You are on page 1of 12

International Journal of Adhesion & Adhesives 31 (2011) 434–445

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives


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

Closed-form solutions for elastic stress–strain analysis in unbalanced


adhesive single-lap joints considering adherend deformations and
bond thickness
Bo Zhao n, Zhen-Hua Lu, Yi-Ning Lu
Department of Automotive Engineering, Tsinghua University, State Key Laboratory of Automotive Safety and Energy, Beijing 100084, PR China

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

Article history: A theoretical model is developed for the stress analysis in adhesive-bonded single-lap joints under
Accepted 26 February 2011 tension, for which the two adherends could have different thicknesses and consist of different
Available online 9 March 2011 materials. A two-dimensional (2D) elasticity theory is adopted in the analysis, which simultaneously
Keywords: incorporates the complete strain–displacement and the complete stress–strain relationships for the
Stress analysis adherends and adhesive. The approach provides a unified treatment for any possible adhesive layer
Lap-shear flexibility and capable of satisfying the stress-free condition at the ends of the bondline. An explicit
Analytical stress analysis closed-form analytical solution is formulated for upper and lower adherends/adhesive stresses (strains)
Simple formulae and tensile, shear and bending loads acting on the adherends along the overlap and then simplified for
Unbalanced joint
practical applications, and simple design formulae for adhesive stresses are produced. The results
predicted by the present full and simplified solutions were compared with the previously theoretical
solution by Bigwood and Crocombe (1989) [35], and the 2D geometrically nonlinear finite element
model using MSC/NASTRAN. The agreement validates the present formulation and solutions for
unbalanced bonded joints. The effects of the stiffness unbalanced parameters on the adhesive stress
distributions were also discussed.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction study the stress distribution in adhesive joints with dissimilar


adherends, for which the two adherends could have different
Adhesive bonded single-lap joints are commonly used to thicknesses and consist of different materials. Hence, this analysis
connect two dissimilar components together in the manufactur- is more general than that of the joints with similar adherends.
ing of lightweight structures such as automobile, train and Balanced adhesive joints have been considered by many
aircraft vehicles due to their superior advantages in comparison investigations such as Refs. [1–19], and others. A brief review of
with other joining technologies [1]. For instance, the fuselage of the previous publications by the authors is presented in Ref. [1].
Boeing’s 787 passenger aircraft is the large structures where In order to derive a closed-form solution for the stress distribu-
composites are adhesively bonded with metals. It is more impor- tion in an adhesive bonded joint, various simplified assumptions
tant than ever for engineers to be able to design and analyze concerning the behavior of the adhesive and adherends have been
unbalanced adhesive joints. In engineering applications, the made in many existing models. Volkersen [2] developed a shear-
methods for determining the stresses in adhesive bonded joints lag model in riveted joints in 1938. This approach has the
therefore need to be applicable to a wide range of both adherend advantage of simplicity, but it ignores the presence of peel stress,
and adhesive material properties. Generally an adhesive bonded which is a key contributor to the failure of adhesive bonded
structure consists of three components with different thicknesses structures. A peel stress field was first captured in the well-
and materials, namely, the two adherends and the adhesive. known work by Goland and Reissner (G–R model) [3] whereby
Because of their nonhomogeneous nature of the material and the behaviors of balanced single-lap joints with rigid and flexible
the fact that it is subject to numerous rigorous boundary condi- adhesives were analyzed, respectively. This model is considered
tions, the exact theoretical treatment of this kind of sandwich to be one of the most significant models of bonded connections,
plate problem is hopelessly complicated. This paper is intended to but it has limitations of the inability to control the boundary
conditions of the adhesive layer and the assumption that the
stresses do not vary through the thickness [1,4–8]. To predict
n
Corresponding author. Tel.: þ86 1062781006; fax: þ 86 1062785708. more accurate stress distribution of the adhesive joint, unceasing
E-mail address: bozhao@tsinghua.org.cn (B. Zhao). attempts to refine G–R model has been proposed by many

0143-7496/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijadhadh.2011.03.002
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 435

researchers [9–15,19] over the ensuing six decades; most notably complexity in such sandwich structure [26,30–32]. Bigwood and
Hart-Smith [9]. Refs. [5,10,19–21] evaluated various theories, and Crocombe [35] (called the B–C solution) derived the simple design
resolved some controversies and inconsistencies between the expressions of unbalanced single-lap joints through the treatment
theories. of ignoring the coupling between the shear and peel adhesive
In practice, adhesive-bonded joints can have a small amount of stresses. Recently, Wang and Zhang [30] developed a three-
material, which squeezes out of the bondline during manufactur- parameter, elastic foundation model where the interfacial adhe-
ing. When this excess material remains and is allowed to cure, it sive peel stress can be predicted and the G–R type models’ violation
is referred to as a spew-fillet, and an adhesive layer without of the equilibrium condition of adhesive layer is removed. Refs.
spew-fillets is referred to as a square-end. Spew-fillets have been [4,16,17,27,31–34] investigated the case of composite adherends and
predicted analytically, numerically and experimentally to have a recently, Luo and Tong [16,17] considered the geometrically non-
beneficial effect on the failure strength of adhesive-bonded linear effect of composite adherends. The authors presented the 2D
joints [4,22–25]. It was suggested that the G–R model reasonably explicit closed-form analytical solutions of adhesive-bonded single-
predict the stress state in the adherend overlap region of a spew lap joints [1] where the treatment was limited to balanced joints due
filleted adhesive layer from these studies. Thus, although they to the overwhelming complexity of obtaining exact closed-form
may have the inability to satisfy zero shear stress condition at the solutions in the case of unbalanced joints. The derivation in this
free edges of the adhesive layer, the G–R type models are still study follows the same assumptions as Ref. [1], considering that the
used by some researchers [5]. adhesive shear stress remains constant across the thickness of the
However, the situation is more complicated if the spew-fillets adhesive layer to obtain closed-form analytical expressions for
are absent. It has been demonstrated from analytical solu- unbalanced single-lap joints [4,5,26].
tions [1,4–8] that a square-ended configuration gives rise to To our best knowledge, it seems that 2D theoretical models of
severe shear stress gradients and peel stress peaks in the vicinity unbalanced joints reported in the literature have less attempts to
of the traction free edges. One of the major limitations of the G–R provide an explicit closed-form solution including the stress,
model and its descendants is the inability to strictly satisfy the strain and in-plane load distributions, and such a solution can
boundary condition of zero shear stress at the free edges of the be followed and implemented conveniently by other researchers.
adhesive layer. According to their analysis, the maximum shear The objective of this investigation is to extend our previous
stress occurs at both ends of the joint instead of actually studies and to derive the full and simple design expressions for
disappearing there. This unavoidably has adverse effects on the elastic stresses in the unbalanced adhesive bonded single-lap
distribution of the maximum adhesive stresses, which are highly joints. It is based on 2D elasticity theory, which simultaneously
responsible for joint failure. Although spew-fillets are frequently includes the complete strain–displacement and the complete
present in real structures, it is more conservative to use the stress–strain relationships for the adherends and adhesive layer.
square end adhesive layer idealization for design purposes, as this Then, through the use of displacement continuity conditions at
has been shown to be the worst case scenario for the peak the interfaces between the adhesive and adherends, an explicit
adhesive stresses [5]. This deficiency was overcome to some extent closed-form solution was obtained that is capable of satisfying the
in Refs. [4–8,26–29]; however, their final numerical solutions are zero shear stress condition at the adhesive free edges. The results
cumbersome and difficult to apply in practical situations. predicted by the present full and simplified solutions were com-
Another major drawback of the G–R type models is that the pared with the previously theoretical solution by Bigwood and
adhesive shear and peel stresses in those theories were generally Crocombe (B–C solution), and the 2D geometrically nonlinear finite
assumed to be uniform across the thickness. However, it is element (FE) model using MSC/NASTRAN. The agreement noted
indicated that adherend failure modes are closely related to the validates the present formulation and solutions for unbalanced
magnitude of the through thickness adhesive peel stresses at the bonded joints. Finally, the effects of the stiffness unbalanced para-
adhesive/adherend interfaces, and an interfacial peel stress near meters on the adhesive stress distributions were also discussed.
the ends of the adhesive layer is critical to evaluating the
potential of debonding and predicting where the debonding can
initiate [5,30]. Furthermore, there is a trend towards the use of 2. Mathematical formulation
substantially thick bond layers (up to  4 mm), for which there is
little data available about the performance of the structure [14]. The 2.1. Equilibrium equations
shear stress in Refs. [12,14] and peel stress in Refs. [4–6,14,15,26,
28,30] were allowed to vary through the thickness of the adhesive, Consider two adherends with the thicknesses of t1 and t2
respectively. bonded together in a single-lap adhesive bonded joint configura-
On the other hand, the connection of two dissimilar adherends tion (see Fig. 1) by an adhesive layer with a thickness of t3. The
is one of the most significant advantages for adhesive bonded length of the overlap region of the two adherends is 2l and the
joints, and the literature with an explicit closed-form solution of non-overlap region length is donated as L. The joint is assumed to
unbalanced adhesive joints is very limited [26–38]. The analysis have a unit width in a direction perpendicular to the (x, y)-plane.
for this kind of unbalanced joint problem is more general and The two unjoined ends of the adherends are simply supported
complicated than that of the balanced joints. However, an explicit and acted on by the tensile load F. The free body diagrams of
formulation is usually difficult to obtain due to the mathematical an infinitesimal element in the overlap region are shown in

Fig. 1. Geometric profile of an unbalanced single-lap adhesive joint under tension.


436 B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445

Fig. 2(a) and (b). The subscripts 1, 2 and 3 denote the upper adherends six (i¼1, 2) may be assumed to vary linearly with the
adherend, lower adherend and adhesive, respectively. Then, let Ei, corresponding transverse coordinate yi (i ¼1, 2), respectively, i.e.,
Gi and mi (i¼1, 2, 3) be the Young’s modulus, shear modulus and
six ¼ fi =ti þ 6Mi ð2ri 1Þ=ti2 , t3 ¼ t3 ðxÞ ¼ tm ðxÞ, ð3Þ
Poisson’s ratio of the three individual components in the joint,
and (six, siy, ti) (i¼1, 2, 3), their stress components, respectively. where 0ryi rti, ri ¼yi/ti (i¼1, 2). From the 2D elasticity theory,
For convenience, three local coordinate systems, xOy1, xOy2 and these stress components should satisfy the equations of equili-
xOy3, were introduced, where the origins of y1 and y2 were brium:
located at their upper surfaces of the upper and lower adherends, @six @ti @siy @ti
respectively, and the origin of y3 was located in the middle plane þ ¼ 0, þ ¼ 0, ði ¼ 1,2,3Þ: ð4Þ
@x @y @y @x
of the adhesive layer. Let su3y and sl3y be the peel stresses at the
upper and lower interfaces of the adhesive, and fi, Qi and Mi (i ¼1, Substituting Eq. (3) into Eq. (4) and introducing Eqs. (1) and
2, 3) be the tensile force, shear force and bending moment of the (2), the shear stress components generated by the integration are
three individual components in the joint, respectively. Qi and Mi identified by imposing the conditions of continuity of stress at the
(i¼1, 2) are the shear force and bending moment in the upper and interfaces between the adhesive and adherend, and the stress-
lower adherends acting at the end of the joint, respectively, with free conditions on the top and bottom surfaces of the joint. Then
the positive sign convention shown in Fig. 2. The values of Qi and one obtains
(
Mi (i ¼1, 2) in the unbalanced joints can be obtained from Cheng t1 ¼ t3 ð3r21 2r1 Þ6Q1 =t1 ðr21 r1 Þ,
et al. [26]. From the equilibrium equations of the adhesive, M3 ¼0, ð5Þ
t2 ¼ t3 ð3r22 4r2 þ 1Þ6Q2 =t2 ðr22 r2 Þ:
Q3 ¼ t3t3 and f3 ¼ s3xt3 can be derived.
By referring to Fig. 2(b), the global equilibrium equations and Similarly, substituting Eq. (5) into Eq. (4), the peel stresses in
boundary conditions can be obtained: the adherends and adhesive yield
8 8
df1
>
>
> þ t3 ¼ 0, df 2
t3 ¼ 0, >
> s1y ¼ ddxt3 t1 ðr31 þ r21 Þ þ su3y ð2r31 þ 3r21 Þ,
< dx dx >
<
dQ1 u dQ2 l d t3 u l
dx þ s3y ¼ 0, dx s3y ¼ 0, dx t 3 ðs3y s3y Þ ¼ 0, ð1Þ s2y ¼ ddxt3 t2 ðr32 þ 2r22 r2 Þ þ sl3y ð2r32 3r22 þ 1Þ, ð6Þ
>
> >
>
> dM
: 1 Q þ t 1 ¼ 0,t dM 2 t
Q þ t 2 ¼ 0, > s ¼ su  dt3 t ðr þ0:5Þ,
:
dx 1 3 2 dx 2 3 2 3y 3y dx 3 3

(
x ¼ l : M1 ¼ M1 , Q1 ¼ Q1 , f1 ¼ F, M2 ¼ Q2 ¼ f2 ¼ t3 ¼ 0, where 0 r ri r1 (i¼1, 2),  0.5r r3 (¼ y3/t3)r0.5.
ð2Þ It can be seen from Eqs. (5) and (6) that the shear stress
x¼l : M2 ¼ M2 , Q2 ¼ Q2 , f2 ¼ F, M1 ¼ Q1 ¼ f1 ¼ t3 ¼ 0:
distribution in the adherends is quadratic, and the peel stress is
The adhesive shear stress is assumed constant (tm) through cubic through the thickness. From the third one of Eq. (6), the
the thickness [5], and the longitudinal normal stresses in the adhesive peel stress at the middle plane (r3 ¼0), sm, at the upper
interface (r3 ¼ 0.5), su3y , and at the lower interface (r3 ¼0.5), sl3y ,
can be easily obtained, then the interfacial peel stresses su3y and
sl3y can be rewritten as
dtm t3 dtm t3
su3y ¼ sm þ , sl3y ¼ sm  : ð7Þ
dx 2 dx 2

2.2. Stress–strain and strain–displacement equations

Let ei, gi, ui and vi (i¼1, 2, 3) be the normal and shear strain and
the longitudinal and transverse displacement, respectively. The
state of plane stress is assumed for convenience here. When a
plane strain state is considered, Ei and mi should be replaced with
Ei =ð1-m2i Þ and mi/(1  mi)(i¼1, 2, 3). The stress–strain equations of
the adhesive and the adherends can be written as
ti
gi ¼ ði ¼ 1,2,3Þ, ð8Þ
Gi

@vi
siy ¼ Ei þ mi six ¼ Ei eiy þ mi six ði ¼ 1,2Þ, ð9Þ
@yi

@ui
six ¼ Ei þ mi siy ¼ Ei eix þ mi siy ði ¼ 1,2Þ: ð10Þ
@x
Consider the case of i¼1, namely the upper adherend. After
some algebraic manipulation and substitution of Eqs. (3) and (6)
into Eq. (9), and integrating qv1/qy1 with respect to y1, the
transverse displacement of the upper adherend v1 becomes
Z r1   
t1 t1 dt3 r4 r3
v1 ¼ v1u þ ðs1y m1 s1x Þdr~ 1 ¼ v1u þ t1  1 þ 1
0 E1 E1 dx 4 3
 4   
r m 6M 1
þ su3y  1 þ r31  1 f1 r1 þ ðr21 r1 Þ , ð11aÞ
2 E1 t1
where v1u is the transverse displacement at the upper surface of
Fig. 2. (a) Global equilibrium along the overlap, and (b) free-body stress the upper adherend. According to the condition of displacement
equilibrium diagram. continuity, and taking Eq. (11a) into account, the transverse
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 437

displacement of the upper adherend at the interface between the Note that all the axial normal stress components must satisfy
adhesive and the upper adherend v1l can be expressed as the following definitions:

t1 dt3 t1 su3y  m1 Z 1 Z 1
v1l ¼ v1 ðr1 ¼ 1Þ ¼ v1u þ þ  f1 : ð11bÞ Mi ¼ six ðri 0:5Þti2 dri , fi ¼ six ti dri ði ¼ 1,2Þ: ð17Þ
E1 dx 12 2 E1
0 0
Using Eq. (11b), Eq. (11a) can be rewritten as After substituting Eqs. (15), (16) and peel stress expressions (6)
into stress–strain Eq. (10), and taking Eq. (1) into consideration,
    
t1 dt3 r4 r3 1 r4 1 and substituting the new longitudinal normal stresses (10) into
v1 ¼ v1l þ t1  1 þ 1  þ su3y  1 þ r31  the above Eq. (17), then one obtains
E1 dx 4 3 12 2 2
  8 2 u
> d2 v 11t2 d3 t 13t d s3y 12su3y
m 6M1 2 > 12M m 0:2 dt
< dx21l ¼  E1 t31 þ 210E11 dx33 þ 1 E1 dx3 þ 35E11 dx2  5E1 t1 ,
 1 f1 ðr1 1Þ þ ðr1 r1 Þ : ð11cÞ
E1 t1 1
2 l ð18aÞ
>
>
2 11t22 d3 t3 m2 0:2 dt3 13t2 d s3y 12sl3y
In exactly the same way as for the upper adherend, the : ddxv22u ¼  12M
E t 3
2
þ 210E 2 dx
3 þ E2 dx
 35E2 dx
2 þ 5E t
2 2
,
2 2

transverse displacement function v2 in the lower adherend is


8 u 2
> t3 3 11t2 d s3y
 
r4 2r32 r22
  4
r
 > du
< dx1l ¼
f1
þ 6M 1 1
 105E d t3 4t1 dt3 1
3 þ 15E dx  210E  m1 0:2 su3y ,
t2 dt3 E1 t1 E t21 1 1 dx 1 1 dx2 E1
v2 ¼ v2u þ t2  2 þ  þ sl3y 2 r32 þ r2 2
E2 dx 4 3 2 2 > t3 3 l
11t2 d s3y
>
: dudx2u ¼ f2
 6M 2 2
þ 105E d t3 4t2 dt3 2
3  15E dx  210E  m2 E0:2 sl3y :
  E2 t2 E t22 2 2 dx 2 2 dx2 2
m 6M2 2
 2 f2 r2 þ ðr2 r2 Þ , ð12Þ ð18bÞ
E2 t2
where v2u is the lower adherend transverse displacement at the
interface. Note that, due to the perfect bonding of the joints, as 2.3. Governing differential equations
shown in Fig. 2(a), the displacement functions are continuous at
the interfaces between the adhesive and adherends. As a result, It is assumed that the longitudinal and transverse displace-
the upper adherend transverse (or longitudinal) displacement at ments of the adhesive linearly vary with the transverse coordi-
the interface, v1l (u1l), should be equivalent to the adhesive nate for thin and moderate thick bonding joints [12,14], that is
transverse (or longitudinal) displacement, v3u (u3u), at the inter- (
u3 ¼ 0:5ðu2u þ u1l Þ þ r3 ðu2u u1l Þ,
face between the adhesive and the upper adherend, and the lower ð19Þ
v3 ¼ 0:5ðv2u þv1l Þ þ r3 ðv2u v1l Þ,
adherend transverse (or longitudinal) displacement at the inter-
face, v2u (u2u), should be the same as the adhesive transverse (or where  0.5r r3 r0.5.
longitudinal) displacement, v3l (u3l), at the interface. That is: Substituting Eq. (19) into shear strain–displacement equa-
v3u ¼ v1l , u3u ¼ u1l , v3l ¼ v2u , u3l ¼ u2u : ð13Þ tion (14) and taking Eq. (8) and s3x ¼ 0 into account yields
  
The shear strain–displacement equations for the adhesive and ðu u Þ 1 dv2u dv1l
tm ¼ t3 ðy3 ¼ 0Þ ¼ G3 2u 1l þ þ , ð20Þ
adherends are written as t3 2 dx dx

@ui @vi @v3 E3


gi ¼ þ , ði ¼ 1,2,3Þ: ð14Þ sm ¼ s3y ðy3 ¼ 0Þ ¼ E3 ¼ ðv2u v1l Þ: ð21Þ
@yi @x @y3 t3
Differentiating Eq. (20) with respect to x once and differentiat-
After substituting Eq. (11c) into Eq. (14), and taking Eqs. (8),
ing Eq. (21) with respect to x twice, and taking Eqs. (18a) and
(5) and (1) into account, and integrating the function qu1/qy1 with
(18b) into account, and eliminating su3y and sl3y with Eq. (7), then
respect to y1, then considering the continuous displacement
one obtains
condition, the longitudinal displacement in the upper adherend 8 6M ð1 þ t =t Þ 6M ð1 þ t =t Þ
u1 can be expressed by the interfacial longitudinal displacement < ða1 D3 a2 DÞtm þ ða7 D2 a8 Þsm ¼ E1f1t1  Ef22t2 þ 1 E1 t2 3 1 þ 2 E2 t2 3 2 ,
1 2
of the upper adherend u1l: : ða7 D3 a8 DÞtm þ ða4 D2 a5 Þsm ¼ 12M1
 12M 2
:
E1 t13 E t3
2 2
"  4 
dv1l t12 d2 t3 r1 r51 r1 1 ð22Þ
u1 ðr1 Þ ¼ u1l  t1 ðr1 1Þ t 1   þ
dx E1 dx2 12 20 12 20
where the differential operator D ¼d/dx. Differentiating Eq (22)
 #   with respect to x and making use of Eq. (1) yields
dsu3y r41 r51 r1 7 1 m1
þ   þ þ t3 t1  ðr31 r21 Þ 8
> 4 2 3 6Q ð1 þ t =t Þ 6Q ð1 þ t =t Þ
< ða1 D a2 D þ a3 Þtm þ ða7 D a8 DÞsm ¼ 1 E t2 3 1 þ 2 E t2 3 2 ,
dx 4 10 2 20 G1 E1 0
   1 1 2 2

m 1 m1 > 4 2 3 12Q1
 12Q
: ða7 D a8 D þ a9 Þtm þ ða4 D a5 DÞsm ¼
0
þ 1 ðr1 1Þ Q1 ð2r31 3r21 þ 1Þ: :
2
 ð15Þ E1 t13 E t3
E1 G1 E1 2 2

ð23Þ
In exactly the same way as for the upper adherend, the
longitudinal displacement function u2 in the lower adherend where the constants ai (i¼1  9) are given in Appendix A. a03 and
can be expressed as follows: a09 are of no use in the subsequent deductions, so their expres-
Z r2   sions are not provided.
@v2 dv2u Differentiating Eq. (23) with respect to x and using Eq. (1)
u2 ðr2 Þ ¼ u2u þ g2 ðr~ 2 Þ t2 dr~ 2 ¼ u2u  t2 r2
0 @x dx leads to two coupled, fifth-order ordinary differential equations
"    # with constant coefficients for the unknown stresses tm and sm :
2
1 d t3 3 r42 r52 r32 dsl3y 2 r52 r42 r22
 t   þ t  þ (
E2 dx2 2
6 20 6 dx 2
20 4 2 ða1 D5 a2 D3 þ a3 DÞtm þ ða7 D4 a8 D2 þ a9 Þsm ¼ 0,
   ð24Þ
1 m2 m ða7 D5 a8 D3 þ a9 DÞtm þ ða4 D4 a5 D2 þ a6 Þsm ¼ 0:
þ t3 t2  ðr32 2r22 þ r2 Þ þ 2 r2
G2 E2 E2
  Elimination of either of tm or sm from Eq. (24) leads to a nine-
1 m2 3 2
Q2  ð2r2 3r2 Þ: ð16Þ order differential equation, whose solution is feasible employing
G2 E2
Ferrari’s classical solution, and the general solution of Eq. (24)
438 B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445

may be expressed as follows: balanced joints. Thus it is necessary that the simplified design
8 formulae are developed which may be applied to both the unba-
>
> tm ¼ C1 cosh l1 x þ C2 sinh l1 x þ C3 cosh l2 x þC4 sinh l2 x
>
> lanced and balanced joints.
>
> þcosh l3 xðC5 cos l4 x þ C6 sin l4 xÞ
>
> It has been demonstrated that the coupled coefficients ai(i¼7–9)
>
< þsinh l3 xðC7 cos l4 x þ C8 sin l4 xÞ þ C9 , in Eq. (24) are relatively small in most cases in the unbalanced joints,
ð25Þ
>
>
>
sm ¼ D1 sinh l1 x þD2 cosh l1 x þ D3 sinh l2 x þ D4 cosh l2 x and they turn into zero in the balanced joints. Thus the coupled
>
>
>
> þcosh l3 xðD5 cos l4 x þ D6 sin l4 xÞ effects of adhesive shear and peel stresses may be neglected, which
>
>
: þsinh l3 xðD7 cos l4 x þ D8 sin l4 xÞ, has been verified feasible in the B–C solution [35]. That is, ai(i¼7–
9)¼0 in the full solution, then Eq. (24) is rewritten as:
where Ci (i¼1–9) and Dj (j¼1–8) are the integration constants (
and they are not independent, and 7 l1, 7 l2 and 7 l3 7il4(here ða1 D5 a2 D3 þ a3 DÞtm ¼ 0,
ð27Þ
i is imaginary unit) are eight eigenvalues of the characteristic ða4 D4 a5 D2 þ a6 Þsm ¼ 0:
equation, and the detailed solving procedure can be seen in
Generally, the solutions of Eq. (27) may be expressed as
references [1,18].
follows:
The complete solutions of the governing differential equations
for tm and sm have nine unknown integration constants to be tm ¼ C1 cosh l1 x þ C2 sinh l1 x þ C3 cosh l2 x þC4 sinh l2 x þ C5 ,
determined by the imposition of appropriate boundary condi- sm ¼ cosh l3 xðD1 cos l4 x þ D2 sin l4 xÞ
tions. It is noted that the present approach yields the explicit þsinh l3 xðD3 cos l4 x þD4 sin l4 xÞ, ð28aÞ
closed-form analytical solutions. where the constants li, Cj, Di (i¼1  4, j ¼1  5) are given
in Appendix B.
2.4. Boundary condition equations Specifically, taking the case of balanced joints with identical
adherends, i.e., E1=E2=E, G1=G2=G and t1=t2=t into account, then
From Eqs. (1) and (2), (22) and (23), the following equations C2 ¼0, C4 ¼0, D2 ¼0, D3 ¼0, and Eq. (28a) can be rewritten as
which can satisfy the rigorous boundary conditions are given by (
Z l tm ¼ C1 cosh l1 x þ C3 cosh l2 x þ C5
: ð28bÞ
tm dx ¼ F, tm 9x ¼ 8 l ¼ 0 ð26a2cÞ sm ¼ D1 cosh l3 x cos l4 x þD4 sinh l3 x sin l4 x
l
8   Furthermore, the peel stress distribution across the adhesive
>  
>
> ða1 D3 a2 DÞtm  þða7 D2 a8 Þsm  and adherends’ thicknesses can be obtained by Eqs. (7) and (6);
>
> x ¼ 8l x ¼ 8l
>
> ( substituting Eq. (28a) into the overall equilibrium Eq. (1) and
>
> ½F þ 6M1 ð1=t1 þ t3 =t12 Þ=ðE1 t1 Þ
>
> , integrating fi, Qi and Mi (i¼1, 2, 3) with respect to x, then all the
>
> ¼
>
> ½F þ 6M2 ð1=t2 þ t3 =t22 Þ=ðE2 t2 Þ expressions for these forces are identified by imposing the rigorous
>
>  
>
>   boundary stress conditions (2) at the ends of the overlap, namely
>
> ðða7 D3 a8 DÞtm  þ ða4 D2 a5 Þsm 
>
> 8
< ( x ¼ 8l x ¼ 8l
>
> f ¼ D1 9x ¼ l D1 , f2 ¼ Ff1 ,
12M 
=ðE 1 t 3
Þ ð26d2iÞ > 1
>
>
>¼ 1 1
, >
< Q1 ¼ D2 9 D2 tm t3 =2, Q2 ¼ D2 D2 9x ¼ l tm t3 =2,
>
> 12M2 =ðE2 t23 Þ x¼l
>
> ð29Þ
>
>   > M ¼ f ðt þ t 3 Þ=2D3 þ D2 9x ¼ l ðxlÞ þ D3 9x ¼ l ,
>
> >
> 1 1 1
> ða7 D4 a8 D2 Þtm 
> þ ða4 D3 a5 DÞsm 
 >
>
>
> x ¼ 8l x ¼ 8l
: M2 ¼ D3 f2 ðt2 þ t3 Þ=2D2 9 ðx þlÞD3 9x ¼ l ,
>
> ( x ¼ l
>
>
>
> 12Q1 =ðE1 t13 Þ
>
> ¼ : where the terms Di (i¼1 3) are presented in Appendix B.
>
: 12Q2 =ðE2 t23 Þ

Hence the nine equations in Eq. (26) will determine exclu- 3. Verification and discussion
sively the nine unknown coefficients for tm and sm . Using Eq. (7),
then the peel stress su3y and sl3y at the interfaces between the The geometry and material parameters for the unbalanced joint
adhesive and adherends, can be obtained. Substituting the explicit in Fig. 1 were taken as: l¼10 mm, L¼40 mm, t1 ¼t2 ¼2 mm, t3 ¼
expressions of t3, su3y and sl3y into the overall equilibrium Eq. (1) 0.2 mm, E3 ¼0.7 GPa, m3 ¼ 0.4, E1 ¼70 GPa, E2 ¼140 GPa, m1 ¼ m2 ¼
and integrating fi, Qi and Mi (i¼ 1, 2) with respect to x, then all the 0.34 and F¼0.6 kN. 5466 four-noded 2D plane strain elements and
expressions for these forces are identified by imposing the 5717 nodes were examined to analyze the geometrically nonlinear
rigorous boundary stress conditions (2) at the ends of the overlap. elastic behavior of single-lap bonded joint under tension, using the FE
Thus, the distribution of stresses (six, siy, ti) (i¼1, 2, 3), at any point code of MSC.Patran/Nastran 2005. A zoom view of the FE models in
(x, yi) in both the adhesive and adherends is entirely determined by the overlap end region appears in Fig. 3. The loadings and boundary
applying these force expressions, Eqs. (25) and (7) to Eqs. (3), (5) and conditions are shown in Fig. 1. The adhesive layer was divided into
(6). Accordingly, the state of strain (eix, eiy, gi) (i¼1, 2, 3) in both the eight four-noded quadrilateral isoparametric elements through its
adhesive and adherends is also entirely determined by substituting thickness. In order to obtain the high peak stresses in the near vicinity
all the stresses (six, siy, ti) (i¼1, 2, 3) into Eqs. (8)–(10). of the overlap edges, a finer mesh with the smallest element of dim-
ensions 0.025 mm  0.03125 mm was used for these areas (Fig. 3). In
2.5. Simplified design formulae the geometrically nonlinear FE implementation, 10 equal load incre-
ments are employed and Newton’s method is selected as the non-
For convenience, the above solutions including Eq. (25) are called linear equation solver.
the full analytical solution for the shear and peel stresses of the The following analytical solution using present approach only
adhesive and adherends in the unbalanced single-lap joints. However, addressed the plane strain state for convenience. The normalized
it is not easy to use for the full solution due to two main factors: the adhesive average shear stress tm =t0 , and average peel stress sm =p
coefficient matrix inversion and the requirement of a4a8–a5a7 a0 in the adhesive middle plane predicted by the present theory, the
during its derivation in the unbalanced joints; when the unbalanced B–C solution and the FE models (baseline model) are plotted
joints become the balanced joints a4a8–a5a7 equals zero, which along the normalized overlap length in the subsequent figures, in
implies the full solution is not suitable for the specific case of the which t0 (F/2/l) and p (F/t2) are the applied average shear stress in
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 439

Fig. 3. Detailed 2D finite element mesh (baseline model).

the adhesive and applied average tensile stress in the adherend,


respectively.

3.1. Comparison with the B–C solution and FE model results

The shear stresses of the detailed FE models in the adhesive


mid plane, at the upper and lower adhesive interfaces are shown
in Fig. 4(a). It is indicated that the variation of adhesive shear
stresses through the thickness is relatively small except for the
region very close to the overlap edges. These numerical results
well correlate with the theoretical assumption of the constant
shear stress across the thickness of the adhesive, so this assump-
tion might be acceptable, which has been already verified theo-
retically in Ref. [5]. The adhesive peel stress distributions across Fig. 4. (a) shear stress distributions of the adhesive layer at the upper and lower
the bondline thickness are plotted in Fig. 4(b) from the present interfaces and middle plane in the detailed 2D FE model, and (b) peel stress
distributions across the adhesive layer in the present analytical and FE solutions.
analytical and FE solutions, and it is observed that their results are
comparable, as also validates the present analytical approach.
The comparison of the adhesive shear stress, and the adhesive
peel stresses at two interfaces and the mid plane, obtained from solution, thus the following analytical solution using present app-
the proposed full and simplified analytical solutions, B–C solution, roach only addressed the results obtained from the simplified solu-
whose adhesive stress expressions are given in Appendix C, and tions for convenience.
those of finite element analysis (FEA), were illustrated in
Fig. 5(a)–(c). It was indicated that the distribution of the adhesive 3.2. Theoretical prediction of the in-plane loads and 2D distribution
shear and peel stresses, particularly of these peak stresses, from of stress components
the new theory has a good correlation with the B–C solution and
the FE results. It can be observed that, after satisfying the zero A great deal of insight into joint behavior can be gained from a
shear stress condition at the ends of the adhesive, the adhesive study of these integrated stress quantities in the adherends. These
shear stresses from the full and simple analytical solutions and quantities include the tensile force (f1, f2), the transverse shearing
FEA reach their maximum values at a very short distance from the (Q1, Q2) and the bending moment (M1, M2) acting on the upper
free adhesive edges, and the peak stresses from the present and lower adherends. Fig. 6(a)–(c) illustrates the in-plane loads
method have better correlation with those of FEA than those of (tensile force, shearing force and bending moment) acting on the
the B–C solution, where the maximum shear stress was predicted adherends, which is predicted by the simplified analytical solu-
to occur at both ends of the overlap region. It shows that the explicit tions developed. These quantities are plotted along the normal-
closed-form analytical solutions established have the capability of ized overlap length direction, and the detailed analysis and
predicting the peel stress distribution through the adhesive thick- similar conclusions can be seen in Ref. [1].
ness, which might be absent in the B–C solution, and this theory In order to illustrate the present method is a (quasi) 2D model
provides a unified treatment for any possible adhesive flexibility. with the explicit close-formed analytical solutions considering all
Above all, it should be concluded that the adhesive stress distribu- the in-plane stress components for the upper and lower adher-
tions predicted by the simple design formulae with the explicit ends and adhesive (neglecting the longitudinal normal stress
constants in Eq. (28) well correlate with those of the full analytical component, however) and satisfying the zero shear stress condition
440 B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445

Fig. 5. Distributions of the adhesive stresses along the dimensionless overlap length for the unbalanced single-lap joint: (a) average shear stresses (tm), (b) average peel
stresses (sm) and (c) peel stresses at the upper and lower interfaces (su3y and sl3y ).

Fig. 6. In-plane load distributions (tensile force, shearing force and bending moment) acting on the upper and lower adherends predicted by the present theory: (a) f1 and
f2, (b) Q1 and Q2 and (c) M1 and M2.
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 441

Fig. 7. 2D stress distributions for the upper and lower adherends and adhesive layer predicted by the present theory: (a) s1x, (b) s2x, (c) s1y, (d) s2y, (e) t1, (f) t2 and (g) s3y.

at the adhesive free edges, Fig. 7(a)–(g) depicts the stress distribu-
tions of the upper and lower adherends and adhesive along the
normalized overlap length and their respective thickness directions,
respectively. An examination of these figures reveals the following
observations:

(1). The theoretical assumption of the linear distribution of


longitudinal normal stress for the upper and lower adherends
across their respective thickness directions in the present
mathematical formulation has been already verified by the
detailed FE model in Ref. [1], and their theoretical stress
distributions are illustrated in Fig. 7(a) and (b).
(2). The boundary stress conditions, including the stress-free
surface conditions in the theoretical formulation, have been
satisfied strictly, as shown in Figs. 7(a)–(g) and 5(a).
(3). The peel stresses for both the upper and lower adherends
reach their maximum values at the left end of the overlap, Fig. 8. Normal strain components (e1x and e1y) in the middle plane of the upper
and both of the two positions are located between their mid adherend.
planes and the interfaces between their respective adherends
and adhesive layer, as shown in Fig. 7(c) and (d). It is noted
that the maximum value of the peel stress in the upper
adherend is a tensile stress, while that of the lower adherend edge of the overlap at the interfaces between their respective
turns into the compressive. adherends and adhesive layer, as shown in Fig. 7(e)–(f). The
(4). The shear stresses of the upper and lower adherends attain distribution of the constant shear stress across the thickness
their maximum values in the region very close to the left of the adhesive layer is shown in Fig. 5(a).
442 B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445

Fig. 9. E2/E1 vs. the adhesive stresses: (a) average shear stresses (tm), (b) average peel stresses (sm), (c) peel stresses at the upper interface (su3y ) and (d) peel stresses at the
lower interface (sl3y ).

(5). It can be seen from Fig. 7(g) that the adhesive peel stress s3y edges. Note that the regions very close to the joint overlap ends
shows a significant variation with the dimensionless coordi- where the strain variation is severe are not included in all the
nate r3, especially near the free edges of the joint. However, curves for the purpose of clear plotting, due to a significant
the adhesive shear stress as seen in Fig. 5(a), unlike those of difference between the present theory and numerical values.
the upper and lower adherends(see Fig. 7(e) and (f)), in
general, does not yield relatively significant changes with
3.4. Parametric study for the stiffness unbalanced joints
the normalized adhesive thickness. Actually, the variation of
the adhesive stresses through the thickness is very small in
3.4.1. Effect of the Young’s modulus ratio E2/E1 between the lower
the center of the overlap region. This region is not a major
and upper adherends on the adhesive stress distributions
concern in engineering applications because it is far away
The effect of Young’s modulus ratio E2/E1 on the theoretical
from the stress concentration areas. The variations of these
distributions of the adhesive shear and the peel stresses was
stresses with the adhesive layer’s thickness also illustrate one
examined by changing the ratio E2/E1 from 1 to 4, while the value
of dangers in plotting the average stresses for predicting the
E1 was held constant at 70 GPa, and the ratio t1/t2 was chosen to be
joint failure.
equal to 1. It can be seen in Fig. 9(a)–(d) that the effect of the
(6). Finally, these stress components predicted by the present
adherends stiffness imbalance on the adhesive shear and peel stress
theory could provide a basis for subsequent joint failure
distributions increases as the ratio E2/E1 is far away 1 (the case of the
analysis including the main failure modes such as adherend
balanced joints), and it is noted that the adhesive peel stress
failure, cohesive failure and adhesive failure.
distributions at the edges of the interfaces increase sharply, as shown
in Fig. 9(c)–(d). This is obviously an unfavorable feature for the stress
concentrations in the adhesive joints because of the wide gap
3.3. Theoretical prediction and numerical validation of the normal
between the two extreme values of the adhesive shear or peel stress.
strains in the upper adherend
These distribution results obtained here coincide with the conclusions
from Sawa et al. [29].
One of the capabilities for the present theory is to provide the
2D strain prediction in both the upper and lower adherends.
Subsequently, taking the case of the upper adherend as an 3.4.2. Effect of the thickness ratio t1/t2 between the upper and lower
example, the two normal strain comparisons between theoretical adherends on the adhesive stress distributions
and numerical results in the mid plane are shown in Fig. 8. In Fig. 10(a)–(d) showed the effect of the thickness ratio t1/t2
general the correlation is good between the two except for the between the upper and lower adherends on the adhesive stress
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 443

Fig. 10. t1/t2 vs. the adhesive stresses: (a) average shear stresses (tm), (b) average peel stresses (sm), (c) peel stresses at the upper interface (su3y ) and (d) peel stresses at the
lower interface (sl3y ).

distributions. Similar to E2/E1, the effect of adherends stiffness Appendix A. Constants in the present full analytical solution
imbalance on the adhesive stress distributions increases as the
ratio t1/t2 increases. Figs. 9 and 10 indicated that the ratios of E2/
E1 and t1/t2, and different combinations of these two ratios, have a 4t13 þ22t12 t3 þ 39t1 t32 4t23 þ 22t22 t3 þ39t2 t32
a1 ¼ þ ,
significant influence on both maximum shear and maximum peel 420E1 420E2
stresses in the adhesive. It implies that there should be an optimal t3 4t1 þ 3t3 þ 9t32 =t1 15m1 t3 4t2 þ 3t3 þ 9t32 =t2 15m2 t3
a2 ¼ þ þ ,
parameter ratio E2t1/E1t2 suggested by Cheng et al. [26], which is G3 15E1 15E2
favorable to the stress intensities in the joint.
1 þ 3ð1 þ t3 =t1 Þ2 1 þ 3ð1þ t3 =t2 Þ2
a3 ¼ þ ,
E1 t1 E2 t2
 
t3 13 t1 t2 12 12
a4 ¼ þ þ , a5 ¼ þ ,
E3 35 E1 E2 5E1 t1 5E2 t2
4. Conclusions 2 2
12 12 11t1 þ39t1 t3 11t2 þ 39t2 t3
a6 ¼ þ , a7 ¼  ,
A general approach is proposed for the elastic stress analysis in E1 t13 E2 t23 210E1 210E2
the unbalanced adhesive joints, considering the complete strain– 1:2t3 =t1 m1 þ 0:2 1:2t3 =t2 m2 þ 0:2
a8 ¼  ,
displacement and the complete stress–strain equations for the E1 E2
adhesive and adherends. The theory can fulfill the zero shear 6ð1 þt3 =t1 Þ 6ð1 þt3 =t2 Þ
a9 ¼  :
stress condition at the adhesive free edges, and has the capability E1 t12 E2 t22
of predicting the peel stress distribution through the adhesive
thickness, and provides a unified treatment for any possible Appendix B. Constants and terms in the present simplified
adhesive flexibility. An explicit closed-form analytical solution is analytical solution
formulated for upper and lower adherends/adhesive stresses
(strains) and the in-plane loads acting on the adherends and then qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
simplified for practical applications, and simple design formulae l21,2 ¼ ða2 7 a22 4a1 a3 Þ=ð2a1 Þ,
for adhesive stresses are developed. The results predicted by the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
present full and simplified solutions for unbalanced bonded joints ð 7 l3 7il4 Þ2 ¼ ða5 7i 4a4 a6 a25 Þ=ð2a4 Þ,
were verified by comparing with the B–C solution and the
detailed 2D geometrically nonlinear FE model. The effects of the 3
C1 ¼ fa2 ½sinh l2 l=ðl2 lÞcosh l2 lFða1 l2 a2 l2 Þsinh l2 l=ð2lÞg=D,
stiffness unbalanced parameters, E2/E1 and t1/t2, on the adhesive
3 3
stress distributions are also discussed. C2 ¼ a1 =½ða a a a
1 l1  2 l1 Þcosh l1 lð 1 l2  2 l2 Þsinh l1 l coth l2 l,
444 B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445

3 stress s3y can be written as


C3 ¼ fFða1 l1 a2 l1 Þsinh l1 l=ð2lÞa2 ½sinh l1 l=ðl1 lÞcosh l1 lg=D,
3 3
C4 ¼ a1 =½ða1 l1 a2 l1 Þcoth l1 l sinh l2 lða1 l2 a2 l2 Þcosh l2 l, a1 cosh l1 x a2 sinh l1 x F a1
C5 ¼ ðC1 cosh l1 l þ C3 cosh l2 lÞ,
t3 ¼ þ þ  ,
sinh l1 l cosh l1 l 2l l1 l

D1 ¼ ðb1 R4 b4 R2 Þ=ðR1 R4 R2 R3 Þ, s3y ¼ cosh l2 xðB1 cos l2 x þB2 sin l2 xÞ
D2 ¼ ðb2 R8 b3 R6 Þ=ðR5 R8 R6 R7 Þ, þsinh l2 xðB3 cos l2 x þ B4 sin l2 xÞ,
D3 ¼ ðb3 R5 b2 R7 Þ=ðR5 R8 R6 R7 Þ,
where
D4 ¼ ðb4 R1 b1 R3 Þ=ðR1 R4 R2 R3 Þ, !
 
4G3 1 1 3E3 1 1
l21 ¼ þ
4
, l2 ¼ þ ,
D ¼ ða1 l31 a2 l1 Þsinh l1 l½sinh l2 l=ðl2 lÞcosh l2 l t3 E2 t2 E1 t1 t3 E1 t13 E2 t23
3
"  #
ða1 l2 a2 l2 Þsinh l2 l½sinh l1 l=ðl1 lÞcosh l1 l, G3 F F 6M2 6M1
a1 ¼ þ þ þ ,
2t3 l1 E2 t2 E1 t1 E2 t22 E1 t12
F=2 þ 3M1 ð1=t1 þ t3 =t12 Þ F=2þ 3M2 ð1=t2 þt3 =t22 Þ "  #
a1 ¼  , G3 F F 6M2 6M1
E1 t1 E2 t2 a2 ¼  þ  ,
2t3 l1 E2 t2 E1 t1 E2 t22 E1 t12
F=2 þ 3M1 ð1=t1 þ t3 =t12 Þ F=2 þ3M2 ð1=t2 þt3 =t22 Þ
a2 ¼   ,
E1 t1 E2 t2 B1 ¼ ðb1 R6 þb4 cosh l2 l cos l2 lÞ=R4 ,
B2 ¼ ðb2 R2 þb3 cosh l2 l sin l2 lÞ=R1 ,
6M2 6M1 6M2 6M1 6Q1 6Q2 B3 ¼ ðb2 R3 þb3 sinh l2 l cos l2 lÞ=R1 ,
b1 ¼ þ , b2 ¼  , b3 ¼  ,
E2 t23 E1 t13 E2 t23 E1 t13 E1 t13 E2 t23 B4 ¼ ðb1 R5 þb4 sinh l2 l sin l2 lÞ=R4 ,
6Q1 6Q2 2 2 ! !
b4 ¼  , c1 ¼ a4 ðl3 l4 Þa5 , c2 ¼ 2a4 l3 l4 ,
E1 t13 E2 t23 3E3 M2 M1 3E3 M2 M1
b1 ¼ 2
þ , b2 ¼  ,
2 2
c3 ¼ l3 ½a4 ðl3 3l4 Þa5 , c4 ¼ l4 ½a4 ðl4 3l3 Þ þ a5 ,
2 2 t3 l2 E2 t23 E1 t13 2 3
t3 l2 E2 t2 E1 t1
3
! !
R1 ¼ c1 cosh l3 l cos l4 lc2 sinh l3 l sin l4 l, 3E3 Q2 Q 3E3 Q2 Q
b3 ¼  1 , b4 ¼ þ 13 ,
R2 ¼ c1 sinhl3 lsinl4 l þc2 coshl3 l cos l4 l, 3
t3 l2 E2 t23 E1 t13 3 3
t3 l2 E2 t2 E1 t1
R3 ¼ c3 sinh l3 l cos l4 l þc4 cosh l3 l sin l4 l,
R4 ¼ c3 cosh l3 l sin l4 lc4 sinh l3 l cos l4 l, R1 ¼ 0:5ðcosh 2l2 lsin 2l2 lÞ,
R5 ¼ c1 cosh l3 l sin l4 l þc2 sinh l3 l cos l4 l, R2 ¼ cosh l2 l cos l2 l þ sinh l2 l sin l2 l,
R6 ¼ c1 sinh l3 l cos l4 lc2 cosh l3 l sin l4 l, R3 ¼ cosh l2 l cos l2 lsinh l2 lsin l2 l,
R7 ¼ c3 sinh l3 l sin l4 lc4 cosh l3 l cos l4 l, R4 ¼ 0:5ðcosh 2l2 l þ sin 2l2 lÞ,
R8 ¼ c3 cosh l3 l cos l4 l þc4 sinh l3 l sin l4 l, R5 ¼ sinh l2 lcosl2 l þ cosh l2 l sin l2 l,
R6 ¼ sinh l2 lcos l2 lcosh l2 l sin l2 l:
C1 C2
D1 ¼ sinh l1 x þ cosh l1 x
l1 l1
C3 C4 References
þ sinh l2 x þ cosh l2 x þ C5 x,
l2 l2
[1] Zhao B, Lu Z-H. A two-dimensional approach of single-lap adhesive bonded
D1 joints. Mech Adv Mater Struct 2009;16(2):130–59.
D2 ¼ ðl3 sinh l3 x cos l4 x þ l4 cosh l3 x sin l4 xÞ [2] Volkersen O. Die Nietkraftverteilung in zugbeanspruchten nietverbindungen
l23 þ l24 mit konstanten laschenquerschnitten. Luftfahrtforschung 1938;15(1/2):41–7.
D2 [3] Goland M, Reissner E. The stresses in cemented joints. J Appl Mech—T ASME
þ ðl3 sinh l3 x sin l4 xl4 cosh l3 x cos l4 xÞ 1944;11(1):A17–27.
l3 þ l24
2
[4] Frostig Y, Thomsen OT, Mortensen F. Analysis of adhesive-bonded joints,
D3 square-end, and spew-fillet-high—order theory approach. J Eng Mech—ASCE
þ ðl3 cosh l3 x cos l4 x þ l4 sinh l3 x sin l4 xÞ 1999;125(11):1298–307.
l23 þ l24 [5] Radice JJ, Vinson JR. On the analysis of adhesively bonded structures: a high
D4 order semi-elastic adhesive layer model. Compos Sci Technol
þ ðl3 cosh l3 x sin l4 xl4 sinh l3 x cos l4 xÞ, 2008;68(2):376–86.
l23 þ l24 [6] Chen D, Cheng S. An analysis of adhesive-bonded single-lap joints. J Appl
Mech—T ASME 1983;50(1):109–15.
D1 h i [7] Adams RD, Mallick V. A method for the stress analysis of lap joints. J Adhes
2 2
D3 ¼ 2 2
ðl3 l4 Þcosh l3 x cos l4 x þ 2l3 l4 sinh l3 x sin l4 x 1992;38(3/4):199–217.
ðl3 þ l4 Þ2 [8] Sawa T, Nakano K, Toratani H. A two-dimensional stress analysis of single-lap
D2 h i adhesive joints subjected to tensile loads. J Adhes Sci Technol
2 2
þ 2 2
ðl3 l4 Þcosh l3 x sin l4 x2l3 l4 sinh l3 x cos l4 x 1997;11(8):1039–62.
ðl3 þ l4 Þ2 [9] Hart-Smith LJ. Adhesive-bonded single-lap joints. NASA Langley Research
D3 h i Center, Report NASA CR-112236, January 1973.
2 2
þ 2 2
ðl3 l4 Þsinh l3 x cos l4 x þ 2l3 l4 cosh l3 x sin l4 x [10] Oplinger DW. Effects of adherend deflections in single lap joints. Int J Solids
ðl3 þ l4 Þ2 Struct 1994;31(18):2565–87.
D4 h i [11] Adams RD, Peppiatt NA. Effect of Poisson’s ratio strains in adherends on
2 2
þ 2 2
ðl3 l4 Þsinh l3 x sin l4 x2l3 l4 cosh l3 x cos l4 x : stresses of an idealized lap joint. J Strain Anal 1973;8(2):134–9.
ðl3 þ l4 Þ2 [12] Ojalvo IU, Eidinoff HL. Bond thickness effects upon stresses in single-lap
adhesive joints. AIAA J 1978;16(3):204–11.
[13] Tsai MY, Oplinger DW, Morton J. Improved theoretical solutions for adhesive
lap joints. Int J Solids Struct 1998;35(12):1163–85.
[14] Luo Q, Tong L. Linear and higher order displacement theories for adhesively
Appendix C. Adhesive stress expressions of the B–C solution
bonded lap joints. Int J Solids Struct 2004;41(22/23):6351–81.
[15] Zhao B, Lu Z-H, Lu Y-N. Simplified finite element formulation of unsymme-
In the model developed by Bigwood and Crocombe (called the trical adhesive joints based on analytical solutions. Mech Adv Mater Struct
B–C solution) [35], the shear and peel stresses are constant across 2009;16(8):597–615.
[16] Luo Q, Tong L. Analytical solutions for adhesive composite joints considering
the adhesive thickness, and the longitudinal deformations of the large deflection and transverse shear deformation in adherends. Int J Solids
adhesive are neglected. The adhesive shear stress t3 and peel Struct 2008;45(22/23):5914–35.
B. Zhao et al. / International Journal of Adhesion & Adhesives 31 (2011) 434–445 445

[17] Luo Q, Tong L. Analytical solutions for nonlinear analysis of composite single- [28] Allman DJ. A theory for elastic stresses in adhesive bonded lap joints. Q J
lap adhesive joints. Int J Adhes Adhes 2009;29(2):144–52. Mech Appl Math 1977;30(4):415–36.
[18] Zhao B. 2D linear and 1D nonlinear analytical stress solutions of adhesive- [29] Sawa T, Liu J, Nakano K, Tanaka J. A two-dimensional stress analysis of single-
bonded single-lap joints. PhD Thesis, Tsinghua University, Beijing, PR China, lap adhesive joints of dissimilar adherends subjected to tensile loads. J Adhes
2009. (in Chinese). Sci Technol 2000;14(1):43–66.
[19] Li G. Deformation of balanced and unbalanced adhesively bonded single-lap [30] Wang J, Zhang C. Three-parameter elastic foundation model for analysis of
joints. PhD Thesis, The University of New Brunswick, Fredericton, NB, Canada, adhesively bonded joints. Int J Adhes Adhes 2009;29:495–502.
2000. [31] Delale F, Erdogan F, Aydinoglu MN. Stresses in adhesively bonded joints: a
[20] Tsai MY, Morton J. An evaluation of analytical and numerical solutions to the closed-form solution. J Compos Mater 1981;15(2):249–71.
single-lap joint. Int J Solids Struct 1994;31(18):2537–63. [32] Shahin K, Kember G, Taheri F. An asymptotic solution for evaluation of
[21] Carpenter WC. A comparison of numerous lap joint theories for adhesively stresses in balanced and unbalanced adhesively bonded joints. Mech Adv
bonded joints. J Adhes 1991;35(1):55–73. Mater Struct 2008;15(2):88–103.
[22] Adams RD, Peppiatt NA. Stress analysis of adhesive-bonded lap joints. J Strain [33] Yang C, Pang SS. Stress–strain analysis of single-lap composite joints under
Anal 1974;9(3):185–96. tension. J Eng Mater—T ASME 1996;118(2):247–55.
[23] Crocombe AD, Adams RD. Influence of the spew fillet and other parameters [34] Taheri F, Zou GP. Treatment of unsymmetric adhesively bonded compo-
on the stress distribution in the single lap joint. J Adhes 1981;13(2):141–55. site sandwich panels-to-flange joints. Mech Adv Mater Struct 2004;11:
[24] Tsai MY, Morton J. The effect of a spew fillet on adhesive stress distribution in 175–96.
laminated composite single-lap joints. Compos Struct 1995;32(1/4):123–31. [35] Bigwood DA, Crocombe AD. Elastic analysis and engineering design formulae
[25] Zhao B, Yue P, Lu Y-N, et al. Stress and stiffness analysis of a simplified for bonded joints. Int J Adhes Adhes 1989;9(4):229–42.
adhesive bonded finite element model considering spew-fillets. J Mech [36] Lee J, Kim H. Stress analysis of generally asymmetric single lap adhesively
Strength 2009;31(6):939–46 (in Chinese). bonded joints. J Adhes 2005;81(5):443–72.
[26] Cheng S, Chen D, Shi Y. Analysis of adhesive-bonded joints with nonidentical [37] Wu ZJ, Romeijn A, Wardenier J. Stress expressions of single-lap adhesive
adherends. J Eng Mech—ASCE 1991;117(3):605–23. joints of dissimilar adherends. Compos Struct 1997;38(1/4):273–80.
[27] Renton WJ, Vinson JR. Analysis of adhesively bonded joints between panels of [38] Williams J. Stresses in adhesive between dissimilar adherends. J Adhes
composite materials. J Appl Mech—T ASME 1977;44(1):101–6. 1975;7(2):97–107.

You might also like