International Journal of Adhesion & Adhesives 57 (2015) 5769

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives

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

Assessment of local strain eld in adhesive layer of an unsymmetrically

repaired CFRP panel using digital image correlation
Mohammad Kashfuddoja n, M. Ramji
Engineering Optics Lab, Department of Mechanical and Aerospace Engineering, IIT Hyderabad, India

art ic l e i nf o

a b s t r a c t

Article history:
Accepted 20 October 2014
Available online 4 November 2014

The present study focuses on experimental investigation of through the thickness displacement and
strain eld in thin adhesive layer in single sided (unsymmetrical) patch repaired CFRP (carbon ber
reinforced polymer) panel under tensile load. Digital image correlation (DIC) technique is employed to
acquire the displacement and strain (longitudinal, peel and shear) eld. Experimental determination of
shear transfer length based on shear strain eld obtained from DIC is introduced to estimate the
optimum overlap length which is an essential parameter in patch design for the repair of CFRP
structures. Further, DIC experiment with magnied optics is performed to get an insight into complex
and localized strain eld over thin adhesive layer especially at critical zones leading to damage initiation.
The failure mechanism, load displacement behavior, damage initiation and propagation are closely
monitored using DIC. The inuence of patch edge tapering on strain distribution in adhesive layer is also
investigated. The DIC successfully captures the global and localized strain eld at critical zones over thin
adhesive layer and further helps in monitoring the damage based on strain anomalies. Strains are found
to have maximum magnitude at the patch overlap edge and the shear strain level in adhesive layer is
higher than the peel strain. Normal tapering increases the peel strain and has negligible inuence on
shear strain level in adhesive layer. The recommended overlap length is found to be consistent with the
recommendation in the literature. Whole eld strain pattern and the overlap length obtained from
experiment are further compared with the nite element analysis results and they appear to be in good
coherence.
& 2014 Elsevier Ltd. All rights reserved.

Keywords:
Adhesive strain
Carbon/Epoxy composite
Digital image correlation
Debonding
Load transfer/shear transfer length
Longitudinal
Peel and shear strain
Repair

1. Introduction
Adhesively bonded composite patch repair is one of the most
prominent techniques to restore the structural integrity of damage
structures. This is because the adhesively bonded repair provides
lightweight, minimal source of stress concentrations and efcient
load transfer from one adherend to another [1] in comparison to
its counterpart and therefore it is being used in various applications especially in aerospace and marine industry. Since the load is
transferred between patch and panel through the adhesive layer,
adhesive plays a crucial role in load carrying capacity of repaired
panel. Adhesive layer being load transfer medium constitutes the
weakest link and failure usually initiates from such location in
adhesively bonded structures.
Several analytical, numerical and experimental works have
been carried out to understand the behavior of adhesively bonded
joints. Davis and Bond [2] have discussed the basic principles and

n
Corresponding author.
Tel.: 91 40 23 01 6078/ 91 40 23 01 6123; fax: 91 40 23 01 6032.
E-mail address: mdkashfuddoja@gmail.com (M. Kashfuddoja).

http://dx.doi.org/10.1016/j.ijadhadh.2014.10.005
0143-7496/& 2014 Elsevier Ltd. All rights reserved.

practices to achieve a strong and durable adhesive bond in

adhesively bonded structural joints and repairs. A micrographic
study on adhesive bonded scarf repair is presented by Whittingham et al. [3] to assess critical features of the bond-line produced
from the different techniques. The area of structural bonded repair
of composites is broadly reviewed by Katnam [4] starting from
damage assessment to automation to identify current scientic
challenges and future opportunities. A comprehensive review of
analytical investigations on adhesively bonded joints pertaining to
both single and double-lap conguration is presented by Banea
and da Silva [5] and da Silva et al. [6]. A comparative study on
different analytical models for adhesively bonded joints is also
reported by da Silva et al. [7]. Most of the existing analytical
models are two-dimensional by nature. Also, they do not account
for change in geometry and boundary condition and they are
formulated assuming linear elastic nature therefore idealizes the
response of joint to avoid complexity. Finite element analysis (FEA)
has been used over the last two decades to overcome the
limitations of analytical models. A detailed review on numerical
modeling of adhesively bonded joints is presented by da Silva and
Campilho [8] and He [9]. To validate the numerical predictions
researchers have used several experimental techniques such as

58

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

strain gauges [1012], moir interferometry [10,13], photoelasticity [14] and digital image correlation (DIC) [1525]. DIC has been
used extensively in recent years because of its advantages over the
other experimental technique [22,2628]. The DIC technique has
the advantage of being applied to any class of material, it requires
no heavy specimen surface preparation and the optics involved is
quite simple. No fringe analysis and phase unwrapping are needed
in this method and also there is no fringe density limitation which
makes the DIC measurement range much larger than the other
technique. Moreover, it is truly a non-contact by nature and
provides full eld data.
Colavito et al. have used the DIC technique to measure
displacement and strain eld within the adhesive layer of adhesively bonded double-lap joints [15,16]. They concluded that the
DIC can be employed to obtain the displacement eld within the
adhesive layer of a double lap joint and it is able to capture a
difference in behavior between the specimens with different
adhesive thicknesses. Moutrille et al. [17] have studied the shear
strain eld in an adhesive joint between composite patch and
aluminum. They concluded that the shear strain decreases with
increase in distance from the free edge and using these measurements they estimated the in-situ shear modulus of the adhesive.
Wang et al. [18] have presented the results of a combined
experimental and nite element investigation on strain/stress
distributions around the overlap ends of laminated composite
single-lap joints. They concluded that the spew llets with high
rectangular segments and composite bulges at the overlap ends
have impacts on the load carrying ability and stresses peak at the
overlap end. Haghani et al. [19] have investigated the effect of
geometrical modications like laminate end tapering or adding
adhesive ller on strain distribution in adhesive joint between
CFRP laminate and steel using the DIC technique. They concluded
that the normal tapering of the laminate did not affect the shear
and principal strain components, but it increased the maximum
peeling strain in the joint and therefore no improvement in joint
strength is observed. Guo et al. [20] used the DIC technique to
evaluate the shear modulus of adhesive layer between aluminum
and aluminum adherend. Ruiz et al. [21] conducted experimental
and numerical study to assess the strain distribution in adhesive
layer of epoxy bonded aluminumaluminum and aluminumCFRP
double lap shear joints under tensile loading. They compared their
experimental results with the numerical predictions and found
that there is a good match between the FEA and DIC results in case
of aluminumaluminum double lap joint as compared to the
aluminumCFRP joint conguration. Comer et al. [22] have used
both 2D and 3D DIC technique to evaluate the evolution of
deformation and strain in composite single lap bonded joints prior
to failure. They concluded that the 3D DIC measurements may be
useful in detecting subcritical damage. Katnam et al. [23] have
used high magnication 2D DIC to analyze strain distributions
near the adhesive llet regions of adhesively bonded composite
single lap joints. They studied the inuence of adhesive ductility
on joint strength and strain distribution at the overlap ends and
concluded that the adhesive with relatively low tensile strength
and high ductility provides a higher joint strength. Crammond
et al. [24] have used full-eld measurement techniques to analyze
the complex stress and strain distributions in adhesively bonded
composite joints using thermoelastic stress analysis (TSA) and DIC.
The experimental stress and strain data in their study is manipulated to give the same output for the purpose of comparison and a
reasonable agreement between the two independent measurement techniques is achieved. Kumar et al. [25] have presented an
experimental approach of measuring peel and shear strains in the
adhesive bondline of composite single-lap joints using DIC. They
compared their experimental results with the theoretical ones and
achieved a similar trend between experimental and numerical

results. Recently, many researchers [2830] have used the DIC

technique in adhesively bonded patch repair domain for damage
assessment and whole eld strain analysis over patch and panel
surface.
Most of the earlier works have been done on adhesive joint
between metalmetal and metalcomposite particularly for single
and double lap joint conguration having thick adhesive layer. Off late,
the focus is on composite/composite adhesively bonded joints due to
increasing demand of lightweight and high strength structures in
various elds. The adhesively bonded composite joint has also
emerged as a potential means of repairing the damaged composite
structures for attaining high structural efciency and improved fatigue
life. Adhesive layer serves as medium to transfer the load between the
patch and panel in a repaired structure. The strain eld is the
representative of progressive stress transfer between a loaded structure and a composite patch used for reinforcement purposes. However, no signicant whole eld experimental work has been reported
yet especially on the behavior of adhesive layer in unsymmetrical
patch repaired CFRP laminates. To improve the performance of
bonded repair of composite structures, it is important to thoroughly
understand the critical stress/strain distribution in adhesive layer
between composite patch and panel. The assessment of adhesive
behavior is also very critical from design point of view.
The present work emphasizes on analyzing the experimental
behavior of thin adhesive layer of about 0.2 mm thickness, typical
of a normal adhesive layer which is being used in practical
applications in bonded patch repair domain. A methodology is
presented to successfully demonstrate the use of a digital image
correlation technique for full eld strain measurement in such a
thin adhesive layer between composite adherends in an unsymmetrical patch repair conguration, which has got more practical
importance and implications as it involves the complication of
secondary bending due to eccentric loading. The global cum local
through the thickness whole eld strain distribution over adhesive
layer is analyzed using DIC in single sided patch repaired
(SSPR) CFRP panel under tensile load. Displacement as well as
longitudinal, peel and shear strain eld in repaired conguration
together with bending phenomena due to single sided
patch (unsymmetrical repair conguration) is examined. Experimental prediction of shear transfer length is carried out to
get the optimum overlap length which is an essential parameter
in patch design. The failure mechanism, load displacement
behavior, damage initiation and propagation are also closely
monitored using DIC. The effect of patch edge tapering on
strain distribution in adhesive layer is also investigated. Whole
eld strain pattern and the overlap length obtained from experiment are further compared with the nite element analysis
results.

2. Specimen fabrication and conguration

The carbon ber used in present work is manufactured by
Hindustan Technical Fabrics Ltd. India, having a weight of 230 gsm.
The matrix is made from epoxy resin LY556 mixed with hardener
HY951 supplied by Huntsman. The average failure strain for matrix is
1.45%. The CFRP laminates are of pure unidirectional (UD) conguration (laminate with only 01 bers in each ply) and they are fabricated
in-house by the hand layup technique and it is elaborately discussed
in Ref. [28]. Because of the pure UD conguration, due care is taken
while laying up the individual layer in the laminate such that the
bers in every layer are maintained parallel to each other and also to
the loading direction. Firstly, the patch and panel are cut from
fabricated laminate to an over dimension (about 23 mm on each
side) using abrasive cut-off wheel mounted on a hand-held saw. The
patch and panel are then accurately machined to the required

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

59

Fig. 1. Specimen geometry of single sided patch repaired (SSPR) panel: (a) front view and (b) side view.

dimensions using a special diamond-coated end mill (JS520100D3S.0Z6-SIRA) supplied by SECO Jabro Tools. Backing plates are used
to avoid the edge delamination. Before machining, careful marking is
done over the laminate parallel to 01 bers and due care is taken such
that the bers are machined parallel to the marking. The typical
model of the repaired panel is shown in Fig. 1. Both patch and panel
are made of same unidirectional Carbon/Epoxy composite laminate.
The length (L), width (W) and the thickness (t) of the panel are
250 mm, 50 mm and 1.6 mm respectively. The stacking sequence in
the panel is [01]4 and bers are oriented along the loading direction. A
circular hole of 10 mm diameter (d) is drilled at the center of the panel
by using a diamond coated drill bit supplied by SECO-Jabro Tools. The
hole is drilled to simulate the effect of damage removal as it happens
in the case of low velocity impact damage. The panel with open hole
is repaired with adhesively bonded single sided rectangular patch
having stacking sequence of [01]3. The length (Lp), width (Wp) and
thickness (tP) of the patch are 60 mm, 50 mm and 1.2 mm respectively. The bonding surface of the specimens is roughened using 200grit sandpaper and then cleaned with isopropyl alcohol to remove any
foreign particle or grit paper residue. Proper marking parallel to the
bers is done over the panel and then the patch is bonded over the
panel along the marking with due care such that the bers in patch
and panel are parallel to each other. Patch is bonded over the
damaged area of the panel using an adhesive. A uniform weight is
then placed over the bonded patch and the squeezed out excess
adhesive from all the sides is cleaned with a surgical blade before
complete curing to avoid the adhesive bulge at the patch overlap edge.
The excess adhesive at the hole is also removed in all the specimens
and then the bonded panel is left for complete curing at room
temperature for 7 days (same curing time is also used for adhesive
samples for evaluating their properties). Beveled aluminum tabs of
dimension 50 mm  50 mm  2 mm are bonded at each end of the

Table 1
Material properties of CFRP laminate and adhesive obtained using the DIC
technique.
CFRP composite laminate
Longitudinal modulus, Exx (GPa)
Transverse modulus, Eyy Ezz (GPa)
Shear moduli, Gxy Gxz (GPa)
Shear modulus, Gyz (GPa)a
Poisson's ratio (xy xz)
Poisson's ratio (yz)a
Araldite 2011 adhesive
Young's modulus E (GPa)
Poisson's ratio ()
a

84.16
7.12
3.30
2.47
0.31
0.43
1.86
0.38

Out of plane properties are evaluated using the procedure given in Ref. [47].

specimen for gripping purpose. In this study, Araldite 2011 adhesive is

used for bonding the patch. It is a two-part epoxy based adhesive
manufactured by Huntsman and has intermediate strength but higher
toughness. The average failure strain for the adhesive (Araldite 2011) is
8.83%. The properties of considered CFRP composite laminate and
adhesive are evaluated as per ASTM standard based on whole eld
strain data obtained from DIC. The detailed methodology for properties evaluation involving the DIC technique are reported in Ref.
[28,31,32] and they are presented in Table 1.

3. Experimental setup and test methodology

The experimental setup consists of a DIC system supplied by
Correlated Solution Inc. and a computer-controlled MTS Landmarks servo-hydraulic cyclic test machine of 100 kN capacity. The
DIC setup consists of a Grasshoppers CCD Camera (POINTGREY-

60

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

Fig. 2. Experimental setup for adhesive strain measurement involving 2D DIC

with tube lens.

GRAS-50S5M-C) having a spatial resolution of 2448  2048 pixels,

Schneider Xenoplan lenses and two LED lamps of 30 W capacity.
The camera is mounted on a tripod having inbuilt spirit level to
ensure horizontal level. Vic-Snap software from Correlated Solution Inc. is used to grab the images during test. To perform DIC
measurement, random speckle patterns are produced over the
specimen surface using an acrylic paint of titanium white/carbon
black color. The procedure for obtaining the random speckle
pattern is discussed in Ref. [33]. The resulting average speckle
diameter is found to be 38 m.
3.1. Adhesive strain measurement methodology
Two dimensional DIC setup is employed to capture the whole
eld strain developing in the adhesive layer between patch/panel
interfaces. At rst, whole eld strain distribution is obtained along
the entire length of adhesive layer such that the global behavior of
adhesive can be analyzed. For this purpose a TAMRON lens (Model:
SP AF 180 mm f/3.5 Di) mounted to the CCD camera is used. The
camera is kept at distance of 0.65 m from the specimen surface so
that entire adhesive layer could be captured during the test and
therefore in global strain analysis ZOI (zone of interest) corresponds
to entire repaired area in the thickness direction. Later, the localized
behavior of adhesive layer is investigated at critical areas such as
patch overlap edge using magnied optics involving InniProbe TS160 lens (from Innity Photo-Optical Company) thereby providing a
magnication range of 016  . The working distance of camera
from specimen surface is 64 mm (see Fig. 2) and therefore in case of
localized strain analysis ZOI corresponds to the area closer to the
outer edge of the patch. The specimen is xed in hydraulic grips and
camera is aligned perpendicular to ZOI. The ZOI near the patch edge
(see Fig. 1(b)) is zoomed in and tensile load is applied at a loading
rate of 1 mm/min and 10 images per second are grabbed.

4. Finite element modeling and analysis

A linear 3-D FEA of repaired panel is carried out using ANSYS 13
software. The geometry and dimensions of panel and patch are
kept the same as that of experimental model. The mesh pattern
around the hole and patch edge is kept very ne to capture the
high stress gradient around it. Around the circular hole there are
15360 elements (96 circumferential; 40 radial; 4 thickness). The
meshing surrounding the hole is chosen based on mesh convergence study [34,35]. Every layer is meshed with one element in
thickness direction for both patch and panel. In thickness direction, the panel is meshed with four elements, adhesive with 10

Fig. 3. Finite element model of SSPR panel (zoomed up view).

elements and patch with three elements. The model is built with
20-noded solid 186 brick element. The patch is bonded on to the
panel over the hole using adhesive layer. Multipoint constraint
(MPC) algorithm is employed for ensuring a perfect bonding
between patch/panel and panel/adhesive interface. The material
properties presented in Table 1 are used in FEA. Fibers in the panel
and patch are aligned parallel to the loading direction. The panel is
xed at bottom face and an in-plane tensile load is applied at the
top face along x-direction so as to simulate the experimental
boundary conditions. The results obtained from FEA are compared
with the experimental data for the same load. The zoomed view of
nite element model of the panel repaired with straight edge
patch is shown in Fig. 3.

5. Results and discussions

The grabbed images are then post-processed in Vic-2D/Vic-3D
software (Correlated Solution Inc.) for obtaining displacement and
strain eld. The adhesive thickness at interface between patch and
panel is measured by an optical microscope (Olympus STM6) using
objective lens (Olympus MPLFLN 10  /0.30) at a magnication of
10  . The adhesive average thickness for a panel repaired with
Araldite 2011 adhesive is 0.21 mm. The results obtained from
various tests are discussed in subsequent sub-sections.
5.1. Whole eld displacement and strain analysis in adhesive layer
In this section, whole eld strain analysis along the length of
adhesive layer is studied thoroughly. A subset size of 21  21
pixels2 along with step size of 5 pixels is chosen for correlation.
The spatial resolution is 26 pixels/mm.
5.1.1. In-plane displacement contours
The longitudinal (u) and transverse (w) displacement eld
obtained from DIC at a load of 38.5 kN is shown in Fig. 4. It is
evident from u and w displacement data that they are quite
smooth and uniformly distributed about the panel center. However, on a closer look (see Fig. 4(a)), one can observe that the width
and angle of some of the fringes above the panel center are
different than the one below it. Therefore, the fringes are relatively
symmetrical about the mid span point of the overlap or panel
center. The reason to this discrepancy could be attributed to the
development of internal damage such as matrix cracking. One can
also notice from w-displacement eld (see Fig. 4(b)) that they are

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

61

Fig. 5. Variation of transverse displacement (mm) of point P in center of SSPR

panel with increasing load using DIC.

Fig. 4. Displacement (mm) contours in SSPR panel at 38.5 kN: (a) longitudinal
displacement (u) and (b) transverse displacement (w).

diagonal and their angle varies along the panel length from
beginning of the adhesive joint at patch overlap edge till the
cutout zone. This observation implies that a differential deformation mechanism is happening in repaired conguration. The large
angles at and near the patch overlap edge infer that a large
amount of load is being transferred from panel to patch around
this zone through the adhesive layer by the shear deformation
mechanism. The band almost becomes horizontal showing near
zero angles just around the cutout zone which indicates that this
zone has no signicant inuence on load transfer. The present
nding is consistent with the results reported in Ref. [36] which
were presented for double strap joint using the Moir technique.
The results in this section are especially presented in pseudorandom contours to facilitate a qualitative comparison with results
reported using the Moir technique [36]. On qualitative comparison between the two results it is found that the diagonal bands in
the present single sided patch repaired panel case have got more
angles in comparison with double strap joint. This nding seems
obvious since there is a slight shift in neutral axis of singe sided
repair conguration and the load acts through an eccentricity
which causes bending effect in addition to applied in-plane tensile
load. To better understand this phenomenon, the transverse
displacement (w) of a point P located in center of the panel is
traced with increasing load using DIC and it is shown in Fig. 5. It is
found that the transverse displacement of considered point varies
linearly up to a certain load which signies the inuence of
bending phenomena in SSPR panel conguration. The variation
remains constant with further increase in load once the neutral
axis coincided with the loading axis. The bending phenomenon in
SSPR panel conguration is elaborately explained in Ref. [37] and
also reported by Hosseini-Toudeshky et al. [38].

5.1.2. Longitudinal strain (xx) distribution

Fig. 6(a) shows whole eld longitudinal strain (xx) distribution
in SSPR panel at a load of 38.5 kN. The longitudinal strain in panel
is found maximum near the patch overlap edge and it is highly
concentrated at the corner of adhesive joint. The variation of
longitudinal strain in panel along line a-a at 38.5 kN and 46.8 kN
which corresponds to 56% and 68% of ultimate load respectively is
shown in Fig. 7. The magnitude of xx near the root is 6800 and
it reduces to 4368 at center of the panel. The reduction in xx
magnitude is only 35% as one move from the root or adhesive
corner to the panel nearer to cutout. This can be attributed to the
presence of bending phenomena in SSPR panel conguration as
discussed in the earlier section. At 68% of ultimate load the
variation of xx shows a similar variation with strain magnitude
shifted to a higher level. The sudden peak in strain magnitude just
near the root signies that major portion of applied load is
transferred from panel to patch within this zone just near the
patch overlap edge.

5.1.3. Peel (zz) and shear strain (xz) distribution

The peel and shear strain distribution in SSPR panel at 38.5 kN
are shown in Fig. 6(b) and (c) respectively. The peel strain is found
maximum over a length near the adhesive/patch interface at
upper end of adhesive layer. Asymmetry in peel strain distribution
is seen at this load with larger strain on upper side of adhesive
layer as compared to the lower one (see Fig. 6(b)). However, it is
found that the peel strain distribution is symmetric at lower load
level. A zone of compressive strain could be identied in panel
over certain length (white line) opposite to higher peel zone and
just below the patch overlap edge. This is due the development of
high peel strain causing the patch to peel away from panel
inducing a compressive strain in the panel. The shear strain is
also found maximum at the patch overlap edge (see Fig. 6(c)).
Shear strain also shows asymmetric behavior with maximum
magnitude being at the same location (dotted black circle) as high
peel strain zone (upper end of adhesive layer). The asymmetry in
strain eld indemnies the damage initiation. It is important to
note here that the peel and shear strain would also be high in the
adhesive layer at the hole edge but experimentally it is not
possible to measure the strain as it is covered by patch and
therefore through the thickness full eld strain measurement is
carried out in this study. The variation of shear strain in adhesive
layer along a line b-b at a load level of 38.5 kN and 46.8 kN is
shown in Fig. 8. The shear strain variation at both the load levels
shows a similar trend. On comparison of shear strain at two load
levels one can nd that the shear strain remains low over a certain

62

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

Fig. 6. Through the thickness strain distribution in single sided patch repair conguration at 56% of failure load: (a) xx (b) zz and (c) xz.

Fig. 7. Longitudinal strain variation in the CFRP panel under single sided patch
at 56% and 68% of failure load.

Fig. 8. Shear strain variation in the adhesive layer under single sided patch at 56%
and 68% of failure load.

length in center of the panel and then depicts a difference in shear

strain magnitude over a zone till the patch overlap edge. This
observation reveals that the load is transferred from panel to patch
through adhesive layer by the shear mechanism and it happens

over this small zone from the overlap edge. This zone is referred as
shear transfer zone and the corresponding length as shear transfer
length (Ls). Shear transfer length is also known as overlap length in
bonded patch repair domain. The overlap length is an essential

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

parameter in establishing the patch design principle for repair of

CFRP structures and helpful in sizing of patch. The shear transfer
length estimated here corresponds to the minimum/optimum
overlap length and it is dened as the length over which the
shear stress in adhesive layer decreases from its maximum value
to nearly zero [37]. The zero shear stress zone means no load is
being transferred there and therefore length beyond this point will
add weight penalty rather than improving the strength of repaired
structure. This parameter Ls, can be obtained from shear strain
variation in adhesive layer along bond length extracted at a load
just before the patch debonding initiation i.e., at 68% of failure load
which is shown in Fig. 8. A curve t is obtained for strain prole
and parameter Ls is deduced. It is found that the value of Ls is
approximately 19.3 mm i.e., about 32% of total the patch length.
Rose in 1998 [39] proposed a methodology for the determination
of load transfer length. According to Rose's approach the load
transfer length can be obtained graphically by drawing the tangent
to the shear stress curve at the point of maximum stress (which is
the end point, or very close to it), and nding the intersection with
the x-axis. Hence, the distance between the end point and that
intersection point is the load transfer length. In the present
work, the same approach is also used for the experimental

63

determination of overlap length and it is found to be approximately 12 mm. The discrepancy between the two approaches
could be attributed to the scatters present in the experimental
data. Therefore, it can be concluded that the minimum overlap
length should be approximately 1.52 times the cutout or hole
diameter. Through a parametric study involving FEA, several
authors [4044] have reported the optimum overlap length using
the shear stress distribution and they also fall in the above
recommended range.

5.1.4. Failure mechanism and load displacement behavior

Fig. 9 shows the damage progression in SSPR panel which is
further correlated with its loaddisplacement behavior as shown
in Fig. 10. As the load reaches to point b (see Fig. 10), where the
high peel/shear strain concentration is observed it suddenly
propagates to 80% of adhesive layer with mixed mechanism of
adhesive and cohesive failure reecting the reduction in load at
point c. Delamination initiation (see Fig. 9(d)) could also be seen
at higher load (point d) signifying the low inter-lamina strength
due to high through thickness or peel strain. The growth of
delamination (see Fig. 9(e)) causes reduction in load to point e

Fig. 9. Damage progression with increasing load in SSPR panel.

64

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

and as the load reaches to point f extensive delamination and

debonding are observed leading to nal failure of panel in a
catastrophic manner. The failure surface of the repaired specimen
corresponding to the patch and panel is shown in Fig. 11. It can be
observed from the gure that the nal failure of the panel happens
with complete debonding of the patch. Further, the adhesive layer
is found intact in certain portion of the patch and panel and the
reason could be attributed to the use of strong patch or poor
bonding quality [29]. Broken bers are also seen locally near the
patch overlap edge zone due to excessive shear/peel strain leading
to surface delamination over there.

The result presented above shows global behavior of adhesive

joint in a repaired panel. It provides an estimate of effective load/
shear transfer length or minimum overlap length as it is an
important parameter very useful in patch design. It is identied
that only a small region at patch overlap edge near the root or
corner of adhesive joints experienced a very high peel and shear
strain and the damage is found to initiate from those region. The
global behavior presents a qualitative representation of damage
initiation site. DIC could not capture the exact localized phenomena at the root or corner of adhesive joint responsible for damage
initiation due to lower spatial resolution. Also it does not provide
an insight into the complex mechanism happening there. To
achieve higher magnication probe lens is used which provides
images with higher spatial resolution at the expense of reducing
the region of interest under investigation.
5.2. Localized behavior of strain distribution in Araldite 2011
adhesive layer
In this section localized strain analysis in adhesive layer of a
panel repaired with single sided straight and tapered edge patch is
discussed.

Fig. 10. Loaddisplacement curve for SSPR panel.

5.2.1. Straight edge patch

The longitudinal (xx), peel (zz) and shear strain (xz) distribution
in adhesive layer of single sided patch repaired panel at a load of
23.9 kN are shown in Fig. 12. The ROI for correlation corresponds to a
zone of 2.42 mm  3.18 mm. The spatial resolution is 591 pixel/mm. A
subset size of 71  71 pixels2 with a step size of 7 pixels is chosen for
correlation. It can be observed from the gures that the magnied
optics facilitates the correlation at extreme edge of the patch near the
root/corner of adhesive joint which were not possible in global

Fig. 11. Failure surface corresponding the debonded patch and failed panel.

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

behavior of adhesive joint due to lower spatial resolution as discussed

earlier. Higher spatial resolution would help us in investigating
localized behavior in critical zones responsible for failure.
It is found that the longitudinal strain is highly concentrated at
the entrance of adhesive joint corner at panel/adhesive interface.
Also, localized high peel strain is observed just at entering root of
panel/adhesive interface. High peel strain concentration would
lead to damage initiation from this location leading to patch
debonding. Further investigation shows that the maximum shear
strain concentration also occurs nearer to overlap edge of the
patch at panel/adhesive interface in the adhesive layer. The steep
band of shear strain represents the load transfer happening
between the patch and panel across the adhesive layer. However
global strain analysis could not reveal such strain localization
occurrences. Global strain analysis shows longitudinal strain to be
maximum across panel width near patch overlap edge rather than
localized at panel/adhesive interface. Also, the zone of maximum
peel and shear strain observed in global analysis overlaps at

65

adhesive and patch which is not obvious. However, local strain

analysis reveals peel strain to be concentrated at panel adhesive
interface and steep band of shear strain at adhesive layer as
discussed earlier.
Fig. 13 shows damage monitoring in adhesive layer based on peel
strain distribution with increasing load. The peel strain is found
maximum (white line) at panel/adhesive interface (see Fig. 13(b)).
With increasing load the maximum peel strain shifted to the
adhesive/patch interface (see Fig. 13(c)) and along the bondline
(black line). The zone of maximum peel strain then advances along
the bondline at patch/adhesive interface (see Fig. 13(d)) with further
increase in load. On close observation one can nd that a zone of
compressive strain is also present adjacent to the maximum peel
strain zone (red line) and it gets intensied similar to the peel strain
with increasing load. The damage initiation could be seen in the form
of discontinuity in correlation in adhesive layer (see Fig. 13(e)) once
the peel strain reaches to critical value. It can also be noticed that the
damage propagates at the adhesive/patch interface along maximum

Fig. 12. Localized whole eld strain distribution in adhesive layer for a panel repaired with straight edge patch at a tensile load of 23.9 kN: (a) xx, (b) zz and (c) xz.

Fig. 13. Damage monitoring in adhesive layer with increasing load based on peel strainzz. (For interpretation of the references to color in this gure, the reader is referred
to the web version of this article.)

66

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

peel strain path. These observations could only be possible due to

localized strain analysis rather than global strain analysis.

5.2.2. Tapered edge patch

The inuence of patch edge tapering on strain level using DIC is
investigated in this sub-section. The line diagram presenting the
geometry of taper made on the joint at the patch edge is shown in
Fig. 14 and the angle achieved on the real joint at the patch edge
(zoomed up view) is also depicted in Fig. 15 for more clarity. The
edge of the patch is provided with a smooth tapering having a
taper angle, 301. To achieve the taper, initially machining is
done to obtain right angle at the patch edges. The angle depicted
in Fig. 14 is carefully marked at the patch edge and then manual
polishing is exercised with due care using a ne grit emery paperP600. To achieve the specied angle at the patch edges, continuous
measurement is made at an interval of 23 passes during polishing
for its accuracy. The area under investigation is then applied with
white acrylic paint followed by black dots to generate random

speckle pattern. On close observation (see Fig. 15), white paint

residue could be seen near the patch/panel edge as well as near
the step of the joint which are left un-cleaned because its cleaning
may hamper the speckle pattern. The presence of paint residue at
edges does not affect the DIC measurement. The ROI for correlation corresponds to a zone of 2.29 mm  3.68 mm. The spatial
resolution is 511 pixel/mm. A subset size of 71  71 pixels2 with a
step size of 7 pixels is chosen for correlation. The longitudinal
(xx), peel (zz) and shear strain (xz) distribution at a load of
23.9 kN are shown in Fig. 16. It is found that the strains are highly
concentrated near the entrance of adhesive joint at panel/adhesive
interface and there is no signicant inuence of patch edge
tapering on longitudinal and shears strain level in the adhesive
layer as compared to the peel strain. The variation of longitudinal,
peel and shear strain in adhesive layer along the bond length close
to patch overlap edge at a load of 23.9 kN is shown in Fig. 17. The
line considered for the plot is shown there. It is found that the
strain decreases with increasing bond length away from patch
overlap edge and also the shear strain is more dominant in
adhesive layer followed by longitudinal and then peel strain in
SSPR panel conguration. This is also true for the panel repaired
with straight edge patch (see Fig. 12). Fig. 18 shows a comparative
plot of peel strain variation along the mid of adhesive layer
considered near the patch overlap edge between straight and
tapered edge patch repaired panel. Due care is taken to extract the
DIC data at the same location and over the same distance while
comparing. The variation of peel strain in adhesive layer between
the two cases is found similar except that the magnitude of peel

Fig. 14. Geometry details of patch edge tapering.

Fig. 15. Actual joint depicting the angle made to achieve the tapered patch edge
(zoomed up view).

Fig. 17. Comparative plot of strain variation in adhesive layer for the panel repaired
with tapered edge patch obtained from DIC at a tensile load of 23.9 kN.

Fig. 16. Localized whole eld strain distribution in adhesive layer for a panel repaired with tapered edge patch at a tensile load of 23.9 kN: (a) xx (b) zz and (c) xz.

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

strain shifted up in case of the panel repaired with tapered edge

patch. The peel stresses/strains in adhesive joints with composites
can be reduced by some of the techniques proposed by da Silva
and Adams [45]. The magnitude of longitudinal, peel and shear
strain is increased from 0.0106, 0.0053 and 0.0118 to 0.0134,
0.0077 and 0.0157 respectively for the case of panel repaired with
tapered edge patch. Similar observation has been found in Ref. [19]
in which steel panel bonded with CFRP composite laminate on
both side is studied. It is important to note here that the taper
length is lower than the experimental prediction thereby resulting
in increased peel strain values at the patch overlap edge. Therefore, one can conclude that any tapering length/ratio may not be
effective in reducing the strain level and a thorough investigation
in terms of tapering parameter needs to be done for appropriate
reduction in peel stress values.

6. Numerical results
In this section the results obtained from FEA for a panel
repaired with tapered edge patch bonded with Araldite 2011
adhesive are presented and further compared with DIC results.
Finite element results are presented by assuming linear elastic

Fig. 18. Comparative plot of peel strain variation in adhesive layer between panels
repaired with straight and tapered edge patch obtained from DIC at a tensile load of
23.9 kN.

67

nature of adhesive material. Firstly, a parametric study is carried

out to study the effect of number of elements in adhesive along
thickness direction and along the adhesive length near the patch
edge on peel and shear strain. It is found that the number of
elements through the thickness of adhesive has negligible inuence on peel and shear strain whereas the renement of elements
in adhesive along its length near the patch edge has signicant
inuence on peel and shear strain showing a corner singularity
behavior. Therefore, the meshing in adhesive near the patch edge
is chosen based on recommendation made in Ref. [8,46]. Similar
meshing pattern is adopted in the patch and panel.
Fig. 19 shows the strain distribution in the adhesive layer
(Araldite 2011) obtained from FEA at a load of 23.9 kN. The FEA
results are presented with adjusted scale to match against DIC
scale. It can be observed from the gure that a high peel and shear
strain concentration exist close to the patch overlap edge at panel/
adhesive interface. Also a zone of compressive strain could be
identied (see Fig. 19(a)) just below the patch overlap edge due to
high peel strain as discussed earlier. The strain distribution
obtained from FEA is found to be consistent with experimental
observation as presented earlier.
6.1. Shear transfer length (Ls) from FEA
In this sub-section a procedure is presented to predict the shear
transfer length from FEA study and it is also compared with the
experimental results presented earlier. To estimate the shear
transfer length analysis is carried out at patch debonding load (in
this case 68% of ultimate load) to obtain the shear strain distribution
in the adhesive layer. The shear strain distribution in the adhesive
layer of the panel repaired with straight edge patch is shown
in Fig. 20 for a load 46.8 kN and it is further compared with the one
obtained from DIC. Shear transfer length is the length over which
the shear strain in adhesive layer reduces from its maximum value
to nearly zero as dened earlier. The value of this length is found to
be 20.2 mm from FEA which is slightly higher than the one
estimated from DIC. Further, the shear strain variation obtained
from FEA and DIC presents certain discrepancies. This deviation
could be because of difference in the location of line plot extracted
from FEA and DIC especially due to unavoidable bending in SSPR
conguration. The other possible reason for the observed discrepancy could be due to the use of bulk adhesive modulus (as
presented in Table 1) in FE simulation rather than in-situ or
effective modulus of the adhesive existing between patch and panel
in experiment. The effective modulus of the adhesive would be

Fig. 19. Whole eld strain distribution in adhesive layer at a tensile load of 23.9 kN obtained from FEA: (a) peel strainzz and (b) shear strainxz.

68

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

Fig. 20. Shear strain variation along adhesive length of a panel repaired with
straight edge patch at 46.8 kN: DIC vs. FEA.

different than bulk adhesive modulus [17]. Therefore, the effective

modulus of the adhesive existing between the patch and panel in
experiment is evaluated based on full eld data extracted locally
within the adhesive zone and the same effective modulus value of
1.25 GPa is then used in FE simulation. The comparative plot of
shear strain variation in the middle of adhesive layer obtained from
experiment and FE simulation carried out using effective adhesive
modulus is shown in Fig. 20. It can be observed from the gure that
FE simulation using effective adhesive modulus shows a much
closer trend to experimental one with a minute discrepancy
between them. Further improvement between the two results
could be achieved by performing FE simulation taking into account
the elasticplastic behavior of the adhesive material. Therefore, on
over all bases one can conclude that there is a close agreement
between the DIC and FEA results besides the difference in magnitude at extreme patch edge which is mainly due to the fact that the
DIC software could not correlate exactly at the edge boundary
(see Fig. 6) due to non-availability of adjacent subset or pixels
required for correlation.

7. Conclusions
In the present work, critical strain analysis in thin adhesive
layer of a single sided patch repaired CFRP panel under tensile load
is carried out using the DIC technique. A methodology is presented
to successfully demonstrate the full eld through the thickness
strain measurement in such a thin adhesive layer by employing
the DIC technique. Both global and local analysis approaches are
executed for analyzing longitudinal, peel and shear strains over
the adhesive layer and they are found to be maximum near the
root of adhesive joint at panel/adhesive interface. Shear strain in
adhesive layer is found to be signicantly high compared to peel
strain. Based on global strain analysis it is found that only a small
zone near the patch edge facilitates the load transfer from panel to
patch through the adhesive layer by the shear deformation
mechanism. The shear transfer length which is an essential
parameter in arriving at an appropriate overlap length in patch
design is estimated based on the shear strain eld in adhesive
layer and it matches closely with the FEA prediction. It is
recommended to have minimum/optimum overlap length in the
range of 1.52 times the cutout diameter, which is found consistent with the data available in the literature on comparison.
Conventional DIC along with magnied optics is capable of exactly
capturing the localized complex strain eld at the root of adhesive

joint. Also damage initiation and progression in adhesive layer are

successfully monitored by capturing localized strain eld with
increasing load. The failure in the panel initiates from high peel/
shear strain concentration zone at the patch overlap edge in the
form of patch debonding and then it propagates with the mixed
mechanism of adhesive and cohesive failure followed by extensive
surface delamination over the panel. The global and local strain
analyses are complementary to each other and help in understanding the complex strain eld over the adhesive layer. It is also
found that the panel repaired with tapered edge patch shows
more strains in adhesive layer near the patch overlap edge in
comparison to the one repaired with a straight edge patch.
Therefore, normal tapering may not be effective in reducing the
strain level and a thorough investigation needs to be done for
arriving at an optimum taper ratio for getting reduction in strain
level. The DIC is found to be suitable for analyzing the localized
and global strain eld over small but critical locations and helps in
monitoring the damage based on developed strain anomalies. The
FEA prediction shows a good coherence with the DIC results
thereby conrming the accuracy of the DIC technique for adhesive
strain measurement in repair domain.

References
[1] Baker AA, Rose LRF, Jones R. Advances in the bonded composite repair of
metallic aircraft structure. Killington: Elsevier Publications; 2002.
[2] Davis M, Bond D. Principles and practices of adhesive bonded structural joints
and repairs. Int J Adhes Adhes 1999;19(23):91105.
[3] Whittingham B, Bakerb AA, Harman A, Bitton D. Micrographic studies on
adhesively bonded scarf repairs to thick composite aircraft structure. Compos
Part A: Appl Sci Manuf 2009;40(9):141932.
[4] Katnam KB, Da Silva LFM, Young TM. Bonded repair of composite aircraft
structures: a review of scientic challenges and opportunities. Prog Aerosp Sci
2013;61:2641.
[5] Banea MD, da Silva LFM. Adhesively bonded joints in composite materials: an
overview. Proc Inst Mech Eng Part L: J Mater Des Appl 2009;223:1. http://dx.
doi.org/10.1243/14644207jmda219.
[6] da Silva LFM, das Neves PJC, Adams RD, Spelt JK. Analytical models of
adhesively bonded joints Part I: literature survey. Int J Adhes Adhes
2009;29:31930.
[7] da Silva LFM, das Neves PJC, Adams RD, Wang A, Spelt JK. Analytical models of
adhesively bonded joints Part II: comparative study. Int J Adhes Adhes
2009;29:33141.
[8] da Silva LFM, Campilho RDSG. Advances in numerical modeling of adhesive
joints. Berlin, Heidelberg: Springer; 2012.
[9] He X. A review of nite element analysis of adhesively bonded joints. Int J
Adhes Adhes 2011;31:24864.
[10] Tsai MY, Morton J. An experimental investigation of nonlinear deformations in
single lap joints. Mech Mater 1995;20(3):18394.
[11] Shenoy V, Ashcroft IA, Critchlow GW, Crocombe AD, Wahab MMA. An
investigation into the crack initiation and propagation behaviour of bonded
single-lap joints using backface strain. Int J Adhes Adhes 2009;29(4):36171.
[12] Derewonko A, Godzimirski J, Kosiuczenko K, Niezgoda T, Kiczko A. Strength
assessment of adhesive-bonded joints. Comput Mater Sci 2008;43(1):15764.
[13] Asundi A. Deformation in adhesive joints using moir interferometry. Int J
Adhes Adhes 1987;7(1):3942.
[14] Schroeder JA. Photoelastic stress analysis of bonded lap shear joints having
thermoplastic adherends. J Adhes 1990;32:23.
[15] Colavito KW, Das M, Hahs D, Gorman J, Madenci E, Smeltzer S. Digital image
correlation technique to extract adhesive strains in lap joints. In: Proceedings
of the 49th AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics, and
materials conference; 2008.
[16] Colavito KW, Gorman J, Madenci E, Smeltzer S. Renements in digital image
correlation technique to extract adhesive strains in lap joints. In: Proceedings
of the 50th AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics, and
materials conference; 2009.
[17] Moutrille MP, Derrien K, Baptiste D, Balandraud X, Grdiac M. Throughthickness strain eld measurement in a composite/aluminium adhesive joint.
Compos Part A 2009;40:98596.
[18] Wang ZY, Wang L, Guo W, Deng H, Tong JW, Aymerich F. An investigation on
strain/stress distribution around the overlap end of laminated composite
single lap joints. Compos Struct 2009;89(4):58995.
[19] Haghani R, Al-Emrani M and Kliger R. Effects of geometrical modications on
behaviour of adhesive joints used to bond CFRP laminates to steel members
experimental investigation. In: Proceedings of the Nordic steel construction
conference (NSCC). Malmo (Sweden); 2009.

M. Kashfuddoja, M. Ramji / International Journal of Adhesion & Adhesives 57 (2015) 5769

[20] Guo BQ, Xie HM, Zhu JG, Wang HX. Study on the mechanical behavior of
adhesive interface by digital image correlation. Sci China: Phys Mech Astron
2011;54(4):57480.
[21] Ruiz PD, Jumbo F, Huntley JM, Ashcroft IA, Swallowe GM. Experimental and
numerical investigation of strain distributions within the adhesive layer in
bonded joints. J Strain 2011;47:88104.
[22] Comer AJ, Katnam KB, Stanley WF, Young TM. Characterising the behaviour of
composite single lap bonded joints using digital image correlation. Int J Adhes
Adhes 2013;40:21523.
[23] Katnam KB, Dhte JX, Young TM. Experimental analysis of the bondline stress
concentrations to characterize the inuence of adhesive ductility on the
composite single lap joint strength. J Adhes 2013;89:6.
[24] Crammond G, Boyd SW, Dulieu-Barton JM. Through-thickness load transfer in
adhesively bonded composite joints. Conf Proc Soc Exp Mech Ser 2013:1114.
[25] Kumar RLV, Bhat MR, Murthy CRL. Experimental analysis of composite single-lap
joints using digital image correlation and comparison with theoretical models. J
Reinf Plast Compos 2013. http://dx.doi.org/10.1177/0731684413500859.
[26] Sutton MA, Orteu JJ, Schreier H. Image correlation for deformation and shape
measurements. New York: Springer; 2009.
[27] Cognard JY, Creachcadec R, da Silva LFM, Teixeira FG, Davies P, Peleau M.
Experimental analysis of the inuence of hydrostatic stress on the behaviour
of an adhesive using a pressure vessel. J Adhes 2011;87:80425.
[28] Kashfuddoja M, Ramji M. Whole-eld strain analysis and damage assessment
of adhesively bonded patch repair of CFRP laminates using 3D-DIC and FEA.
Compos Part B 2013;53:4661.
[29] Caminero MA, Pavlopoulou S, Lopez-Pedrosa M, Nicolaisson BG, Pinna C,
Soutis C. Analysis of adhesively bonded repairs in composites: damage
detection and prognosis. Compos Struct 2013;95:50017.
[30] Caminero MA, Lopez-Pedrosa M, Pinna C, Soutis C. Damage monitoring and
analysis of composite laminates with an open hole and adhesively bonded
repairs using digital image correlation. Compos Part B: Eng 2013;53:7691.
[31] Kashfuddoja M, Ramji M. Adhesive strain measurement in patch repaired
CFRP laminate using 2D DIC. In: Proceedings of the 19th international
conference on composite materials (ICCM - 19). Montreal (Canada); July 28
August 2, 2013.
[32] Kashfuddoja M, Ramji M. Critical analysis of adhesive layer behaviour in patch
repaired CFRP panel involving digital image correlation. J Compos Mater 2014,
http://dxdoi.org/10.1177/0021998314541312.
[33] Ubaid J, Kashfuddoja M, Ramji M. Strength prediction and progressive failure
analysis of carbon ber reinforced polymer laminate with multiple interacting

[34]

[35]

[36]

[37]
[38]

[39]

[40]
[41]

[42]
[43]

[44]
[45]
[46]
[47]

69

holes involving three dimensional nite element analysis and digital image
correlation. Int J Damage Mech 2013. http://dx.doi.org/10.1177/1056789513504123.
Bhise VS, Kashfuddoja M, Ramji M. Optimization of circular composite patch
reinforcement on damaged carbon ber reinforced polymer laminate involving both mechanics-based and genetic algorithm in conjunction with 3D
nite element analysis. J Compos Mater 2013. http://dx.doi.org/10.1177/
0021998313501922.
Kashfuddoja M, Ramji M. An experimental and numerical investigation of
progressive damage analysis in bonded patch repaired CFRP laminates. J
Compos Mater 2014. http://dx.doi.org/10.1177/0021998314521058.
Tsai MY, Morton J. An investigation into the stresses in double-lap adhesive
joints with laminated composite adherends. Int J Solids Struct 2010;47:
331725.
Duong CN, Wang CH. Composite repair, theory and design. 1st ed.. The
Netherlands: Elsevier Publications; 2007.
Hosseini-Toudeshky H, Saber M, Mohammadi B. Mixed-mode 3-D crack
propagation of repaired thin aluminium panels using single-side composite
patches. J Fract Mech 2008;153:10516.
Rose LRF. Theoretical analysis of crack patching. In: Baker AA, Jones R, editors.
Bonded repair of aircraft structure. Dordrecht, The Netherlands: Martinus
Nijhoff Publishers; 1988. p. 77105.
Soutis C, Hu FZ. Design and performance of bonded patch repairs of composite
structure. Proc Inst Mech Eng Part G 1997:211.
Campilho RDSG, de Moura MFSF, Domingues JJMS. Modelling single and
double-lap repairs on composites materials. Compos Sci Technol 2005;65:
194858.
Hu FZ, Soutis C. Strength prediction of patch-repaired CFRP laminates loaded
in compression. Compos Sci Technol 2000;60:110314.
Soutis C, Duan DM, Goutas P. Compressive behaviour of CFRP laminates
repaired with adhesively bonded external patches. Compos Struct 1999;45:
289301.
Liu X, Wang G. Progressive failure analysis of bonded composite repairs.
Compos Struct 2007;81:33140.
da Silva LFM, Adams RD. Techniques to reduce peel stresses in adhesive joints
with composites. Int J Adhes Adhes 2007;27:22735.
Kumar S, Mittal KL. Advances in modeling and design of adhesively bonded
systems. New York, United States: Wiley-Scrivener; 2013.
Mallick P. Fiber reinforced composites: materials, manufacturing, and design.
3rd ed. New York, United States: CRC Press; 2007.