increasing adhesive thickness. Same tendency reference model.

Table 3
Reference model, material and geometrical parameters.

Parameter Eadhesive (MPa) Fiber volume (%) Tadhesive (mm) Tinner adherend (mm) Overlap (mm) Width (mm) Orientation

Value 1000 60 0.1 4 14 12 0 (parallel to X-axis)

28 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31

Fig. 7. Reference model results: (a) average shear stress and (b) homogeneity coefficient in terms of time.

adhesive Young’s modulus. More precisely, stiffer adhesive

enhances average shear stress but reduces homogeneity. This
result is in line with the conclusion made in Section 3.3.1.

3.4.2. Adherend fiber volume effect

In order to investigate the effect of the substrates’ stiffness, we
considered in this section the influence of the fiber volume
percentage. From Fig. 14, it can be concluded that fiber volume
has the same effect as the substrate thickness and the opposite
effect of the overlap length. The average shear stress increases
while homogeneity coefficient decreases with increasing fiber
volume. According to mixing law, the longitudinal stiffness
increases while the shear stiffness decreases as the fiber volume
is augmented. Here also, the average shear stress becomes higher
for longitudinally stiffer substrates. On the other hand, the stress
heterogeneity in the joint is enhanced as the shear stiffness of the
Fig. 8. Normal stresses in the outer adherend in the X–Z plane of symmetry adherends becomes greater.
(impact reference model).
Challita and Othman [8], showed that the homogeneity coeffi-
cient decreases with the isotropic substrate’s Young’s modulus. It
seems here that the shear out-of-plane modulus G13 is the most
Fig. 11. The effects of the overlap length are opposite to the effects important parameter.
of the substrates thickness. This can be explained by the fact that
the overlap length has opposite effects on the longitudinal and
shear stiffness of the substrates. In other terms, the increase of the 3.4.3. Effect of substrate orientations
overlap length increases the substrates’ shear stiffness and In order to investigate the effect of anisotropy of substrates, we
decreases the longitudinal stiffness. Hence, the average shear chose four different configurations for the adherend fiber orienta-
stress and homogeneity coefficient increase respectively with tions in addition to the reference one. Fig. 15 illustrates the results.
increasing longitudinal and shear stiffness’s of the substrates. This The 01 orientation is longitudinal parallel to the load, while the
observation is in line with the conclusions of Section 3.3.2. laminate (0, 90) means the first ply is 01 then 901 then 01 and
subsequently, same for the laminate (45,
45).
It was shown that the toughest joint is the joint with fiber
3.3.4. Width effect
parallel to the loading direction and a huge decrease in average
The last geometrical parameter investigated in this work is
shear stress appears when no fiber percentage is parallel to the
joint width. Results are depicted in Fig. 12 which shows that width
loading direction. It does not appear that the 451 fiber is better
has almost no effect either on average shear stress nor on stress
than the 901. Indeed, the longitudinal stiffness of substrates is
homogeneity. These results are quite expected as the problem can
higher for 01 fibers. Thus, the average shear stress is the best for
be assumed to be two-dimensional (Fig. 13).
the highest longitudinal stiffness, i.e., when all fibers are in 01
direction.
3.4. Effect of material parameters Stress homogeneity coefficient is higher for 451, 901 or ( þ451/

451). Indeed, G12 is higher for these configurations. We have
3.4.1. Adhesive Young’s modulus effect assumed that G12 ¼ G13 . The shear substrates stiffness is higher for
Five values of the adhesive Young’s modulus were investigated 451, 901 or ( þ451/
451). Hence, we can conclude one more time
in this section. The parametric study shows that both average that stress field is more homogeneous for low substrate’s shear
shear stress and homogeneity coefficient increase with increasing stiffness.
 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31 29

Fig. 9. Effect of the adhesive thickness on average shear stress (a) and homogeneity coefficient (b).

Fig. 10. Effect of the adherend thickness on average shear stress (a) and homogeneity coefficient (b).

Fig. 11. Effect of the overlap length on average shear stress (a) and homogeneity coefficient (b).

4. Conclusion layer. The stress is due to either structural or wave propagation

effects. Structural or geometrical heterogeneity is independent of
A parametric three-dimensional finite element study for double time. Its contribution does not alter from quasi-static to impact
lap joints of similar composite adherends under quasi-static and case. On the opposite, heterogeneity due to wave propagation
impact loads was accomplished, revealing the geometrical and effects was very high at the beginning of the simulation and
material influence on the shear stress distribution in the adhesive disappears within some micro-seconds. As we were interested in
30 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31

Fig. 12. Effect of the joint width on average shear stress (a) and homogeneity coefficient (b).

Fig. 13. Effect of the adhesive’s Young’s modulus on average shear stress (a) and homogeneity coefficient (b).

Fig. 14. Effect of the fiber volume on average shear stress (a) and homogeneity coefficient (b).

the permanent regime, only structural effects are influencing with increasing substrates’ thickness, adhesive Young’s modulus
stress state in the adhesive joint. Therefore, close results were and substrates’ fiber volume fraction. It was decreasing with
obtained for both quasi-static and impact cases. For the same increasing adhesive thickness and overlap length. On the other
imposed displacement, the average shear stress was increasing hand, the homogeneity coefficient was increasing with increasing
 R. Hazimeh et al. / International Journal of Adhesion & Adhesives 56 (2015) 24–31 31

Fig. 15. Effect of fiber orientation on average shear stress (a) and homogeneity coefficient (b).

overlap length and adhesive Young’s modulus. It was decreasing [3] Vaidya UK, Gautam ARS, Hosur M, Dutt P. Experimental–numerical studies of
with increasing adhesive and substrates’ thicknesses and fiber transverse impact response of adhesively bonded lap joints in composite
volume fraction. It was also shown that the highest average shear structures. Int J Adhes Adhes 2006;26:184–98.
[4] Carlberger T, Stigh U. An explicit FE-model of impact fracture in an adhesive
stress is achieved for fibers oriented in the same direction as the joint. Eng Fract Mech 2007;74:2247–62.
load. This orientation yields also to the best stress state homo- [5] Sawa T, Nagai T, Iwamoto T, Kuramoto H. A study on evaluation of impact
geneity. These results can be interpreted by the fact that the strength of adhesive joints subjected to impact shear loadings. ASME IMECE
2008;15:55–61.
average shear stress increases with increasing adhesive shear
[6] Silberschmidt VV, CasasRodriguez JP, Ashcroft IA. Impact fatigue of adhesive
stiffness and substrate longitudinal stiffness; whereas, the stress joints. Adv Strength Mater 2009;399:71–8.
heterogeneity increases with increasing adhesive and substrate [7] Park H, Kim H. Damage resistance of single lap adhesive compoiste joints by
shear stiffness. This is the main finding of this study. Indeed, we transverse ice impact. Int J Impact Eng 2010;37:177–84.
[8] Challita G, Othman R. Finite-element analysis of SHPB tests on double-lap
succeeded to separate the influence of substrates’ longitudinal and
adhesive joints. Int J Adhes Adhes 2010;30:236–44.
shear stiffness as they were considered as orthotropic materials. [9] Liao L, Sawa T. Finite element stress analysis and strength evaluation of epoxy-
On the opposite, this can hardly be done by assuming isotropic steel cylinders subjected to impact push-off loads. Int J Adhes Adhes 2011;31:
elastic substrates. Finally, this study was carried out assuming 322–30.
[10] Liao L, Kobayashi T, Sawa T, Goda Y. 3-D FEM stress analysis and strength
elastic behavior for both adhesive and substrates. Therefore, no evaluation of single-lap adhesive joints subjected to impact tensile loads. Int J
failure should occur neither in the adhesive nor in the adherends. Adhes Adhes 2011;31:612–9.
Consequently, the results of this work should be understood in this [11] Dharan CKH, Hausser FR. Determination of stress–strain characteristics at very
framework. high strain rates. Exp Mech 1970;10:370–6.
[12] Villoutreix J, Acetarin JD. Polyétheréthercétone. Techniques de l’ingénieur
AM3394.
[13] Guillon D. Fibres de verres de renforcement. Techniques de l’ingénieur A2 110.
Acknowledgements [14] Odru P. Calcul et conception des structures composites. Techniques de
l’ingénieur A7 792.
[15] Chevalier Y. Comportement élastique et viscoélastique des composites. Tech-
Khaled Khalil (MMC Team) would like to thank l’Ecole Doctor-
niques de l’ingénieur A7 750.
ale des Sciences et Technologies for their collaboration in this [16] HUNTSMAN adhesives, Araldite 2031; February 2009.
work. [17] Higuchi I, Sawa T, Suga H. Three-dimensional finite element analysis of single-
lap adhesive joints under impact loads. J Adhes Sci Technol 2002;16:
1585–601.
References [18] Higuchi I, Sawa T, Suga H. Three-dimensional finite element analysis of single-
lap adhesive joints subjected to impact bending moments. J Adhes Sci Technol
[1] Sawa T, Ichikawa K. A stress analysis and strength estimation of stepped lap 2002;16:1327–42.
adhesive joints under static and impact tensile loadings. ASME IMECE [19] Sawa T, Suzuki Y, Kido S. Stress analysis and strength estimation of butt
2005:819–25. adhesive joints of dissimilar hollow cylinders under impact tensile loadings.
[2] Kim H, Kayir T, Mousseau SL. Mechanisms of damage formation in transver- J Adhes Sci Technol 2003;17:943–65.
sely impacted glass-epoxy bonded lap joints. J Compos Mater 2005;39:
2039–52.