SCIENCE CHINA

Physics, Mechanics & Astronomy

• Review • January 2014 Vol.57 No.1: 2–11
Special Topic: Airworthiness and Fatigue doi: 10.1007/s11433-013-5274-3

Adhesive bonding of composite aircraft structures: Challenges and

recent developments
PANTELAKIS Sp. & TSERPES K. I.*
Laboratory of Technology & Strength of Materials, Department of Mechanical Engineering & Aeronautics, University of Patras, Patras
26500, Greece

Received April 11, 2013; accepted May 30, 2013; published online December 10, 2013

In this review paper, the challenges and some recent developments of adhesive bonding technology in composite aircraft
structures are discussed. The durability of bonded joints is defined and presented for parameters that may influence bonding
quality. Presented is also, a numerical design approach for composite joining profiles used to realize adhesive bonding. It is
shown that environmental ageing and pre-bond contamination of bonding surfaces may degrade significantly fracture tough-
ness of bonded joints. Moreover, it is obvious that additional research is needed in order to design joining profiles that will en-
able load transfer through shearing of the bondline. These findings, together with the limited capabilities of existing
non-destructive testing techniques, can partially explain the confined use of adhesive bonding in primary composite aircraft
structural parts.

adhesive bonding, fracture toughness, finite element analysis

PACS number(s): 81.70.Bt, 81.05.-t, 61.41.+e, 62.20.Mk, 02.70.Dh

Citation: Pantelakis S, Tserpes K I. Adhesive bonding of composite aircraft structures: Challenges and recent developments. Sci China-Phys Mech Astron,
2014, 57: 211, doi: 10.1007/s11433-013-5274-3

1 Introduction The joining of thin-walled composite structures by adhe-

sive bonding, without using fasteners, is a promising meth-
Current applicable conventional design rules and standards od to reduce aircraft cost and weight [1–5]. Adhesive bond-
for joining primary composite structures usually lead to ing also brings several technical advantages such as the
local thickening of the fastener area and to a large number avoidance of detrimental stress concentrations and the
of fasteners distributed over several lines to prevent or delay avoidance of fiber cuts associated with the drilling of holes
the different failure modes at the interface. The certification for introducing fasteners [1–5].
approach follows a strict fail-safe philosophy with the sizing For secondary structures, adhesive bonding is common
strategy that the additional fasteners are capable of carrying practice but until today the certification rules, that are ap-
full Limit Load. Therefore, compared to their counterparts plicable for primary bonded structures, prevent the use of
in metallic structures, the interfaces in composite thin- boltless bonded joints for primary structures, as a result of
walled structures induce a significant weight and cost pen- earlier experiences, where the interpretation of the rules led
alty and mitigate the technical and economic benefits ex- to in-service premature failure incidents on adhesively
pected from the massive introduction of composites in aero- bonded joints. The use of bonded joints in primary struc-
structures. tures is also prevented by the inability of the existing
non-destructive testing techniques to capture all types of
*Corresponding author (email: [email protected]) defects in the bondline, the requirement to redesign aircraft

© Science China Press and Springer-Verlag Berlin Heidelberg 2013 phys.scichina.com link.springer.com
 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 3

parts such as the load to be transferred by pure shear of the of CFRP bonded joints was investigated experimentally. In
bondline, the susceptibility of the joint (adhesive and adhe- the study, two different structural adhesives were used: the
sive/substrate interface) into environmental ageing and well-established structural adhesive Epibond 1590 A/B and
pre-bond contamination [1,5,6] as well as the unknown fa- the new LMB adhesive. The experimental procedure and
tigue behavior of bonded joints [4]. main findings of this work are described in the following
In this review paper, some of the above mentioned chal- paragraphs.
lenges will be discussed in detail and some recent develop-
ments on the durability of bonded joints, on the parameters 2.1 Materials and specimens
that may influence bonding quality as well as on the nu-
merical design of composite joining profiles will be pre- The laminated adherents were made from 8 plies of the pre-
sented. preg AS4/8552 composite material. The lay-up of the plates
was quasi-isotropic. The adherents were bonded using the
Epibond 1590 A/B [13] adhesive and the epoxy LMB adhe-
2 Effect of accelerated environmental ageing sive. The Epibond 1590 A/B adhesive is a structural adhe-
sive designed for aerospace applications. It is a two-part
Aircraft structures during service are subjected to various paste epoxy adhesive capable of service up to 177°C. The
environmental conditions (temperature, humidity, radiation, LMB adhesive is a prototype two-part paste adhesive with
etc). This is why the engineering community makes the improved peel performance, hot/wet Tg and processing
great effort to investigate the possible effects of environ- (slump) compared to the Epibond 1590 A/B adhesive. In ref.
mental ageing on the structural integrity of aerostructures [12] the effect of adhesive thickness was also included by
[6–11]. The investigation refers to both metallic structures, considering two different thicknesses, namely 0.5 mm and
where corrosion is the main problem, and to polymeric 1.5 mm.
structures such as composite materials and structural adhe-
sives. Several research papers have reported on this subject. 2.2 Mechanical testing
For instance, in ref. [7] six structural adhesives were tested
under different loading conditions after being exposed for 8 Mode I tests were conducted on the Double Cantilever
years into different environments. A similar work was per- Beam (DCB) specimen according to AITM 1.0053 [14]
formed in ref. [8] where the degradation in strength due to standard. A schematic sketch of the DCB specimen is given
accelerated ageing of two epoxy adhesives was experimen- in Figure 1(a). In facilitating crack propagation (debonding),
tally measured. In ref. [8] the critical strain energy release a pre-crack of length between 35 and 40 mm was initially
rate of wet-aged and reference specimens was measured created at one side of the specimen. Afterwards, the speci-
under several mixed mode loading ratios. It was found that men was unloaded until pre-crack was completely closed
the combined effect of temperature and moisture degrades and reloaded until the desired crack length (~100 mm) with
significantly the respective critical strain energy release rate
of the aged specimens. In ref. [9] mechanical tests were
conducted to assess the effect of moisture content on the
static strength of uncured adhesively bonded steel joints.
The results showed that exposure for 1008 h in a hot humid
environment decreases the strength of adhesively bonded
steel joints by 96%. In addition to ref. [9], the effect of age-
ing on the mechanical performance of adhesive bonded
Carbon Fiber Reinforced Plastics (CFRP) and aluminum
specimens was investigated in refs. [10,11], respectively.
The study on CFRP bonded structures was performed using
high-frequency dielectric analysis after exposure to condi-
tions of 70oC and 100% humidity. It was found that speci-
mens exhibit a significant loss of strength over the pro-
longed exposure period ref. [10]. Moreover, in ref. [11] the
effects of environmental moisture and temperature on fa-
tigue threshold and crack growth behavior of P2-etched and
commercial coil-coated aluminum bonded joints were stud-
ied under mixed-mode loading conditions.
In the frame of the investigation of durability of compo-
site bonded joints used in aeronautical structures, in ref. Figure 1 (a) Schematic sketch of the DCB specimen, (b) schematic
[12], the effect of accelerated ageing on fracture toughness sketch of the ENF specimen.
4 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1

a displacement rate of 1 mm/min. From the test, the mode I

critical energy release rate GIC was derived using
A
GIC   10 6 , (1)
a W
where A is the total energy required to achieve the total
crack length equals to the total area bounded by the load–
displacement curve, a is the total crack length (final minus
initial crack length) and W is the specimen’s width [14]. Eq.
(1) can be related to mode I critical stress intensity factor
KIC for plane strain. Note that KIC is the valid ASTM frac-
ture toughness value for plane strain. Plane stress toughness
is not valid according to ASTM standard.
Mode II tests were conducted on the End-Notch Flexure
(ENF) specimen according to standard AITM 1.0006 [15].
A schematic sketch of the ENF specimen is given in Figure
1(b). The specimens used for mode II tests had an overall
length of at least 110 mm. The displacement rate used was
again 1 mm/min. From the test, mode II critical energy re-
lease rate GIIC was derived using

9 Pa 2 d  1000
GIIC  , (2)
2W (1 / 4 L3  3a 3 ) Figure 2 (a) DCB specimen during mode I testing, (b) ENF specimen
during mode II testing.
where d is the displacement at crack propagation onset, P is
the critical load required for crack onset, a is the initial
crack length and L is the span length [15]. Eq. (2) can also
be related to mode II critical stress intensity factor KIIC for
plane strain. The ratio GIC/GIIC is directly related to
(KIC/KIIC)2 [16].
Figure 2(a) illustrates a DCB specimen during mode I
testing while Figure 2(b) an ENF specimen during mode II
testing.

2.3 Thermal and wet ageing

Accelerated ageing aims to simulate the varying environ-

mental conditions to which the aircraft structure is subjected. Figure 3 (Color online) Thermal cycle applied during the thermal ageing.
Ageing comprises thermal ageing, which is due to the varia-
tion of temperature between extremes that deviate a lot from
room temperature (20oC–22oC), and wet ageing, which is
due to the increased humidity. Both types of ageing have
been found to affect the mechanical performance of poly-
meric materials such as CFRPs and structural adhesives.
In ref. [12] thermal ageing was realized by subjecting the
specimens into 1000 thermal cycles described in Figure 3.
On the other hand, wet ageing was realized by subjecting
the specimens for 176 d (about 4200 h) to a relative humid-
ity of 85% under an elevated temperature of 70°C. Figure 4
plots the variation of weight gain in the specimens with re-
gard to square root of exposure time. Figure 4 (Color online) Evolution of weight gain in the bonded joints
due to moisture absorption as a function of time.

2.4 Experimental results

The tests were decided to enable investigation of the com-
The test matrix of the mechanical tests is shown in Table 1. bined effect of accelerated ageing (wet and thermal) and
 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 5

Table 1 The test matrix for mode I and mode II tests

Mode I tests
Epibond 1590 LMB
t=0.5 mm t=1.5 mm t=0.5 mm t=1.5 mm
Reference 5 5 Reference 5 5
Thermal ageing 5 5 Thermal ageing 5 5
Wet ageing 5 5 Wet ageing 5 5
Mode II tests
Epibond 1590 LMB
t=0.5 mm t=1.5 mm t=0.5 mm t=1.5 mm
Reference 5 5 Reference 5 5
Thermal ageing 5 5 Thermal ageing 5 5
Wet ageing 5 5 Wet ageing 5 5

adhesive thickness on fracture toughness of the CFRP than those of the Epibond 1590 A/B adhesive revealing an
bonded joints. advantage of the LMB over the Epibond 1590 regarding the
The load-displacement curves measured during mode I resistance to shear loads. Thermal ageing caused a decrease
tests are illustrated in Figure 5. The maximum load value in GIIC for both adhesives. Moreover, the increased adhesive
measured for each group of specimens gives a qualitative thickness led to the reduction in GIIC values of both adhe-
measure of the bonded joint performance as it represents the sives.
load at which crack initiation occurred. Moreover, the By using the experimentally measured values of GIC and
maximum load is of major importance for the aeronautic GIIC, the critical mixed mode fracture toughness SC of the
industry as it is used for designing bonded joints. specimens can be derived using the Strain Energy Density
Figure 6 compares the average GIC values of the speci- theory [16] through the critical mode I and mode II stress
mens tested. As can be seen, adhesive thickness has a con- intensity factors KIC and KIIC as mentioned earlier. This der-
tradictory effect on fracture toughness for the two adhesives: ivation is very important because it allows assessing the
positive for the Epibond 1590 A/B and negative for the effect of ageing on the fracture toughness of the joints in-
LMB. This is probably due to the differences in the interac- dependently of loading mode which varies according to
tion mechanisms between yielding and thickness occurring flight conditions.
in the two adhesives. Thermal ageing leads to a decrease in
toughness of the bonded joints, which is more significant
for the LMB. On the contrary, wet ageing leads to an in- 3 Effect of pre-bond contamination
crease in toughness of the bonded joints. This is due to plas-
ticization of the adhesive caused by the presence of water. It The physicochemical parameters that may influence fracture
is noted that larger exposure periods are expected to cause a toughness of bonded joints may be divided into the manu-
decrease in the toughness of the joints. A very interesting facturing-related parameters (pre-bond contamination) and
finding is that the bonded joints assembled with the LMB in-service-related parameters (after-bond contamination).
adhesive with the small thickness (0.5 mm) have in all cases Pre-bond contamination mainly affects the physicochemical
the highest fracture toughness. conditions of the substrate, thus affecting bonding quality
Figure 7 compares the average GIIC values of the speci- [17–25]. After-bond contamination mainly influences the
mens tested. The GIIC values of the LMB are much greater integrity of the adhesive as described in sect. 2.

Figure 5 Load displacement curves of mode I tests: (a) reference specimens, (b) thermally aged specimens, and (c) wet aged specimens.
6 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1

T700/M21 composite material. The plates were bonded us-

ing one layer of the film adhesive FM300 K.05. The dimen-
sions of the specimens are 200 mm×25 mm×1.56 mm. The
thickness of the specimens was 3.2 mm. The lay-up of the
CFRP laminates was [0°2/90°]S.

3.2 Pre-bond contamination

The scenario of the poorly cured adhesive applies to both

adhesive joining and repairing. If a parameter (e.g. temper-
ature) of the curing process deviates from the one pre-
scribed in the respective specification, the quality of the
cured part may be influenced. In ref. [26] in simulating poor
curing conditions, a lower curing temperature was applied
Figure 6 (Color online) Comparison of mean GIC values for the different for the entire process; the other parameters retained their
set of tests.
suggested values.
Release agent is a film used to facilitate removing of
parts from molds. In the manufacturing of composite parts,
Si-based release agents are being used. The Si contamina-
tion on the CFRP surface may influence adhesion. In ref.
[26] the samples were exposed to a solution of Frekote 700
NC which is a Si-based release agent film.
The presence of humidity may affect the quality of adhe-
sion and may also degrade strength of the CFRP material. In
[26] the specimens were stored for four weeks in an immer-
sion bath put inside a container covered with aluminum foil
to prevent evaporation of water and stored in an oven at
70°C. The measured average weight increase of the samples
due to water uptake was about 1.32 wt% .
Skydrol is a fire resistant aviation hydraulic liquid exten-
sively used in aircraft hydraulic systems. Skydrol may in-
Figure 7 (Color online) Comparison of mean GIIC values for the different trude into material surfaces due to a possible leakage. For
set of tests.
the sake of simulating possible contamination of a compo-
site surface before being covered by a bonded patch due to a
In ref. [26] an experimental program was conducted to repair, in ref. [26] tests were done using specimens having
assess the effect of pre-bond contamination on fracture the surfaces of CFRP adherents contaminated by Skydrol.
toughness of CFRP bonded laminates by conducting mode I Thermal degradation may occur in several aircraft parts
tests. To this end, five different representative pre-bond due to local overheating caused by various phenomena.
contamination scenarios were considered; namely, poor Exposure to extremely high temperatures or to lower tem-
curing of the adhesive, release agent contamination, mois- peratures may damage the matrix of the CFRP, thus influ-
ture uptake, Skydrol contamination and thermal degradation encing the physicochemical conditions of the adher-
of the composite substrate. The first three scenarios may ent/adhesive system. In ref. [26] test samples were stored
appear during manufacturing of the joints while the latter for 1 hour at different temperatures in an air circulation ov-
two are related to maintenance and could appear, for exam- en.
ple, during implementation of a bonded repair in a cracked
part. All five contamination scenarios lead to the creation of 3.3 Mechanical testing
defects in the adherent/adhesive interface which have the
form of kissing bonds and cannot be detected by existing Mode I experiments were conducted to measure fracture
NDT techniques. In the next paragraphs, the experimental toughness of CFRP bonded joints (DCB specimen) accord-
procedure and main findings of the study conducted in ref. ing to ISO 15024 [27] and ISO 25217 [28] standards. The
[26] will be described. experimental procedure followed is the same as the one
described in sect. 2.2 except from the length of the
3.1 Materials and specimens pre-crack (3–5 mm), the loading rate (2 mm/min) and the
total desired crack length (115 mm). From the tests, GIC
The laminated plates were made from 6 plies of the prepreg referring to crack initiation was derived using the corrected
 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 7

beam theory [27]. According to this theory, GIC is given by

3 P
GIC  , (3)
2b ( a   )

where P is the crack initiation load value, δ the load line

displacement at this point, b the specimen width, a the total
delamination length, F the large-displacement correction
and Δ is a correction factor derived experimentally from the
plot of the cube root of the compliance ( / P)1/3 as a
Figure 8 The average crack initiation GIC values for the six scenarios.
function of crack length a.

3.4 Experimental results

The average values of GIC for the six different scenarios are
compared at the histogram of Figure 8. GIC values are
strongly affected by the maximum load. The poor curing led Figure 9 Schematic displays of joining profiles.
to a decrease of the GIC value by almost 95%; this finding
reveals the criticality of curing process in bonded joints.
The dramatic decrease in GIC is explained by the failure beam realized by means of a Pi-shaped profile is illustrated
mode of the poorly cured specimen in which a complete in Figure 10. Design of the multi-functional profiles, for
lack of cohesion is ascertained. Release agent and thermal example to ensure their integrity as well as the integrity of
degradation also decrease dramatically GIC of the joint by the bondline is still a major challenge 29. Important tools in
70% and 62%, respectively. Moisture and Skydrol have also the design of multifunctional profiles are the continuum
a considerable effect on the fracture toughness of the joints damage modeling technique [29,31–33], the cohesive zone
(a 25% and a 27% decrease in GIC, respectively). modeling technique [30] and Digital Image Correlation
[31,33].
In ref. [32], numerical design optimization of an NCF
4 Numerical design of joining profiles H-shaped joining profile for bonded joints was conducted
with respect to failure load for the H-tension (pull-out)
Realization of adhesively bonded joints requires in some load-case. In the following paragraphs the methodology and
cases redesign of the structural parts to be assembled as main numerical findings will be described briefly.
current designs were selected so as to implement bolted
connections. The main objective of the redesign of the 4.1 Geometry and materials
structural parts is the load to be transferred between the as-
sembled parts through shear stresses and not normal stresses. The initial geometry of the H-shaped multi-functional pro-
Therefore, large adhesive surfaces loaded in shear have to file is schematically shown in Figure 11(a). The profile is
be created. A very useful tool in the realization of bonded used to adhesively join two aluminum inserts. The aim of
joints is multifunctional profiles of different shapes which this work is to propose an optimal geometry which will give
act as joining elements between the assembled parts. The higher strength for the H-tension load-case, schematically
multifunctional profiles are made of textile composite mate- described in Figure 11(b). The H profile was made from the
rial and can be of different shapes as shown in Figure 9. An HTS/RTM6 NCF material [32]. For the adhesive, the EA
example of a bonded joint between a composite skin and a 9695 adhesive film is used. More details about the materials

Figure 10 Bonded joining of a composite skin with a composite beam by means of a Pi-shaped joining profile.
8 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1

Figure 12 FE mesh of the 45/45 RVE.

Figure 11 (a) Schematic sketch and dimensions of the H-shaped profile,
(b) Schematic sketch of the H-shaped bonded joint loaded in tension.
4.4 Optimization module

can be found in ref. [32]. The optimization module, integrated in the ANSYS© com-
mercial FE code was employed to perform optimization
4.2 Design optimization methodology [32]. The module employs three types of variables to char-
acterize the design process: the design variables, the state
The proposed design optimization methodology, combining variables and the objective function. These variables are
the method of progressive damage modeling with numerical represented by scalar parameters in the ANSYS© Parametric
optimization, comprises the following steps: Design Language [34].
(1) Characterization of the mechanical performance of
the NCF material using local homogenized progressive 4.5 Analysis of the reference geometry
damage modeling;
(2) Development of the global FE model of the multi- In developing the 3D FE model of the H-profile, the
functional profile; ANSYS© FE code was used. A generic view of the 3D FE
(3) Assignment of the mechanical properties of the RVE model can be seen in Figure 13. All components were mod-
to the elements of the FE model of the multi-functional pro- eled using SOLID185 element which is defined by 8 nodes
file and implementation of progressive damage modeling to and 3 degrees of freedom per node. Progressive damage
simulate the global mechanical performance of the joint; modeling on the bonded joint with the reference geometry
(4) Analysis of the reference geometry to assess strength of the H-profile is initially performed to predict strength and
and select the objective functions (critical stresses); identify the critical stresses that govern the main failure
(5) Decision on design variables, design constraints and mechanisms that cause final failure of the structural part.
state variables (dimensions of H, geometry constrains); Maximum values of these stresses correspond to the objec-
(6) Performing optimization; tive functions which are to be minimized. Consequently, the
correlation between objective functions and various dimen-
(7) Analysis of the optimal geometry.
sions of the H-profile is necessary. Dimensions with the
In the following sections, each of the above steps will be
stronger influence on the values of the objective functions
described shortly. More details can be found in ref. [26].
will serve as design variables in the optimization module. In

4.3 Behavior of the NCF material

The mechanical behavior of the NCF material was charac-

terized using local homogenized progressive damage mod-
eling. For this purpose, two representative volume elements
(RVEs) have been selected, namely one for the dual layer
0/90 and one for the 45/45. The geometry of the RVEs has
been built using the WiseTex software and transferred to the
ANSYS© FE software to be solved. Figure 12 shows the FE
mesh of the RVE for the 45/45 dual layer. The results of
the material characterization are in the form of material
stiffnesses and strengths at each material direction for ten-
sion and compression load-cases [32]. Figure 13 FE mesh of the H-shaped bonded joint.
 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 9

the optimization, the restrictions, in which design variables

must be subjected, are also considered. Usually, these re-
strictions come from the manufacturing sector. Since these
restrictions were not available, a variation within ±20% of
the initial design variables was chosen. Excessive con-
straints can lead to designs that may have better properties
but are impossible to construct. Geometry of the optimum
design is then chosen, based on the results of the optimiza-
tion. The final step is to analyze the optimum geometry.
The failure modes predicted just before failure, at 0.8
mm are: failure at the X direction (fiber failure) of the NCF
material, mostly at the 45/45 dual layers, failure at the Z
direction of the NCF material (delamination failure) and
adhesive failure (debonding). Among three failure modes Figure 15 Feasible design sets.
the most critical, which causes the load-drop at the load-dis-
placement curve, was fiber failure at the X direction of the which is the one that gave the lowest value of SX. Values of
NCF material. The state of this failure mode just before H2, T4 are in mm and SX is in MPa.
final failure is shown in Figure 14(a). The stress that causes
and governs this failure mode was the tensile stress SX.
Debonding and delamination at the NCF material were also 4.7 Analysis of the optimal geometry
present but of secondary importance. Therefore, it is decid- The final step of the design optimization process, which is
ed that the minimization of SX will serve as the objective also the validation step, is the progressive failure analysis of
function at the optimization loop under the assumption, the optimal geometry. The percentage of increase in
which is not far from reality, that the other stresses do not strength of the optimal geometry with regard to strength of
significantly affect fiber failure of the NCF material. the reference geometry is the criterion for the success of the
optimization.
4.6 Optimization analysis Figure 16 compares the load-displacement curves of the
reference and optimal geometries for the H-tension load-
The dimensions influencing SX have been found from a set case. As can be seen, an improvement in the overall me-
of stress analyses to be the tapered length H2 (initial value chanical performance of the joint has been achieved: stiff-
12 mm) and the thickness of the adhesive T4. These two ness has been slightly increased while the failure load and
dimensions serve as the design variables in the optimization. displacement to failure have been considerably increased by
The design variables were chosen to vary within the range 4.7% and 10%, respectively. Of course, these values depend
of ±20% of the reference values. The extremes of this range on the mesh density adopted at the FE model of the joint.
represent the state variables. During optimization loop 10 The failure load corresponds to the maximum load in the
design sets were created and stress analyses with a dis- curve. This finding validates the design optimization meth-
placement of 1mm were performed for each design. Figure odology. At the optimal geometry, the developed SX
15 illustrates the design sets that were built from the opti-
mization loop. Marked with stars is the optimum design,

Figure 14 Predicted accumulation of fiber failure at the X direction of the

45/45 dual layers at the applied displacement of 0.8 mm for a. the reference Figure 16 Comparison between the load-displacement curves of the
geometry and b. for the optimal geometry. reference and optimal geometries for the H-tension load-case.
10 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1

stresses, driving the main failure mechanism of the joint site aircraft structures: A review of scientific challenges and opportu-
(fiber failure at the longitudinal direction), are smaller and nities. Prog Aerospace Sci, 2013, http://dx.doi.org/10.1016/j.paerosci.
2013.03.003
the progression of fiber failure is slower as can be seen in 2 Banea M D, Da Silva L F M. Adhesively bonded joints in composite
Figure 14(b); it is shown that at the same load-level (8 mm) materials: an overview. Proc IMechE Part L-J Mater Design Appl,
the accumulation of fiber failure is smaller at the optimal 2009, 223: 1–18
geometry than at the reference geometry. 3 Parashar A, Mertiny P. Adhesively bonded composite tubular joints:
Review. Int J Adhesion Adhesives, 2012, 28: 58–68
4 Abdel Wahab M M. Fatigue in adhesively bonded joints: A review.
5 Conclusions ISRN Mater Sci, 2012, 2012: 1–25
5 Baldan A. Adhesively-bonded joints and repairs in metallic alloys,
polymers and composite materials: Adhesives, adhesion theories and
The challenges that have to be met in order for the use of surface pretreatment. J Mater Sci, 2004, 39: 1–49
adhesive bonding to be extended from secondary to primary 6 Pitt S, Jones R, Peng D. Characterization of the durability of adhesive
composite aircraft structures were reviewed. The basic bonds. Fatigue Fract Eng Mater Struct, 2012, 35: 998–1006
7 Su N, Mackie R I, Harvey W J. The effects of ageing and environ-
challenges, as identified by the authors, are: durability of ment on the fatigue life of adhesive joints. Int J Adhes Adhes, 1992,
bonded joints, pre-bond contamination of bonded surfaces, 12: 85–93
effective design of joining profiles to realize adhesive 8 Wylde J W, Spelt J K. Measurement of adhesive joint fracture prop-
bonding, development of extended non-destructive tech- erties as a function of environmental degradation. Int J Adhes Adhes,
niques capable of detecting all types of defects in the 1998, 18: 237–246
9 Lu Z, Wang P, Lin J, et al. Effect of moisture content in uncured ad-
bondline and fatigue.
hesive on static strength of bonded galvanized DP600 steel joints. Int
Recent developments from the research conducted by the J Adhes Adhes, 2011, 31: 202–208
authors in the areas of durability of bonded joints, pre-bond 10 Halliday S T, Banks W M, Pethrick R A. Dielectric studies of adhe-
contamination of bonded surfaces and effective design of sively bonded CFRP/epoxy/CFRP structures–design for ageing.
joining profiles were also presented herein. Compos Sci Technol, 2000, 60: 197–207
Durability of two composite adherent-adhesive systems 11 Datla N V, Papini M, Ulicny J, et al. The effects of test temperature
and humidity on the mixed-mode fatigue behavior of a toughened
was studied in combination with adhesive thickness by
adhesive aluminum joint. Eng Fract Mech, 2011, 78: 1125–1139
conducting mode I and mode II tests on specimens that have 12 Katsiropoulos Ch V, Chamos A N, Tserpes K I, et al. Fracture
been subjected to accelerated thermal and wet ageing. The toughness and shear behaviour of composite bonded joints based on a
experimental results showed a contradictory performance of novel aerospace adhesive. Composites Part B-Eng, 2012, 43: 240–
the two adhesives under the different load conditions and 248
13 Epibond 1590 A/B aerospace structural adhesive, technical data sheet
thicknesses. Thermal ageing degraded the performance of
by HUNTSMAN, 2006
the joints. Wet ageing enhanced the fracture toughness of 14 AITM1-1.0053, Airbus test method: determination of fracture tough-
the bonded joints due to plasticization of the adhesive. ness energy of CFRP bonded joints – mode I, 2006
However, subjection to a humid environment for a larger 15 AITM1-1.0006, Airbus test method: determination of fracture tough-
period is expected to lead to a degradation of fracture ness energy of CFRP bonded joints – mode II, 2006
toughness of the bonded joints. 16 Sih G C. Mechanics of Fracture Initiation and Propagation: Surface
and Volume Energy Density Applied as Failure Criterion. Kluwer
The effect of pre-bond contamination on fracture tough-
Academic Publishers, 1991
ness of adhesively bonded CFRP laminates was studied by 17 Joshi S B, Gray T F, Banks W M, et al. Environmental ageing of ad-
conducting mode I tests. To this end, five different repre- hesively-bonded joints. II. Mechanical studies. J Adhesion, 1997, 62:
sentative pre-bond contamination scenarios were considered. 317–335
All considered scenarios were found to affect significantly 18 Aydin M D, Temiz S, Özel A. Effect of curing pressure on the
fracture toughness of adhesive bonds. This finding high- strength of adhesively bonded joints. J Adhesion, 2007, 83: 553–571
19 Jeenjitkaew C, Luklinska Z, Guild F. Morphology and surface chem-
lights the criticality of the physicochemical conditions of istry of kissing bonds in adhesive joints produced by surface con-
the bonded surfaces in the quality and strength of the tamination. Int J Adhes Adhes, 2010, 30: 643–653
bondline. 20 Parker B M. Surface pretreatment of carbon fibre-reinforced compo-
Finally, a numerical design optimization methodology sites for adhesive bonding. Composites, 1982, 13: 280–288
for joining profiles was presented. Using the methodology 21 Comyn J. Durability of Structural Adhesives. In: Kinloch A J, ed.
London and New York: Applied Science Publishers, 1983. 85–129
the geometry of an NCF H-shaped joining profile was opti-
22 Drain F, Guthrie J, Leung C L, et al. The effect of moisture on the
mized with respect to strength. The final geometry has an strength of steel–steel cyanoacrylate adhesive bonds. Int J Adhes
increased strength compared to the initial one by 4.7%. The Adhes, 1984, 17: 71–81
methodology is parametric and can be applied for the opti- 23 Frigione M, Aiello M A, Naddeo C. Water effects on the bond
mization of any joining profile made of textile composite strength of concrete/concrete adhesive joints. Constr Build Mater,
material. 2006, 20: 957–970
24 Comyn J, Groves C L, Saville R W. Durability in high humidity of
glass-to-lead alloy joints bonded with an epoxide adhesive. Int J
Adhes Adhes, 1994, 14: 15–20
1 Katham K B, Da Silva L F M, Young T M. Bonded repair of compo- 25 Cognard J. Influence of water on the cleavage of the adhesive joints.
 Pantelakis S, et al. Sci China-Phys Mech Astron January (2014) Vol. 57 No. 1 11

Int J Adhes Adhes, 1988, 8: 93–99 30 Sun S Y, Chen H R. The interfacial fracture behavior of foam core
26 Markatos D N, Tserpes K I, Rau E, et al. The effects of manufactur- composite sandwich structures by a viscoelastic cohesive model. Sci
ing-induced and in-service related bonding quality reduction on the China-Phys Mech Astron, 2011, 54: 1481–1487
mode-I fracture toughness of composite bonded joints for aeronauti- 31 Tserpes K I, Cinquin J, Pantelakis S. On the mechanical performance
cal use. Composites Part B-Eng, 2013, 45: 556–564 of H-shaped non-crimp fabric adhesively bonded joints. J Composite
27 ISO 15024. Fibre-reinforced plastic composites – determination of Mater, 2011, 45: 1607–1619
mode I interlaminar fracture toughness, GIC, for unidirectionally re- 32 Tserpes K I, Koumpias A S, Pantelakis S. Numerical optimization of
inforced materials, 2001 composite multi-functional joining profiles with respect to strength.
28 ISO 25217. Adhesives – determination of the mode 1 adhesive frac- Int J Terraspace Sci Eng, 2012, 5: 77–82
ture energy of structural adhesive joints using double cantilever beam 33 Guo B Q, Xie H M, Zhu J G, et al. Study on the mechanical behavior
and tapered double cantilever beam specimens, 2009 of adhesive interface by digital image correlation. Sci China-Phys
29 Tserpes K I, Ruzek R, Pantelakis S. Strength of Pi-shaped non-crimp Mech Astron, 2011, 54: 547–580
fabric adhesively bonded joints. Plastics Rubber Composites, 2012, 34 Design Optimization, Release 11.0 Documentation for ANSYS©,
41: 100–106 2009