FAILURE OF ADHESIVELY BONDED JOINTS.

ANALYSIS OF EXPERIMENTAL RESULTS BY

NOMINAL AND LOCAL APPROACHES.

A. Barroso, F.París, V. Mantič

Group of Elasticity and Strength of Materials, School of Engineering, University of Seville,

Camino de los Descubrimientos s/n, Sevilla, Spain.

ABSTRACT
The failure in structural adhesively bonded lap joints can be analyzed following different approaches:
using a nominal stress state typically given by Strength of Material models, a local singular stress state
induced at the multimaterial corners of the joint, or using cohesive constitutive laws for the adhesive layer
or other damage progression models. The aim of this paper is to analyze the suitability of models based
on local stress states to predict a failure in metal-to-composite double-lap joints and their capability to
predict a failure when the geometrical parameters such as overlap length, adherent thickness or stacking
sequence change. For this objective, a preliminary test program has been carried out, which has shown
some limitations in approaches using local stress states when a gross yielding takes place.

1.- INTRODUCTION
The failure prediction of adhesively bonded lap-joints between composite materials is a
complex problem, Mathews et al [1]. There exist different approaches, some of them
based on either the nominal stress state, typically evaluated by means of the Strength of
Materials formulas, or the local (singular) stress state at the multimaterial corners which
appear in the overlap zone. There are also other approaches, such as Continuum
Damage Mechanics, which try to characterize the failure process using more complex
constitutive laws.
Among the approaches based on nominal stress states the work by Hart-Smith [2] is an
excellent example. Hart-Smith found, for example, that in double lap joints in tension
the failure key parameter is the strain energy in shear per unit lap-area. This fact was
successfully confirmed by the experimental results by Leman and Hawley [3].
Crocombe [4] proposed an alternative criterion based on a plastic collapse mechanism
in the adhesive layer. Kairouz [5] studied the role of the stacking sequence in the failure
of composite lap joints. Adams and co-workers have investigated in depth the role of
the adhesive fillet in the failure, Adams et al [6], and have improved analytical models
for the stress analysis of the adhesive joints, Adams and Mallick [7].
The approaches based on the local (singular) stress states have also been widely
investigated in literature. The presence of points, in the overlap zone of the joint, where
both geometry and mechanical properties change abruptly have attracted attention of
several studies to the analysis of the singular stress state induced at these corners and
their influence on the initiation of failure. Some of these proposals were based on the
presence of a crack at the corner, Ferlund and Spelt [8], or use interfacial fracture
mechanics concepts, Hutchinson and Suo [9]. Singular stress states at these points, not
considering previous cracks emanating from the corner tip, were analyzed by Gradin
[10], Groth [11], Hattori [12], Reedy [13], Reedy and Guess [14], Lefebvre et al [15]
and Quaresimin and Ricotta [16]. Other works not directly related with adhesive joints
but with general corner problems are useful in our problem as they provide powerful
techniques and methods for asymptotic stress characterization, e.g., Leguillon and
Yosibash [17] or Grenestedt and Hallstrom [18] analyzing the presence of a very small
crack at the notch tip and found a relationship between the stress intensity factor of this
small crack and the stress intensity factor of the uncracked corner.
Other approaches, e.g., use a progressive anisotropic degradation of the material,
Laschet and Stas [19], define a critical damage zone length by using Finite Elements,
Sheppard et al. [20], or use cohesive models, Mohammed and Liechti [21].
The objective of the present work is to check the representativity of the singular stress
field in the failure of composite to metal double lap joints. In Section 2, the asymptotic
stress field at the different multimaterial corners is evaluated by using a semi-analytical
tool developed by the present authors together with numerical Boundary Element
Method (BEM) models. In Section 3, the numerically analyzed geometries are tested in
shear by tension using double lap joint specimens, and then, in Section 4, the
comparison of numerical and experimental results is discussed.

2.- CHARACTERIZATION OF THE LOCAL (SINGULAR) STRESS FIELD

Let us consider a polar coordinate system (r,θ) centred at the corner tip. Let us assume
stress singularities of the type Ο(rλ−1), then the asymptotic stress and displacement
fields at the corner can be written as a series expansion

n n
Kk
σ ij (r , θ ) = ∑ f
1− λk ijk
(θ ), u i ( r , θ ) = ∑ K k r λk g ik (θ ) , i , j = 1,2,3 (1)
k =1 r k =1

where λk are the characteristic exponents, (1- λk) being the order of stress singularities,
and Kk are the Generalized Stress Intensity Factors (GSIF). Functions fijk(θ) and gik(θ)
are characteristic angular shape functions of the circumferential coordinate θ. It is well
known that λk, fijk(θ) and gik(θ) only depend on the local geometry, material properties
and local boundary conditions at the corner neighbourhood and can be evaluated
analytically, except of the numerical evaluation of roots of an analytic function, in most
of the cases, Costabel et al [22], Barroso et al [23]. Situations which lead to stress
singularities different than O(rλ−1) represent exceptions, see Sinclair [24] for a few
examples of singularities of the type O(rλ−1ln r), and will be not considered here for the
sake of simplicity.
The computation of GSIF Kk, requires the knowledge of the whole geometry and
loading conditions, so for non trivial geometries, numerical models are needed to
calculate Kk. The multimaterial corners appearing in a typical metal-to-composite
[0/90/0] double lap joint are depicted in Figure 1. The mechanical properties
(orthotropic) of the unidirectional composite (AS4/8552) are (E11=141.3 GPa,
E22=E33=9.58 GPa, G12=G13=5.0 GPa, G23=3.5 GPa, ν12=ν13=0.3, ν23=0.32, α1=-1·10-6
ºC-1, α2=α3=26·10-6 ºC-1) while the adhesive has (E=3.0 GPa, ν=0.35, α1=45·10-6 ºC-1)
and the aluminium (E=68.67 GPa, ν=0.33, α1=24.5·10-6 ºC-1). The evaluation of the
characteristic exponents and the angular shape functions was performed by means of a
semi-analytical tool developed by Barroso et al [23], which is able to deal with
multimaterial corners including any number of linearly elastic anisotropic materials,
mathematically non-degenerate and degenerate in the framework of Stroh Formalism.
The in-plane orders of stress singularities for the corners in Figure 1, computed by this
tool, are shown in Table 1. All the characteristic exponents λk leading to singular stress
terms in (1) and the first regular (non-singular) term for each corner configuration have
been obtained, which is enough for an accurate description of the stress and
displacement fields at the corner tip, Barroso [25].

adhesive
metal
composite material

0o 0o adhesive 0o
adhesive
adhesive
adhesive metal adhesive
metal 90o

Figure 1.- Multimaterial corners in a composite to metal double lap joint.

It should be stressed that due to the weak singularities (Re(λ)>0.5) obtained for the
corners in Figure 1, which are summarized in Table 1, the most singular term alone is
not enough for an accurate representation of the stress field, except at an unrealistic
close distance to the corner tip, and all terms obtained in Table 1 have to be taken into
account.
adhesive
0o adhesive 0o
adhesive

0o
23º
adhesive adhesive
Aluminium 90o Al.
1 2 3 4 5
λ1=0.763236 λ1=0.986914 λ1=0.901497 λ1=0.686272 λ1=0.905312
λ2=0.889389 λ2=1.926197 λ2=1.01447 λ2=0.696605 λ2=1.700273
λ3=1.106980 λ3=0.791014
λ4=1.152813
Table 1.- Characteristic exponents of the multimaterial corners.

In Figure 2 the angular shape functions fijk and gik (i,j=r,θ) for corner 1 (see Table 1) and
for the first term (λ1=0.763236) are depicted, the first quadrant of each figure
representing the 0º layer of the composite and the rest representing the adhesive, the
circumference is the zero reference value for each function. For the sake of brevity not
all terms of all corners are shown, but there is huge information obtained from the stress
and displacement field divided into the terms of the series expansion in (1).
10
15 15
40
0º CFRP 0º CFRP 8
Adhesive 0º CFRP Adhesive 10
Adhesive 30 Adhesive 0º CFRP
10
6

20
4
5 5

10
2

0 0 0 0
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15 -40 -30 -20 -10 0 10 20 30 40
-10 -8 -6 -4 -2 0 2 4 6 8 10

-10 -2
-5 -5

-20
-4

-10 -10 gr 1
fθθ1 frθ1 frr1 -30
-6
gθ1
zero value zero value zero value -8
-40
-15 -15
zero value
-10

Figure 2.- Angular shape functions fijk and gik (i,j=r,θ) for the first term (λ1) in corner 1.
The contribution of the different terms can be analyzed and the possibility of a failure
criteria proposal based on allowable values of the Kk is then opened. The GSIF were
obtained for different configurations, shown schematically in Table 2. Both the
thickness of the aluminium (1.6 and 3.2 mm) plate and the composite laminate (1.6, 2.2
and 2.9 mm) were changed, also two different stacking sequences were used [0]n and
[0n/90n]s. Moreover one configuration had a different overlap length (20 mm instead of
12.5 mm). An aluminium-aluminium configuration was also analyzed.
All these configurations share the same multimaterial corner configurations shown in
Table 1, the differences in the global configuration (thickness, stacking sequence,
overlap length...) makes the Kk to change while λk, fijk(θ) and gik(θ) do not change as
locally the corners do not change, except for the Al-Al case.

[0]8 (1.6 mm) [0]12 (2.2 mm) [0]16 (2.9 mm)

Al (1.6 mm)
L=12.5 mm

Al (3.2 mm)
L=12.5 mm

Al (3.2 mm)
L=20 mm

[02,902]s (1.47 mm) [03,903]s (2.2 mm) Al (1.6 mm)

Al (1.6 mm)
L=12.5 mm

Al (3.2 mm)
L=12.5 mm
Table 2.- Different configurations numerically simulated using BEM.

Each configuration was numerically analyzed using a BEM code. As an example, a

detail of the left hand side of the overlap zone of the BEM model for configuration
Al(1.6 mm)/[03/903]s(2.2 mm), Loverlap=12.5 mm, is shown in Figure 3.

3.75 y x 3 plies - 0º

3.25

2.75 Deformed shape 6 plies - 90º

y (x20)

2.25

Adhesive 3 plies - 0º
1.75

Aluminium
1.25
46.00 46.50 47.00 47.50 48.00 48.50 49.00 49.50 50.00 50.50 51.00
x
Figure 3.- Detail of the BEM model, deformed and undeformed shape.
The basic features of the BEM model carried out are: linear elastic 2D plane strain,
perfect adhesion between the composite layers, the adhesive and the aluminium, a
progressive refinement of the mesh towards the corner tip using 10-6 and 10-8 mm at
different corners, uniform temperature effects included (due to the cure cycle of the
adhesive) and around 1500 linear elements used. Using a robust least squares procedure,
Barroso et al [26], the values of the Kk for each corner and for each configuration in
Table 1 were numerically computed and normalized following Pageau et al [27]. In this
procedure, the knowledge of λk, fijk(θ) and gik(θ), with a high accuracy, makes the Kk the
only unknowns in (1). The summary of Kk values for the corners 1, 2 and 3, see Table 1,
is shown in Table 3, the units being MPa·mm(1-λ). Also corner 1 for the Al-Al
configuration (#10) is included. Results for corner 3 is obviously only available when
the [0n/90n]s stacking sequence is used.

corner 1 (corner 1b ) corner 2 corner 3

config. Al Stack. Loverlap k1 k2 k3 k1 k2 k1 k2
1 [0º]8 12.5 -0.00275036 0.0273839 -0.0114328 0.000723935 0.000313232
2 3.2 [0º]12 12.5 -0.00265869 0.0296293 -0.0113959 0.000728197 0.000796011
3 [0º]16 12.5 -0.00267612 0.0319616 -0.0106351 0.000727391 0.00033738
4 [0º]8 12.5 -0.00427067 0.0387508 -0.0171819 0.00134598 0.00118731
1.6
5 [0º]12 12.5 -0.00396993 0.0391659 -0.0165938 0.00138476 0.00121834
6 [02/902]s 12.5 -0.00253126 0.0225756 -0.0100326 0.00088457 0.00017434 0.00357711 0.00188352
3.2
7 [03/903]s 12.5 -0.00258788 0.0263395 -0.0101892 0.000856521 0.000231622 0.00284841 0.00385512
8 [02/902]s 12.5 -0.00413257 0.0331505 -0.0158801 0.00168495 0.000276338 0.00481455 0.00140211
1.6
9 [03/903]s 12.5 -0.00418547 0.0347244 -0.0142997 0.00169283 0.000276204 0.00356942 0.00476516
10 3.2 Al 1.6 12.5 0.00151012 -0.0137057 0.0310423 0.00112665 -0.000596641
11 3.2 [0º]12 20.0 -0.00264264 0.028152 -0.0110226 0.000860062 0.000250501

Table 3.- GSIF values for the different corners and configurations.

3.- EXPERIMENTAL TESTING

Double-lap shear test were carried out following the requirements of ASTM D3528 [28]
for all configurations introduced in Table 2. Two surface treatments were employed for
the aluminium plates, namely: anodized and scaled (A and D in Table 4).
50 mm
50 mm

Table 4.- Detail of the double-lap shear test in tension, and summary of results.
For each configuration, 5 specimens were tested, the mean value of the apparent shear
strength (τR), the standard deviation and variation coefficient (VC%) having been
evaluated and presented in Table 4. Configurations 2 and 4 were observed to have high
dispersion (VC%) while the rest of configurations presented good VC values for
comparison purposes. Some additional tests were carried out at partial loadings (up to
90% of the observed mean value of the apparent shear strength) in order to check if a
preliminary damage at the corners can be observed prior to the final catastrophic failure.
Some of these inspections, for configurations 2 and 11 (both having 3.2 mm aluminium
and [0]12 unidirectional CFRP laminate, with the only difference in the overlap length,
respectively 12.5 and 20 mm) are shown in Figure 4. No damage has been observed in
the neighbourhood of the corner tips.

Adhesive
CFRP 0º
Adhesive

CFRP 0º

0.1 mm 0.1 mm

Aluminium Aluminium

Al(3.2 mm) / [0]12 Loverlap=12.5 mm Al(3.2 mm) / [0]12 Loverlap=20 mm

Figure 4.- Detail of corner #1 at partial loading for configurations 2 and 11.

4.- REPRESENTATIVITY OF THE SINGULAR STRESS STATE IN FAILURE

In this section the analysis of numerical and experimental results will be performed
first. Then, the comparison of tendencies in both numerical and experimental results
will be carried out to investigate the suitability of the parameters defining the local
(singular) stress state for its possible use in the proposals of failure criteria.

4.1.- Numerical results.

From Table 3 we can observe that:
• For corner #1, all Kk values are similar in 3.2 mm Al configurations and a
40%∼50% lower than Kk values for 1.6 mm Al configurations, which are also
similar among them. In the following table this fact can be clearly observed for
K1 values, the reference values for 3.2 and 1.6 mm configurations being
respectively configurations 1 and 4.

Configuration 1* 2 3 6 7 4* 5 8 9
Differences in K1 (%) 0 -3.33 -2.7 -7.97 -5.91 0 -7.04 -3.23 -2
* The reference for 2, 3, 6 y 7 is conf. 1, for 5, 8 y 9 is conf. 4

• For corner #2, the same trend is observed with same relative values for 3.2 mm
and 1.6 mm configurations, having similar values in their groups and 40%∼50%
differences between these two groups, higher values being observed for the 1.6
mm Al configurations. In the following table, values for K1 are presented.

Configuration 1* 2 3 6 7 4* 5 8 9
Differences in K1 (%) 0 0.589 0.477 22.19 18.31 0 2.881 25.18 25.77
* The reference for 2, 3, 6 y 7 is conf. 1, for 5, 8 y 9 is conf. 4
• Corner #3 appears only in these configurations having [0n/90n]s laminates, but
the same general trend is observed. It is also clear the influence of the number of
layers [02/902]s vs. [03/903]s in the value for K1 is higher for the [02/902]s
configurations. This effect was not observed in the [0]n laminates, the number of
plies not being relevant for significant variations of K1.
The observed tendencies can be easily justified:
• For corner #1, no significant changes appear with either the number of 0º plies
or the stacking sequence, but only with the Al thickness. A satisfactory
explanation is the high difference in the stiffness of both adherents. Three
unidirectional laminates [0]8, [0]12 and [0]16 were used. Shear stresses in the
adhesive are higher at the overlap end with the lower stiffness. The most
balanced configuration (Al 3.2 mm / CFRP [0]8) is still very unbalanced, and
additional increases in the number of CFRP plies just make the shear stress
distribution to be more severe. However, changes in the Al thickness, the softer
adherent greatly affect the stiffness ratio and the shear stress distribution.
• In corner #2 the explanation is basically the same, but the dependency observed
with the stacking sequence is due to the local load transmission through the
adhesive fillet, greatly affected by the presence of the 90º layer. Anyway, the
tendency is not as clear as in the former case.
Configurations 2 and 11 are equal with the only difference of the lap length. No
significant differences have been observed in the Kk values with the exception of corner
2, where K1 shows a 18% variation.

4.2.- Experimental results.

With reference to results in Table 4, no significant differences were observed in the
surface treatment (Anodized or Scaled) in the present static testing, although they could
be expected to appear in fatigue testing. There is a clear influence of the Al thickness in
the experimental results, which can be observed in the configuration pairs 6-8, and 7-9.
In the following table, the results obtained for τR have been compared for [0n/90n]s
(n=2,3) configurations with 1.6 and 3.2 mm Al. The values are 15%∼26% lower in
configurations having 1.6 mm Al.

[02/902]s-Al(A) [03/903]s-Al(A) [02/902]s-Al(D) [03/903]s-Al(D)

Al. 3.2 mm 25.01 26.38 25.56 25.56
Al. 1.6 mm 18.43 (↓26%) 20.91 (↓21%) 19.70 (↓23%) 21.75 (↓15%)

The influence of the stacking sequence on the experimental results is significant for
configurations having 3.2 mm Al thickness, while this influence is not clear in
configurations with 1.6 mm Al, as shown in the following table. In the former case,
configurations with [0n/90n]s (n=2,3) show higher (10%∼21%) τR values when
compared with the corresponding [0]n (n=8,12) configurations.

[0]8 [02/902]s [0]12 [03/903]s

Al A 22.26 25.01 (↑12%) 21.85 26.38 (↑21%)
(3.2 mm) D 22.03 25.56 (↑16%) 23.30 25.56 (↑10%)
Al A 22.03 18.43 21.76 20.91
(1.6 mm) D 17.97 19.70 24.44 21.75
The influence of the overlap length on the experimental results is clear: the failure load
F(KN) clearly increases (71%∼93%) from Loverlap=12.5 mm to Loverlap=20.0 mm, as well
as the apparent shear strength (τR) (6%∼20%). In the following table, all these results
are summarized.
Al(3.2)-[0]12 Al(3.2)-[0]12
F(KN)/τR(MPa)
(Loverlap=12.5 mm) (Loverlap=20.0 mm)
Aluminium A 13.8/21.85 26.6 (↑93%)/26.20 (↑20%)
Aluminium D 14.7/23.30 25.1 (↑71%)/24.75 (↑6%)

4.3.- Numerical-experimental correlation

Surface treatment: No significant influence was observed in the experimental results of
anodized or scaled aluminium plates under static load. The surface treatment was not
included in the numerical models as they do not modify the stiffness properties of the
materials.
Aluminium thickness: If configurations 2 and 4 are excluded due to their high VC%, the
influence on the mean apparent shear strength was clearly observed between the
configuration couples 6-8 and 7-9, which only differ in the aluminium thickness. The
experimental variation between these configurations was around 15%∼26%. The
numerical models also presented a clear influence 40%∼50% in the computed values of
Kk. Thus, a qualitative correlation exists between experimental and numerical results.
CFRP thickness: Neither the experimental nor the numerical results show a significant
influence on the laminate thickness. This lack of influence was explained by the high
unbalance between both adherents in the lowest case with [0]8 laminate, this unbalance
was even higher with [0]12 and [0]16 which do not change qualitatively the unbalance of
the joint. It is expected that more balanced joints would be more affected by initial
changes in the CFRP thickness.
Stacking sequence: With the same number of total plies, the stacking sequence affects
the tensile stiffness drastically. This fact is slightly observed in the numerical results,
when comparing configurations 1-6, 2-7 and 4-8, as K1 decreases with [0n/90n]s
laminates instead of [0]n. Experimentally, a higher failure load was observed in the
[0n/90n]s configurations. Thus the same qualitative trend is observed between the
numerical and experimental results.
Overlap length: The comparison of the results of configurations 2 and 11, which only
differ in the overlap length, was not significant in the numerical results but very clear in
the experimental results. The configuration with 20 mm overlap has a higher failure
load. Partial load test did not show any failure at the corner tip. Thus, in this case, the
qualitative trend between numerical and experimental results is not the same. Using
Hart-Smith [2] approach of nominal stress values, it is obtained that only the 40% of the
overlap length remain elastic in the 20 mm configuration while the 99.7% is elastic in
the 12.5 mm configuration. This fact shows that configuration 11 (Loverlap=20mm) with
gross yielding at both sides of the overlap zone can not fully develop a failure
mechanisms based on local (singular) stress states.
5.- CONCLUSIONS
The failure of adhesively bonded joints between aluminium plates and CFRP laminates
has been analyzed. Using the asymptotic fields (stresses and displacements) at the
multimaterial corners as the key factor for failure initiation, a complete characterization
of these corners was performed using a semi-analytic tool for the evaluation of
characteristic exponents and angular shape functions and using numerical models (based
on the Boundary Element Method) for the computation of the Generalized Stress
Intensity Factors.
A preliminary test program was carried out using double lap shear joints loaded in
tension to evaluate static failure loads and apparent shear strengths. Different
combinations of Al-CFRP double lap joints were tested to study the influence on the
failure load of: Al thickness, CFRP thickness, stacking sequence and overlap length.
Numerical predictions and experimental results were compared to study the suitability
of a failure criteria based on the parameters which define the singular stress state at
these corners. The numerical-experimental comparison has shown qualitatively the
same trends with every parameter variation. Only one exception was found with overlap
length variation, while a clear variation was observed in the experimental results, no
significant variation was obtained in the numerical models. Using an elastic-plastic
model by Hart-Smith, the yielded zone in the overlap area was determined to be in a
great extent (60%) of the lap distance. With the presence of this gross yielding, failure
mechanisms based on local stress states can not be developed. The numerical-
experimental correlation was found to exist with low yielding extent in the adhesive
layer. This fact allows failure criteria based on singular stress states to be used in this
type of structures.

ACKNOWLEDGEMENTS.
This work was supported by the Spanish Ministry of Education and Science, through
the Projects TRA2005-06764 and TRA2006-08077, and by the Junta de Andalucía,
through the Projects of Excellence TEP1207 and TEP2045.

REFERENCES
1. Matthews,F.L.,Kilty,P.F. & Godwin,E.W.“A review of the strength of joints in fibre
reinforced plastics.Part 2.Adhesively bonded joints”, Composites ,1982;13:29-37.
2. Hart-Smith, L. J., “Analysis and design of advanced composite bonded joints”,
NASA CR-2218, 1974.
3. Leman, G. M. and Hawley, A. V. “Investigations of joints in advanced fibrous
composites for aircraft structures”, Technical Report No AFFDL-TR169-43 Vol 1
(US Air Force, June 1979).
4. Crocombe, A. D. “Global yielding as a failure criterion for bonded joints”,
International Journal of Adhesion and Adhesives, 1989;9:145-15 .
5. Kairouz, K. C. “The influence of stacking sequence on the strength of bonded CFRP
joints”, PhD Thesis (Imperial College, London), 1991.
6. Adams, R. D., Comyn, J. and Wake, W. C., Structural adhesive joints in
engineering (second edition), Chapman & Hall, 1997.
7. Adams, R. D. and Mallick, V. “A method for stress analysis of lap joints”, Journal
of Adhesion, 1992;38:199-217.
8. Fernlund, G. and Spelt, J. K. “Failure load prediction of structural adhesive joints”,
International Journal of Adhesion and Adhesives, 1991;11: 213-220.
9. Hutchinson, J.W. and Suo, Z., Advances in Applied Mechanics 1992;29:63-191.
10. Gradin, P. “A fracture criterion for edge bonded bimaterial bodies”, Journal of
Composite Materials, 1982;16: 448-456.
11. Groth, H. L. “Stress singularities and fracture at interface corners in bonded joints”,
Int. Journal of Adhesion and Adhesives, 1988;8: 107-113.
12. Hattori, T. “A stress-singularity-parameter approach for evaluating the adhesive
strength of single lap joints”, JSME Int. Journal, Series I, 1991;34:326-331.
13. Reedy, Jr., E. D. “Connection between interface corner and interfacial fracture
analyses of an adhesively bonded butt joint”, International Journal of Solids and
Structures, 2000;37:2443-2471.
14. Reedy, Jr., E. D. and Guess, T. R. “Interface corner failure analysis of joint strength:
effect of adherent stiffness”, I. Journal of Fracture,1998;94: 305-314.
15. Lefebvre, D. R., Dillard, D. A. and Dillard, J. G. “A stress singularity approach for
the predictions of fatigue crack initiation in adhesive bonds. Part 2: Experimental”,
Journal of Adhesion, 1999;70: 139-154.
16. Quaresimin, M. and Ricotta, M. “Fatigue behaviour and damage evolution of single
lap bonded joints in composite materials”, Comp. Sci. Technology,2006;66:176-187.
17. Leguillon, D. and Yosibash, Z. “Crack onset at a v-notch. Influence of the notch tip
radius”, International Journal of Fracture; 2003;122:1-21.
18. Grenestedt, J. L. and Hallstrom, S. “Crack initiation from homogeneous and
bimaterial corners”, J. of Applied Mechanics,1997;64: 811-818.
19. Laschet,G. and Stas,A. “Finite element failure prediction of adhesive joints using a
simple damage model”,Report SA-158,Str.Aer.(University de Liege,Belgique),1992.
20. Sheppard, A., Kelly, D. and Tong, L. "A damage zone model for the failure analysis
of adhesively bonded joints", Int. J. Adhesion and Adhesives,1998;18: 385-400.
21. Mohammed, I. and Liechti, K. M. “Cohesive zone modelling of crack nucleation at
bimaterial corners”, J. of the Mechanics and Physics of Solids,2000;48:735-764.
22. Costabel, M. Dauge, M. and Lafranche, Y., “Fast semi-analytic computation of
elastic edge singularities”, Université de Rennes I, 1998.
23. Barroso, A., Mantič, V. and París, F. “Singularity analysis of anisotropic
multimaterial corners”, Int. Journal of Fracture,2003;119: 1-23.
24. Sinclair, G.B. “Logarithmic stress singularities resulting from various boundary
conditions in angular corners of plates in extension”, J.Appl.Mech,1999;66:556-559.
25. Barroso, A. “Characterization of singular stress states at multimaterial corners.
Application to adhesively bonded joints with composite materials”, Ph.D. Thesis (in
Spanish), University of Seville, Spain, 2007.
26. Barroso, A., Mantič, V. and París, F. “Evaluation of generalized stress intensity
factors in anisotropic elastic multimaterial corners”, ECCM-11, Rodhes (Greece),
2004, (on CD, paper A032).
27. Pageau, S. P., Gadi, K. S., Biggers, Jr., S. B. and Joseph, P. F. “Standardized
complex and logarithmic eigensolutions for n-material wedges and junctions”,
International Journal of Fracture,1996;77: 51-76.
28. ASTM D3528-96 (2002) Standard Test Method for Strength Properties of Double
Lap Shear Adhesive Joints by Tension Loading.