Composites Part B 77 (2015) 129e138

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

End geometry and pin-hole effects on axially loaded adhesively

bonded composite joints
Ibrahim Fadil Soykok*
Celal Bayar University, Hasan Ferdi Turgutlu Faculty of Technology, Department of Mechatronics Engineering, 45400 Turgutlu, Manisa, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: An experimental investigation was performed to analyze the potential impacts of varying joint region
Received 21 July 2014 geometries and adhesive filled pin holes on adhesively bonded composite structures. Tapers, especially
Received in revised form half-length ones are observed to provide an anticipated progress in single lap joints. Besides, scarf joints
8 January 2015
with aligned adherends in the same plane exhibited enhanced stiffness and strength in consideration of
Accepted 9 March 2015
single lap joints. In terms of the stiffness and strength, thickening of adherends was also found to be
Available online 17 March 2015
impressively efficient on composite single lap joints as well as scarf joints. Contrary to the expectation of
that the hardened adhesive previously filled into the holes during adhesion would create a pin effect in
Keywords:
A. Glass fibers
load bearing, holey specimens exhibited poor performance and induced degradation in joint quality.
B. Adhesion © 2015 Elsevier Ltd. All rights reserved.
C. Damage mechanics
E. Joints/Joining

1. Introduction uncured prepreg. This test device was designed to simulate the
automated tape lay-up (ATL) and automated fiber placement (AFP).
As a typical application of composites, fiber reinforced plastics The study gave consistent results with a standard deviation of 16%
(FRP) have been commonly used in many fields of industry but the results including stiffness and tack values were found to be
including aviation, automotive, marine and wind energy structures. inconsistent with the levels specified by manufacturers. Kaufmann
This type of composites has been produced using different et al. [4] proposed a novel optimization framework for hand-laid
manufacturing techniques depending on the areas of application. carbon fiber prepregs by extending an existing one. Implement-
One of the most popular ways of manufacturing FRP is to use ing of a draping simulation tool an optimal draping strategy in
prepregs. Prepregs are made of reinforcements pre-impregnated by terms of seed point and reference angle was selected.
matrix material in liquid phase. After excess resin is fully removed, During the current experimental work, composites are pro-
it is subjected to a partial curing and becomes ready to use as a duced with Vacuum Assisted Resin Transfer Molding (VARTM)
tacky solid flexible layer. Applying both temperature and pressure method, which is another preferred way to obtain FRP. Despite
for a specific period of time is enough to provide a complete offering some advantages such as shorter start-up time, and lower
hardening during composite production. Many studies concerning tooling cost, VARTM process has also special challenges. In fact,
this production technique have been included in the literature. impregnation of the liquid resin to reinforcing fibers under vacuum
Jiang and Huang [1] tested the effectiveness of the Near Infrared is a time-consuming process especially when dealing with large-
Spectroscopy (NIR) analysis on non-contact quality inspection of scale structures. Hence, a new high-speed VARTM process was
laid fabric prepreg and found the method sufficiently rapid and employed by Han et al. [5] to reduce impregnation time by
reliable. In order to introduce a noncontact analytical method to changing resin flow direction from in-plane to through thickness
improve prepreg quality, Jiang and Huang [2] have used also NIR for the perform. Surface properties of fibers are key factors con-
combined with partial least squares (PLS) for analyzing the trolling flow characteristics of polyester resin in VARTM processes.
impregnation degree of prepreg. Crossley et al. [3] developed a new Glass fiber surfaces were modified by using g-Methacryloxypropyl
peel test which quantifies the tack and dynamic stiffness of trimethoxy silane (g-MPS) in a study of Lee et al. [6]. Lower surface
energy of glass plate was measured due to g-MPS treatment onto
the glass fiber surface. The flexural strength and modulus of the
* Tel.: þ90 236 314 10 10; fax: þ90 236 314 20 20. cured glass fiber/UPE composites have improved in comparison to
E-mail address: [email protected]. those of untreated ones. g-MPS treatment have also reduced the

http://dx.doi.org/10.1016/j.compositesb.2015.03.031
1359-8368/© 2015 Elsevier Ltd. All rights reserved.
130 I.F. Soykok / Composites Part B 77 (2015) 129e138

void content of glass fiber/unsaturated polyester (UPE) composite. a simplified one dimensional model to investigate single and
Zhang et al. [7] reported the results of using quick and preheating double lap joints with different type adherend materials. Maximum
processes to fabricate carbon fiber reinforced epoxy resin com- shear stresses were calculated at the free end and nearer to the
posites by means of VARTM. Preheating process was found to be a material with higher stiffness. Mattos et al. [16] developed a
suitable method to improve the processing efficiency in terms of phenomenological model for performing failure analysis of com-
the total processing time and flexural strength of composite. posite adhesive single lap joints with arbitrary glued area. They
In constructing of composite structures several parts need to be identified a shape factor in order to correlate between rupture
joined together using an appropriate technique. Amongst, adhe- forces of joints having different glued areas.
sively bonding has become a widely used one in laminated com- Failure of bonded joints can be either adhesive or cohesive in
posites, due to its superior advantages in comparison with nature. Environmental conditions and stresses degrade the inter-
mechanically fastening. Considering negative impacts of drilling face as well as bulk of the adhesive in the bond line [12]. For
holes, aero-aqua dynamic inconvenience and weight penalty in characterization failure behaviors of joints with different (brittle
bolted or pinned joints, bonding plates provides a more homoge- and ductile) adhesives and different overlap lengths Neto et al. [17]
nous stress distribution, light weight, higher fatigue resistance and carried out experimental tests. Ductile adhesive (SikaForces 7888)
ability to join dissimilar materials. However, it should not be has given cohesive failure results for all overlap lengths, while
ignored some points in design process of these type connections. inter-laminar composite failure was observed for overlap lengths
For example, peek shear and peel stresses arising typically near the up to 30 mm as in case of brittle adhesive (AV138).
interface between adherends and adhesive layer, because of The main problem is the low transverse tensile strength, which
discontinuous geometry at the ends of the joint region, still stands is of the same order or lower than that of the matrix. Bonded joints
as challenge in front of the designers. Besides, small flaws occurring experience peel loading so the composite undergo transverse ten-
during service life or production are difficult to inspect and the load sion before the adhesive fails. Like the shear stresses, the peel
transmission capacity of joint may depend on several factors such stresses peak at the ends of the overlap and can induce composite
as adhesive type (brittle or ductile), surface roughness and also failures, due to the low transverse (through the thickness) tensile
thickness of adhesive layer. strength. The inner composite adherend splits apart locally due to
Different methods have been utilized in order to detect joint these peel stresses, thereby destroying the shear transfer capacity
properties for special loading types. Balzani et al. [8] characterized between the inner and the outer adherends [18]. Some authors
fracture toughness of a film adhesive in conditions of different tried to design joints that produce reduced peel stresses to tackle
mode loading types and tried to fit suitable criteria for calculation this problem. Silva and Adams [18] used an internal taper and ad-
critical energy release rates for any given mode mixing ratio. In a hesive fillet to observe the increases in joint strength. However, the
study of double-lap unidirectional and quasi-isotropic adhesive technique they employed was not useful when the thermal stresses
joints, Tsai and Morton [9] used moire interferometry for experi- are high enough especially at low temperatures. One of the findings
mental investigations in addition to numerical finite element and obtained by Kilic et al. [19] from their numerical study utilizing a
analytical one-dimensional closed-form solutions. Adherend shear global element coupled with traditional elements in finite element
deformations were found to be significant in the laminated double method was that adhesive overflow had a significant impact on the
lap composite joints, which is not taken usually into account in peel stress reduction in single-lap joints. Lee et al. [20] tested
theoretical modeling. Kumar et al. [10] studied composite adhesive double-strap and supported single lap joints in an experimental
lap joints for evaluation of kissing bond regions using digital image study. According to the data obtained from a set of strain gauges
correlation (DIC) method. The effectiveness of digital image cor- attached on the joint specimens, peel stresses exist in supported
relation technique in detection of dry contact kissing bonds was singe lap joints over than those in double strap joints. Elastic-
tried to find out. This noncontact technique was able to indicate the plastic stress analysis was conducted by Sayman et al. [21] in fi-
existence of kissing bonds at less than 50% of failure load. The ber glass-epoxy adherends with DP460 ductile adhesive. Analytical
propagation of fundamental Lamb Mode through an adhesive joint and multi-linear finite element results proved that when the
between GFRP composite laminate and aluminum was studied by thickness of the adherend is increased, the intensity of the peel
Hoskoti et al. [11] numerically in FEM. Generating the waves at the stress raises.
adhesive joint, interaction of lamb wave with adhesive were The authors many of whom focused on adhesively bonding in
observed by placing the sensors at the upper surface of the model. composite joints have concerned also with the topic of peeling
Finally, reflection factor and transmission factor were observed to stresses; however, studies aiming the solution have remained very
be dependent on the adhesive orientation and the excitation limited. In the present experimental investigation, this undesired
frequency. loading case is intended to eliminate geometrically by using
Some studies were devoted to the investigation of composite- different joining techniques. Among many adhesively bonded joint
eadhesive interface properties and the determination of influ- applications, special geometries may constitute effective solutions
encing factors in composite joints. Kumar et al. [12] characterized to minimize peel stress effects and thereby contribute in-plane
interfacial imperfections using ultrasonic signatures obtained from tensile strengths. In this context, tapered and scarf joints can be
experimental data. Mahoney et al. [13] studied numerically and considered as an option to simple lap joints for increasing
performed Taguchi analysis to assess the sensitivity of the finite maximum failure loads via peel stress reductions. Therefore, both
element model. The interface fracture energy and the adhesive of the joint geometries were employed in our present experimental
ductility are found to play a key role in controlling the static failure work and failure responses were observed.
load. On the other hand, peel stresses, which intensify at the free ends
There are also available studies dealing with some special issues of bond region, arise due to a bending moment resulting from
such as impact effects, different adherend or adhesive material and discontinuous structure of joint region in conducting axial loads.
bonding area. In a study by Sayman et al. [14] joints were exposed Because, stiffening the adherends may be here another beneficial
to transverse impacts of different energy levels prior to the ex- practice for alleviating peel stresses, some specimens were sup-
periments at different temperature conditions. Decreases in load ported by increasing the thickness before assembly.
capacity observed as a result of high and low temperature condition In an effort to observe whether the adhesive material can act as
tests and higher impact energy applications. Her SC [15] employed a pin fastener and contribute load-bearing, some specimens were
I.F. Soykok / Composites Part B 77 (2015) 129e138 131

Table 1
Mechanical properties of the glass fiber/epoxy composite material.

E1 (MPa) E2 (MPa) n12 Xt (MPa) Yt (MPa) Xc (MPa) Yc (MPa) S (MPa) Tg ( C)

27,200 12,800 0.25 586.5 ± 9.16 133.7 ± 2.62 324.7 ± 11.10 200.2 ± 5.64 61.96 ± 0.76 85

priorly drilled from contact surfaces, thoroughly and the holes were Resin inlet and vacuum links were connected with the bag by
filled with adhesive material during bonding process. using sealing paste. Both link were made of flexible hoses with
In order to design safe and efficient adhesively bonded com- valves located at the ends to control resin and air passage.
posite joint structures, allowable stresses in tension and shear are In order to ensure that the vacuum bag is fully air proof, the inlet
required. The controlling stress type on single-lap joint failures is valve was closed at first and the air in the bag was evacuated under
usually not the shear stresses, but tensile stresses are the key vacuum. The outlet valve remained open until the air was fully
determinant. One of the most effective parameter controlling the discharged. In order to check any leakage, the bag was left standing
tensile stress and the apparent shear strength is the geometric for half an hour while both valves are closed. Because any air pas-
structure of the samples. Therefore, the geometric variables sage into the bag during resin impregnation may cause the
investigated in the current study are crucial especially for tensile manufacturing process to be interrupted and repeated.
loading in adhesively bonded joint design. Prior to the resin impregnation, the final stage of production, the
With the knowledge given in this paper, the designer has an table was started to heat gradually up to 120 C, which is the given
opportunity to make a choice between geometrical preferences curing temperature of resin matrix. By the way, the mixture of resin
which are useful methods to avoid structural weaknesses associ- and fast hardener was led in to the bag by turning on resin inlet and
ated with shear and peel stress concentrations in lap region. For a vacuum valves. The mixture was penetrated gradually into the fiber
specific adhesive and adherend material, the reader is assisted to fabrics and then the valves were turned off after a full impregnation
select the most proper joint design. was assured. The product was left on the table at a constant tem-
perature value of 120 C for 2 h which is the essential time for resin
curing. At the end of this period, the composite plate was
2. Experimental procedure dismantled from the table before removing upper and lower
separator layers.
2.1. Production of composite plates
2.2. Determination of mechanical properties
The composite material of which joins tested during the current
investigation was produced using vacuum assisted resin infusion Mechanical properties related to the manufactured composite
technique. The production was carried out on a heater table, which material were obtained by conducting a series of experiments.
provides matrix material the required curing temperature and Compressive, tensile and rail-shear tests were implemented for
gives composite product a plane shape. The table consists of a metal finding material constants which are listed in Table 1. The data
plate and heating resistances under the entire surface of the plate. It given in the table are the longitudinal modulus E1, the transverse
is also equipped with a control panel to be able to monitor and set modulus E2, the longitudinal tensile strength Xt, transverse tensile
instantaneous temperature as well as the vacuum pressure, strength Yt, the longitudinal compressive strength Xc, transverse
generated by a pump assembled with the table body. compressive strength Yc, the rail shear strength S, the Poisson's
For preparation of reinforcing material, 0/90 woven and 45/45 ratio v12, and the glass transition temperature Tg
fabric glass fibers were first cut in size of 1 m 1 m in suitable
orientations, compatible with [0/90/45/45] s composite plate. A 2.3. Adherend preparation and joint formation
release film, glass fibers and a peel ply were spread out on the
heater table, respectively. A resin distribution facilitator network To form the adherends, the composite plate having a fiber
was laid then on the top of the layers to provide quick and ho- orientation of [0/90/45/45] s was prepared by cutting into speci-
mogenous resin dispersion. All these layers were then wrapped fied dimensions according to the standard test method, (ASTM
hermetically with a vacuum bagging film by taping the film from D5868 e 01) [22] for adhesive bonding in fiber reinforced plastics
edges on to the heater table. (FRP). After dimensioning specimens as illustrated in Fig. 1, their

Fig. 1. Simple single lap joint of composite test specimen with dimensions.
132 I.F. Soykok / Composites Part B 77 (2015) 129e138

joint regions were shaped by abrasive machining method according Specimens were primarily surface roughened and cleaned before
to the geometric parameters tested in this investigation. The test joint formation according to ASTM D2093 [23]. Loctite epoxy adhe-
samples were formed for different geometric characteristics, sive was employed to create single lap joints between glass fiber/
named as simple single lap (Fig. 2a), holey (Fig. 2c, d, g), tapered epoxy composite adherends having bonding zones with various
(Fig. 2i), half tapered (Fig. 2h), and scarf (Fig. 2f, g) joint geometries. geometric features. The properties of the adhesive material were
Three type holey samples were produced to investigate the pin- given in Table 2 While assembling adherends after spreading the
hole effects. The samples with single central hole (Fig. 2c), two adhesive on bonding surface in the form of a thin film layer, the
longitudinal holes (Fig. 2d) and two crosswise holes (Fig. 2e) were thickness of the bond-line was set to 0.75 mm by using a thicker plate
prepared. The center of single hole was located in the middle of the under the free end of the upper adherend and by applying a constant
bonding surface, and 12.5 mm away from both the side edges. Two pressure on the entire surface. In order that adhesive could gain its full
longitudinal holes were also equidistant to the sample side edges strength, the joints were kept in standard atmospheric conditions for
and the distance between the centers was 15 mm. The center of the 7 days period before testing, according to the product data sheet.
first hole was 5 mm away from the end edge. Two crosswise holes
were placed in the diagonal line of the square bonding area so as to 2.4. In-plane loading tensile tests
divide it into three equal parts.
In addition to those mentioned, some specimens were doubled In order to get ultimate tensile strengths, failure loads and
in thickness size by sticking two equivalent samples using epoxy loadedisplacement (Force-Disp.) characteristics, samples of each of
adhesive which is the same as that used to create joints (Fig. 2b,g). joint geometry were exposed to in-plane tensile loading until a
Single lap joints and scarf joints were formed with these double complete failure occurred. The experiments performed in Shi-
thickness adherends to compare them with simple single lap joints. madzu AG-100 testing device with a capacity of 100 KN. Prior to

Fig. 2. CAD images and photos (front and side) of joint samples: a) Simple sing. lap, b) Simp. sing. lap double thickness, c) Single hole, d) Double longitudinal hole, e) Double
crosswise hole, f) Scarf, g) Scarf double thickness, h) Half tapered, i) Tapered joints.
I.F. Soykok / Composites Part B 77 (2015) 129e138 133

Table 2
Typical properties of adhesive material (Loctite Hysol-9466).

Peel strength/unit width Max.lap shear strength Tensile strength Tensile Modulus Elongation Shore Hardness Glass Trans. Temp.
(on steel) (N/mm) “ISO 4587” (N/mm2) “ISO 527-3” (Mpa) “ISO 527-3” (Mpa) “ISO 527-3” % “ISO 868” (Durometer D) “ASTM E 1640” ( C)

8 37 32 1.718 3 60 62

fixing the samples between the wedge grips, doublers with the
same material and same thickness as composite plate were stuck to
the free ends of specimens where they are clamped and axially
loaded. By supplementing the material this way, it was intended to
minimize bending effect caused by the asymmetric nature of single
lap joint structure. A crosshead speed of 1 mm/min was set in
Trapezium software which is connected to the testing device and
synchronously performs data acquisition and processing.

3. Results and discussion

Adhesively bonded, glass fiber/epoxy composite to composite lap

joints with different geometric features were analyzed comparatively
during the current experimental work. Through applying different
geometries in joining adherends, it was principally aimed to develop
the values of joint strengths by enabling reduced peel stress forma-
tions in the free ends of bonding regions. In line with this purpose,
analyses were performed for half and full-length tapered joints, scarf
joints, and also joints of double thickness adherends. The behavior of
joints with a central, two longitudinal and two crosswise holes was
another concern dealt during the current study. The objective of this
second type experimental process was to observe whether the cured
adhesive, filled during bonding adherends into the priorly drilled Fig. 4. Effects of adherend thickness in scarf joints on loadedisplacement curves.
holes, may create a pin effect on joints by making a contribution in
resisting axial tensile forces.

3.1. Effect of adherend thickness on the loadedisplacement either event, improvements are evident not only in joint stiffness
response but also in failure strengths of joints associated with increased
thickness of adherends. Double thickness adherends make a change
Fig. 3 and 4 show the effect of adherend thickness on joint in joint stiffness of simple single laps at a level of 76.96%, whereas
strengths of simple single lap and scarf joint forms, respectively. In the increase remains at 48.57% in scarf joint forms. The increase in
both of the conditions can not be regarded as an unusual occur-
rence. However, the difference between failure loads of single and
double thickness adherends in simple single lap joints seems to be
significant. Despite load is transmitted through equal-sized adhe-
sive layers and bonding areas in both of the joint forms, joints with
thicker adherends give higher strength values in tensile loading.
This can be attributed to reduced peel stress formation in double
thickness adherends in comparison to single layer ones. During
loading of single lap joints in tension, the adherends are deformed
and shear stresses are concentrated at the ends of the overlap
section. Moreover, normal stresses which peak also at both of two
overlapping ends develop in the bonding region due to the bending
moments arising out of the eccentricity of loading. These peel
stresses lead the joint strength to drop below the theoretical
maximum shear strength value of the adhesive. It is possible to
achieve a more uniform shear stress distribution and minimizing
peel stresses by means of increasing the stiffness ratio between
adherend and adhesive layer. In the current study, this ratio is
controlled by increasing specimen thickness so that higher strength
values are obtained in double-thickness samples for equal overlap
lengths.
Scarf joint which eliminates the eccentricity in loading is an
efficient method to remove peel stresses. However, the thickness of
Fig. 3. Effects of adherend thickness of simple single lap joints on loadedisplacement
adherend restrains the available bonding area. Hence, the double-
curves (* Failure Loads are the Force values corresponding to the peak points of the thickness application in scarf joint discussed in the present study
curves). appears to be a practical technique to improve the joint efficiency.
134 I.F. Soykok / Composites Part B 77 (2015) 129e138

For scarf joint forms with double thickness adherends, the increase simple and tapered lap joints, such a moment is not expected to
in failure load is most likely to happen as a consequence of larger take place theoretically in scarf joints except for a very little po-
bonding area. Nevertheless, it may have been influenced as well by tential assembly defects. Angular deformation generated by the
some decline in peeling stress formation which stems from angle bending moment in simple and tapered joint regions implicitly
increase between adhesive interface and outer surface of adher- increases the total axial elongation. Moreover, failure elongation in
ends. As a matter of course, greater angle means a greater thickness tapered joints appears to be further due to thinned ends of
for a certain distance from sharp edges of adherends. adherends.

3.2. Effects of different end geometries and bonding techniques on 3.3. Effects of pin holes on the loadedisplacement response
the loadedisplacement response
When holey adherends were combined, liquid adhesive leaks
Fig. 5 gives an opportunity to compare the loadedisplacement into the holes of both sides. The adhesive material takes a pin-like
characteristics of adhesively bonded joints between adherends, all shape through the hole after hardening. At first, the cured adhesive
of which have an equal thickness but different bonding region pins were thought to enhance the joint's apparent shear strength.
geometries. Half and full length tapered joint and scarf joint are However, the envisaged pin-reinforcement effect could not be
demonstrated in the present figure by comparing with simple observed during the tests of joints having adhesive-filled holes
single lap form. Of these, the diagram related to scarf joint draws (Fig. 6). This phenomenon was ascribed to the formation of the
attention with the appearance of higher joint stiffness in proportion concentrated shear stresses around the hole locations which climb
to other curves. This fact supports the idea of that such joints may up over the adhesive shear strength, locally. Although the average
have relatively more stable structure with aligned adherends in the shear strength (18,008 N/mm2 for 25 25 mm2 overlapping area)
same plane. However, failure strength of scarf joint remains under remained under the maximum lap shear strength of adhesive
the level of tapered ones even though it could exceed the value of (37 N/mm2 in Table 2), the stresses around the holes could exceed
simple single lap joint. that value due to geometric stress concentrations, in just the same
When analyzed the curves, a sudden load discharge is observed way as observed in the overlap ends. Such stress peaks developed
after peak points and no plastic deformation is previously detected, in the vicinity of the hole boundaries lead to specific average
generally. However, the zigzag which can be noticed in the curve of strength losses for each hole number and location. Because of the
simple single lap joint at 3.8 mm displacement level shows a small- enhanced notch effects, increased number of holes and non-aligned
scale local instantaneous separation arising from excessive peel multi-holes appear to exacerbate the weakening phenomena of lap
stresses at the ends of the lap regions. Such a local premature joints.
separation does not exist in other joint configurations, in which
peaks of the peel stresses are minimized, geometrically. 3.4. Comparison between the failure loads of different joining
When distances to failure are analyzed in graphics, it is evident methods
that scarf joints and tapered joints constitute minimum and
maximum displacements, respectively. On the other hand, failure Many factors are effective on adhesively bond strength, such as
displacement of simple single lap joint seems to exist nearly in the adhesive type, the bond line thickness, overlap length etc.
middle of these values with a rate of 4.8 mm. This situation is Considering the adhesive material and the composite structure
related presumably with the resultant bending moment in the being tested, failure is observed to be due to induced through
joining region due to non-coaxial loading in single lap joints. Unlike thickness stresses in the laminate. Therefore, rather than adhesive

Fig. 5. Effects of different end geometries and bonding technics on loadedisplacement Fig. 6. Effects of pin holes drilled in different orientations and quantities on
curves. loadedisplacement curves.
I.F. Soykok / Composites Part B 77 (2015) 129e138 135

Table 3
Effects of different end geometries and pin holes on failure load values.

Joint design Adherend thickness Number of Failure load (kN) Comparison with
specimen simple single lap
Mean StDev
(% change in failure load)

a Simple sing. lap Single 3 11.255 0.304 0.00

b Simple sing. lap Double 3 12.600 0.651 11.93
c Single-Central Hole Single 3 10.663 0.507 5.29
d Double-Longitudinal Hole Single 3 9.625 1.308 14.49
e Double-Crosswise Hole Single 3 6.833 0.542 39.29
f Scarf Single 3 12.277 0.873 9.03
g Scarf Double 3 13.890 0.548 23.39
h Half-tapered Single 3 13.255 0.021 17.76
i Tapered Single 3 12.765 0.742 13.42

or adhesive-adherend interfacial bonding strength, through- bonding region may be more beneficial rather than forming a
thickness shear strength of composite material is here the main perforated joint region.
characteristic governing the overall joint performance. Hence, Tapered joints are mostly preferred for reduced peel stress for-
reduced joint strength is predicted as a consequence of reducing mation and fewer peaks in shear stress distribution at the end
overlapping distance. On the contrary, extended overlap area brings sections of adherends. By this way, it would be possible to achieve
about more loading capacity until its strength reaches to that of the increased shear strength at adhesive layer and, in turn, to enhance
tensile strength of cross section. However, extending overlap length the load carrying capacity of joint up to a certain value. Aiming to
may lead to negative consequences in terms of cost and labor is- clarify whether applying the taper along the entire adhesive layer
sues. In such a case, different joint designs would be helpful to or to a smaller portion is more efficient, our experiments indicate
improve the structural performance. that full-length tapered joints are more resistant to axial forces by
Effects of different end geometries and pin holes on failure load 13.42 percent compared with simple single lap joints (Table 3).
values are given in Table 3. Also, a graphical representation of However, joints having half-length tapers give relatively higher
failure load variations with respect to the given parameters can be failure strengths comparing to full length ones. Half-length tapers
found in Fig. 7. Examining the figure, decreases in failure load are contribute joint strength by 17.76 percent, which is 1.32 times
easily visible for any case of pin hole application but nonetheless, bigger than those of full-length tapers. As a consequence, defining
increased failure loads are noteworthy as well for all other end an optimum taper length specific to each joint has appeared to be
geometries. Each pin hole drilled through-thickness on bonding essential.
area induces degradation in joint quality to some extent, even Kumar et al. [24] have found a constant decline in tensile
though they were wholly filled with adhesive material. The lowest strength with increasing scarf angle, which was also consistent
observed strength values were here related to those with two- with FEM predictions. The knockdown in tensile strength was 89%
crosswise holes after two-longitudinal holes. In comparison to in comparison to the strength values of neat composite for joints
the simple single lap one, losses in failure loads of joints with having 5 scarf angle. During the present work, two types of scarf
single-central, two-longitudinal, and two-crosswise pin holes joints having different adherend thicknesses and scarf angles were
appeared to be 5.29, 14.49, and 39.29 in percentages, respectively. examined. The scarf angle of natural-thickness specimens was 4,5 ,
In a previous study [14], however, a crater created on bonding while it was twice of that in double-thickness specimens. Contrary
interface by applying a 20J transverse impact was reported to to the aforementioned study, the scarf joints with angular value of
contribute the joint strength, despite the damage to the adhesive 9 have given higher failure strength by 13.2% in proportion to
layer. This situation suggests that, an embossed structure in those with 4.5 scarf angle. This contradiction between these two
separate studies can be explained with the state of doubling in
adherend thickness and also a slight increase in bonding area.
Although there is not much expansion in adhesive layer, reinforcing
joint region provide a significant increase in load capacity, as is
seen.
Reinforcing adherends by thickening up gives extra endurance
also to simple single lap joints in axial tension. The joints with
adherends doubled in thickness by attaching two identical speci-
mens to each other have appeared to be more resistant to axial
forces by 11.93% compared to those having single-layer adherends.
This development in bonding strength is most likely due to the
reduction in angular displacement of joint region under tension
which causes there a decrease in peeling stress formations,
consequently. In a previous study by Lee et al. [20] single-lap joints
were also supported in a similar manner. Unlike those of the cur-
rent study, reinforcements were attached to adherends on the
overlap side. Therefore, the reinforcement was effective there only
until the beginning of the overlap region. Such a local thickening of
adherends has not displayed an excellent performance as expected.
Fig. 7. Box plots of failure load data obtained with three repeated tests of different end
Their test results have revealed a decrease of 44% in load capacity
geometries and pin holes. a) Simple sing. lap, b) Simp. sing. lap double thickness, c)
Single hole, d) Double longitudinal hole, e) Double crosswise hole, f) Scarf, g) Scarf compared with double strap joints. On the other hand, it is possible
double thickness, h) Half tapered, i) Tapered joints. to say that the full-length reinforcement method practiced during
136 I.F. Soykok / Composites Part B 77 (2015) 129e138

the current study is more efficient in terms of reducing peel stresses of some local ADH failure cannot be ignored, as well. The surface of
and improving joint strength. half-tapered joint indicates here the most intensive adhesive frac-
ture according to the macro-scale surface contours (Fig. 9-h). Even
if ADH is not regarded as desirable as COH, this interfacial fracture
3.5. Effects of different joint geometries on failure modes
can give favorable results in composites comparing with ones,
resulting in substrate decomposition such as FT and LFT. The in-
Failure modes were mainly classified into six categories in ASTM
crease in comparative apparent shear strength of half-tapered
D5573 [25] according to the failure shapes between adherend and
joints is related to that kind of phenomena. Minimizing adhesive
adhesive. In case the separation appears to be at the interface of
peel stresses by means of geometric feature of adherend tips in-
adhesive and adherend, the failure mode is called Adhesive Failure
duces the failure mode to transform from LFT to ADH, which means
(ADH) but if the separation occurs within the adhesive itself, it is
also an improvement in joints payload capacity. In other words, the
called Cohesive Failure (COH) in fiber-reinforced plastics (FRP)
more failure path is controlled by adhesive, the more joint strength
joints. Thin-Layer-Cohesive Failure (TLC) is very similar to Cohesive
approximates the adhesive true shear strength representing the
Failure except for its closeness to the adhesiveesubstrate interface.
value in the absence of normal stresses.
Stock-Break Failure (SB) stems from composite material weakness
Referring to the current experimental findings, scarf joint does
and occurs outside the joint region. Fiber-Tear Failure (FT) develops
not ensure a sufficiently prominent property increase to the joint
within the FRP matrix and such a failure is usually characterized by
unless it was doubled in thickness. Furthermore, this method leads
the appearance of fibers on both surfaces after rupture. Differently
to some difficulties experienced in proper matching up of especially
from FT, few or no glass-fibers transferred from the substrate to the
long bonding surfaces. Although simple single lap joint with double
adhesive layer in the Light-Fiber-Tear Failure (LFT) mode and only a
adherend thickness offers a noticeable improvement in tensile
thin layer of the FRP matrix can be monitored on the adhesive layer.
adhesive strength, some inconveniences such as weight penalty,
In fact, cohesive type failure (COH) is usually preferred for joints
excessive adhesive and adherend material consumption and labor
as an ideal failure mechanism. Different ways can be proposed to
intensity should not be ignored if this kind of joint is considered.
provide the formation of this desirable mode. For instance, the
Given all of these factors, however, tapered joint appears to be the
formation of a strong interface would turn the failure mode into a
most preferable version of joint geometry with its outstanding
cohesive on micro-scale [26]. For peel ply treatments, it has been
characteristics. Besides its being a feasible geometry from the
reported that the joint strength can vary significantly depending on
points of labor, material cost and weight-saving, tapered joints give
the adhesive used instead of adherend material. Besides, other type
satisfactory results in pulling tests comparing to simple lap joints.
surface treatments, such as plasma, laser and blasting are alterna-
Moreover, half-length taper should be preferred to full-length one
tive methods that allow controlling the failure mode and thus the
for achieving a more optimal design, because small tip angles cause
overall failure response.
adherend plate to be rather vulnerable and leads to a thin tip
When examined carefully the fracture surfaces of adhesively
damage as shown in Fig. 9-i.
bonded joints of specimens with different end geometries in Fig. 8
Consequently, although the current study has contributed to the
and Fig. 9, glass fibers are easily visible on the entire or a section of
knowledge of adhesively bonding technique in composite joints,
the ruptured surfaces. Although a thin layer of matrix material
further researches are required to be performed; especially the
appears to be removed from bonding region, there is few glass fi-
effects of temperature extremes and environmental aging condi-
bers seem to be transferred to the adhesive side during separation
tions on each joint geometry should not be ignored. Besides that,
(Fig. 9-a,d). Based upon the joint decompositions in the figures, it is
other type reinforcing materials and different fiber orientations
possible to make an inference that light-fiber-tear failure (LFT) is
would be an interesting topic for future studies. In selecting the
the main failure mode for all tested joints. Because LFT is a failure
joint type with a special geometric characteristic, the material
mode that is closely related to the peel stresses, it is obvious that,
properties of adherends should also be taken into consideration.
different joint geometries are not able to completely eliminate the
peel stress peaks where tensile stresses in the adhesive generally
control the joint failure. 4. Conclusions
On the other hand, adhesive-adherend interface separation seen
in the macrographs of certain joints (Fig. 9) denotes also the The geometry is known as a key factor controlling the adhe-
occurrence of an adhesive type failure (ADH). LFT is mainly sively bonded composite joint behavior under in-plane loading
observed formation for each of the joint geometry; however, signs case due to the resulting dissimilar stress distributions. In order to

Fig. 8. Failed composite joints with different adhesive bonding geometries (Image of 24.2 megapixels digital camera).
I.F. Soykok / Composites Part B 77 (2015) 129e138 137

Fig. 9. Macrographs of fracture surfaces of composite joints with different adhesive bonding geometries.
138 I.F. Soykok / Composites Part B 77 (2015) 129e138

investigate the response of these various type joint geometries, [5] Han SH, Cho EJ, Lee HC, Jeong K, Kim SS. Study on high-speed RTM to reduce
the impregnation time of carbon/epoxy composites. Compos Struct 2015;119:
adhesively bonded glass fiber/epoxy composite scarf joints, tapered
50e8.
joints with inclined surfaces along the half and full overlapping [6] Lee GW, Lee NJ, Jang J, Lee KJ, Nam JD. Effects of surface modification on the
length and joints with double-thickness adherends were examined. resin-transfer moulding (RTM) of glass-fibre/unsaturated-polyester compos-
Besides, the effects of adhesive-filled one centripetal, longitudinal ites. Compos Sci Technol 2002;62:9e16.
[7] Zhang K, Gu Y, Li M, Zhang Z. Effect of rapid curing process on the properties
and crosswise oriented double hole were also analyzed, individu- of carbon fiber/epoxy composite fabricated using vacuum assisted resin
ally. Based on the results, it is possible to make the following infusion molding. Mater Des 2014;54:624e31.
inferences: [8] Balzani C, Wagner W, Wilckens D, Degenhardt R, Busing S, Reimerdes HG.
Adhesive joints in composite laminates d A combined numerical/experi-
mental estimate of critical energy release rates. Int J Adhesion Adhesives
▪ Increasing the sample thickness improves the joint performance 2012;32:23e38.
in composite single lap joints as well as in scarf joints. Doubling [9] Tsai MY, Morton J. An investigation into the stresses in double-lap adhesive
joints with laminated composite adherends. Int J Solids Struct 2010;47:
in thickness is found to stiffen the joint by 76.96 and 48.57 % in 3317e25.
simple and scarf joints, respectively. [10] Kumar RLV, Bhat MR, Murthy CRL. Evaluation of kissing bond in composite
▪ Taper in single lap joints provide a significant advantage over adhesive lap joints using digital image correlation: preliminary studies. Int J
Adhesion Adhesives 2013;42:60e8.
simple single lap ones in terms of axial tensile strength. Instead [11] Hoskoti L, Kumaraswamy A, Babu J. Numerical modelling of propagation of
of along the entire adhesive layer, applying the taper to a smaller the fundamental symmetric lamb mode through a metal composite adhesive
portion was found to be more efficient. joint. Procedia Eng 2013;64:426e34.
[12] Kumar RLV, Bhat MR, Murthy CRL. Some studies on evaluation of degradation
▪ Scarf joints have relatively more stable structure with aligned
in composite adhesive joints using ultrasonic techniques. Ultrasonics
adherends in the same plane. Such joints were found to be more 2013;53:1150e62.
resistant to in-plane loading and have higher joint stiffness [13] Mahoney DCO, Katnam KB, Dowd NPO, McCarthy CT, Young TM. Taguchi
compared to the simple single lap joints. analysis of bonded composite single-lap joints using a combined inter-
faceeadhesive damage model. Int J Adhesion Adhesives 2013;40:168e78.
▪ Adhesive-filled holes in overlapping joint region have not given [14] Sayman O, Arikan V, Dogan A, Soykok IF, Dogan T. Failure analysis of adhe-
the expected pin effect. Contrary to the belief of its contribution sively bonded composite joints under transverse impact and different tem-
to the joint strength, the through-thickness drilled holes peratures. Compos Part B 2013;54:409e14.
[15] Her SC. Stress analysis of adhesively-bonded lap joints. Compos Struct
induced degradation in joint quality to some extent. 1999;47:673e8.
▪ Regardless of the bonding method and end geometry, the [16] Costa Mattos HS, Monteiro AH, Palazzetti R. Failure analysis of adhesively
dominating failure mode was observed as light-fiber-tear failure bonded joints in composite materials. Mater Des 2012;33:242e7.
[17] Neto JABP, Campilho RDSG, Silva LFM. Parametric study of adhesive joints
(LFT) for all joints tested. with composites. Int J Adhesion Adhesives 2012;37:96e101.
[18] Silva LFM, Adams RD. Techniques to reduce the peel stresses in adhesive
joints with composites. Int J Adhesion Adhesives 2007;27:227e35.
Acknowledgment [19] Kilic B, Madenci E, Ambur DR. Influence of adhesive spew in bonded single-lap
joints. Eng Fract Mech 2006;73:1472e90.
The author would like to thank Dokuz Eylul University, [20] Lee HK, Pyo SH, Kim BR. On joint strengths, peel stresses and failure modes in
adhesively bonded double-strap and supported single-lap GFRP joints. Com-
Department of Mechanical Engineering for providing the necessary pos Struct 2009;87:44e54.
experimental material and Laboratories. [21] Sayman O, Arikan V, Dogan A, Soykok IF. Multi-linear stress analysis in
adhesively bonded double-lap joint. J Compos Mater 2014;48:1497e504.
[22] ASTM D5868-01. Standard test method for lap shear adhesion for fiber rein-
References forced plastic (FRP) bonding. 2014.
[23] ASTM D2093-03. Standard practice for preparation of surfaces of plastics prior
[1] Jiang B, Huang YD. Quality inspection of laid fabric epoxy resins prepreg by to adhesive bonding. 2011.
near infrared spectroscopy. Compos Part A 2008;39:712e7. [24] Kumar AB, Sridhar I, Sivashanker S, Osiyemi SO, Bag A. Tensile failure of
[2] Jiang B, Huang YD. Investigation of the impregnation degree of the prepreg by adhesively bonded CFRP composite scarf joints. Mater Sci Eng B 2006;132:
near infrared spectroscopy. Compos Part B 2011;42:946e8. 113e20.
[3] Crossley RJ, Schubel PJ, Warrior NA. The experimental determination of pre- [25] ASTM D5573-99. Standard practice for classifying failure modes in fiber-
preg tack and dynamic stiffness. Compos Part A 2012;43:423e34. reinforced-plastic (FRP) joints. 2012.
[4] Kaufmann M, Zenkert D, Akermo M. Cost/weight optimization of composite [26] Kanerva M, Saarela O. The peel ply surface treatment for adhesive bonding of
prepreg structures for best draping strategy. Compos Part A 2010;41:464e72. composites: a review. Int J Adhesion Adhesives 2013;43:60e9.