ARTICLE IN PRESS

International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives

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

A new method for the determination of bending moments in single lap joints
X. Zhao a, R.D. Adams a,b, Lucas F.M. da Silva c,
a
Department of Mechanical Engineering, University of Bristol, Queen’s Building, University Walk, Bristol BS8 1TR, UK
b
Department of Engineering Science, University of Oxford, Parks Road, Oxford, UK
c
Departamento de Engenharia Mecânica, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

a r t i c l e in fo abstract

Article history: This paper presents a new method of determining bending moments at the overlap ends of single lap
Accepted 9 September 2009 joints. It is based on the assumption that the overlap region does not deform under load. This
assumption is supported by the observation that for an overlap of up to 50 mm, the large deformations
Keywords: occur only at the ends of the overlap. Physical insight into the mechanics of the overlap region shows
C. Lap-shear that the rotation is merely a geometric effect, which helps alignment of the load path during the loading
C. Finite element stress analysis process.
Analytical stress analysis Examples show that the method proposed here is better than those of the Goland and Reissner and
Bending moment factor of the Hart-Smith for overlaps up to 25 mm long. It is more useful for unbalanced joints, where Goland
and Reissner’s method fails to work and Hart-Smith’s method involves difficult mathematics. This
method may also be easily extended to deal with non-linearities in the adherend.
& 2009 Elsevier Ltd. All rights reserved.

1. Introduction overlap. The bending moments are then used as boundary

Knowledge of the state of stresses inside the adhesive layer of
an adhesively bonded joint is essential for joint strength
prediction and joint design. There are two methods for the stress
analysis of lap joints, namely analytical and numerical methods.
Analytical methods using closed-form algebra employ classical
linear theories in which some simplifications are used. The finite
element (FE) method, on the other hand, is a well established
numerical technique which can handle complex structures and
non-linear material properties where classical methods generally
fail to work. Although the closed-form solutions have their
limitations, they are easy to use, especially for parametric studies.
The FE method needs large computer power and experienced
personnel. Consequently, the former is widely used for joint
design and the latter for research. The aim of this paper is to
improve the analysis of the closed-form solutions so the classical
method can be more widely used. The FE method is only used to
calibrate the closed-form solution.
Analytical solutions for single lap joints have been developed
since 1938 when Volkersen analysed the shear-lag problem [1].
However, he ignored the bending moments applied to the joint
because of the eccentricity of the load. It was Goland and Reissner
who were among the first to produce solutions for peel stresses as
well as for shear stresses [2]. There are two steps in their method.
The first is to determine bending moments at the ends of the
 Corresponding author. Tel.: + 351 225081706; fax: + 351 225081445.
E-mail address: lucas@fe.up.pt (L.F.M. da Silva).
conditions to analyse the overlap region. They considered two
cases, in which the adhesive layer is either stiff or very flexible. To
overcome the limitations involved in the work of Volkersen,
Goland and Reissner and other earlier workers, extensive devel-
opment has been made since then, notably by Renton and Vinson
[3] and Allman [4]. All of the methods attempt to improve the
analyses of the stresses inside the overlap, but they follow the first
step in Goland and Reissner’s method to determine the bending
moments. It was Hart-Smith who first recognized the limitation in
Goland and Reissner’s determination of the bending moments [5],
claiming that Goland and Reissner overestimated the bending
moments. To improve the determination of bending moments, he
developed a new bending moment factor, which involved
complicated mathematics. It should be noted that both Goland
and Reissner’s and Hart-Smith’s bending moment factors are only
applicable to identical adherends. Joints with different adherends
(e.g. composite and aluminium) or with different thickness are
widely used in industry and this requires the determination of
bending moments at the ends of the overlap if analytical methods
are to be used. Some authors tried to obtain more general closed-
form solutions by including non-identical adherends (thickness
and material properties) or composite adherends [3–14]. How-
ever, as the model gets more general, the governing equations
become increasingly complicated and require the use of a
computer for solution [15,16]. For example, the solution given
by Hart-Smith can only be obtained by numerical iteration [5]. It
can be concluded from the above that a more accurate and
convenient determination of bending moments is required for
both identical and, in particular, non-identical adherends. In this

0143-7496/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijadhadh.2009.09.001

Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
 ARTICLE IN PRESS
2 X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]
paper, a new method is proposed for determining the bending
moments for single lap joints with either identical or non-
identical adherends, and it will be compared with the Goland and
Reissner and Hart-Smith methods as well as the results from a FE
analysis.
2. Theory
The proposed method is based on the assumption that the
overlap region does not deform under load. This assumption is
based on the realisation that the bending stiffness of the overlap is
much higher than that of either adherend (since bending stiffness
is proportional to the cube of the thickness). Experiments also
show that the overlap region does not deform much during the
loading process. More evidence is given in Fig. 1, where the results
are calculated using the FE method for joints with a 12.7 mm
Fig. 1. Bending moments at the ends of the overlap for different properties of the
overlap determined by using the finite element method.
Fig. 2. Geometry and material properties of joints analysed (not to scale).
Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
overlap and 1.6 mm thick adherends. Other parameters are listed
in Fig. 1. The load per unit width is varied up to 1000 kN/m which
is a typical value encountered in practice for structural adhesives
(see Section 3). The bending moments were calculated by
numerical integration with stresses available at the Gauss points
from the FE results and then the bending moments were
extrapolated to the edges of the overlap. It can be seen that the
stiffness of both the adherends and the adhesive and the thickness
of the adhesive (in the range of 0.125–0.3 mm) in the overlap
region (the stiffness of adherends outside the overlap is kept
constant) have little effect on the bending moments. With this
assumption, the following analysis is able to utilise basic
cylindrically bent plate theory.
Fig. 2 shows the geometry of a single lap joint and its material
properties. The overlap length of the joint is l, the lengths of the
two sheets outside the joint are l1 and l2, and their thicknesses are
t1 and t2. The joint width is assumed to be large enough compared
with the sheet thickness so that only the cross section need
be considered. The joint is loaded with tensile forces T per unit of
joint width. The overlap is assumed to be rigid so that only the
two sheets need to be analysed using the coordinate system
shown in Fig. 2. The w coordinate represents transverse deflection
of the two sheets from their undeformed positions. The effects of
the sheet deflections are included in the determination of the
bending moments.
The bending moments per unit width in the sheets may be
expressed as follows:
M1 ¼ Tðan x1  w1 Þ; 0 r x1 r l1 ð1Þ
M2 ¼  T½an ðl2  x2 Þ þ w2 ; 0 r x2 r l2 ð2Þ
where an is the angle between the x1 (or x2) axis and the line of
the load path as shown in Fig. 2 and is given by
an ¼ ðt1 þ t2 Þ=2ðl1 þ l2 þ lÞ ¼ ðt1 þ t2 Þ=2L ð3Þ
where L=l1 +l2 + l, w1 and w2 are the deflections of the two sheets
from their undeformed shape and the thickness of the adhesive is
ignored.
 ARTICLE IN PRESS
X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]] 3

According to the classical theory of the small bending of thin, The moment at the left end of the overlap become
cylindrically bent plates, the differential equations for the !
d2 w1
transverse deflections of the two sheets can be written as ML ¼ M1jx1 ¼ l1 ¼  D1
dx21
jx1 ¼ l1
d2 w1 M1 Tðan x1  w1 Þ 0:5Tðt1 þ t2 Þz2 cosh z2 l2 sinh z1 l1
¼  ¼  ð4Þ ¼
dx21 D1 D1 z2 sinh z1 l1 cosh z2 l2 þ z1 z2 l cosh z1 l1 cosh z2 l2 þ z1 cosh z1 l1 sinh z2 l2
ð13Þ
d2 w2 M2 T½an ðl2  x2 Þ þ w2 
¼  ¼ ð5Þ
dx22 D2 D2 And the moment at the right end of the overlap is
!
where D1 and D2 are the flexural rigidities of the two sheets, d2 w2
MR ¼ M2jx2 ¼ 0 ¼  D2
respectively, i.e. D1 = E1t31/12(1  n12), D2 =E2t32/12(1 n22) and n is the dx22
jx2 ¼ 0
Poisson’s ratio. 0:5Tðt1 þ t2 Þz1 cosh z1 l1 sinh z2 l2
¼
Writing z21 = T/D1 and z22 = T/D2, then z2 sinh z1 l1 cosh z2 l2 þ z1 z2 l cosh z1 l1 cosh z2 l2 þ z1 cosh z1 l1 sinh z2 l2
ð14Þ
d2 w1 2 2
 z1 w1 þ z1 an x1 ¼ 0 ð6Þ
dx21
But if the values of z1l1 and z2l2 are sufficiently large, we can
say that
d2 w2 2 2
 z2 w2  z2 an ðl2  x2 Þ ¼ 0 ð7Þ sinh z1 l1  cosh z1 l1 ð15aÞ
dx22
and
These are the standard, second order, linear differential
equations and have solutions of the form sinh z2 l2  cosh z2 l2 ð15bÞ

w1 ¼ A1 cosh z1 x1 þB1 sinh z1 x1 þ an x1 ; 0 rx1 r l1 ð8Þ Then we have

an Lz2 sinh z1 x1
w1 ¼ þ an x1 ;
w2 ¼ A2 cosh z2 x2 þB2 sinh z2 x2  an ðl2  x2 Þ; 0 r x2 r l2 ð9Þ z2 sinh z1 l1 þ z1 z2 l cosh z1 l1 þ z1 cosh z2 l2
0 r x1 r l1 ð16Þ
The four constants can be determined from the following four
boundary conditions, as shown in Fig. 3
and
w1 ¼ 0 at x1 ¼ 0 ð10aÞ
an Lz1 cosh z2 x2 an Lz1 sinh z2 x2
w2 ¼   an ðl2  x2 Þ; 0 r x2 r l2
w2 ¼ 0 at x2 ¼ l2 ð10bÞ z2 þ z1 z2 l þ z1 z2 þ z1 z2 l þ z1
ð17Þ
dw1 dw2 The moment at the left end of the overlap becomes
¼ at x1 ¼ l1 ; x2 ¼ 0 ð10cÞ
dx1 dx2
0:5Tðt1 þ t2 Þz2
ML ¼ ð18Þ
z2 þ z1 z2 l þ z1
dw1
w1  w2 ¼  l at x1 ¼ l1 ; x2 ¼ 0 ð10dÞ Similarly, at the right end
dx1
0:5Tðt1 þ t2 Þz1
where Eqs. (10c) and (10d) are from the assumption that the MR ¼ ð19Þ
z2 þ z1 z2 l þ z1
overlap is rigid. By applying the boundary conditions, the
deflections of the two plates can be written as For identical adherends, z1 = z2, then we have

an Lz2 cosh z2 l2 sinh z1 x1 0:5Tt 0:5Tt

w1 ¼ ML ¼ MR ¼ ¼ ¼ 0:5Ttk ð20Þ
z2 sinh z1 l1 cosh z2 l2 þ z1 z2 l cosh z1 l1 cosh z2 l2 þ z1 cosh z1 l1 sinh z2 l2 1 þ0:5lz 1þ cz

þ an x1 ; 0 r x1 r l1 ð11Þ where z = z1 = z2, c is half the overlap length, i.e. 2c= l and k is
defined as the bending moment factor, which is
1
an Lz1 sinh z2 l2 cosh z1 l1 cosh z2 x2 k¼ ð21Þ
w2 ¼ 1þ cz
z2 sinh z1 l1 cosh z2 l2 þ z1 z2 l cosh z1 l1 cosh z2 l2 þ z1 cosh z1 l1 sinh z2 l2
an Lz1 cosh z2 l2 cosh z1 l1 sinh z2 x2 Hart-Smith’s result can be written as

z2 sinh z1 l1 cos hz2 l2 þ z1 z2 l cosh z1 l1 cosh z2 l2 þ z1 cosh z1 l1 sinh z2 l2 1
an ðl2  x2 Þ; 0 r x2 r l2 ð12Þ M ¼ 0:5Tðt þ ta Þ 2
ð22Þ
1 þ zc þðz c2 =6Þ
where ta is the thickness of the adhesive. Therefore, his bending
moment factor is
1
kHS ¼ 2
ð23Þ
1 þ zc þ ðz c2 =6Þ
Comparison of Eq. (21) with Eq. (23) shows that the difference
between the present result and Hart-Smith’s is the term z2c2/6.
Hart-Smith’s form includes the thickness of the adhesive, which is
ignored in the present method.
In order to compare the results of the present method with
Goland and Reissner’s, their formulae were used in the following
forms
Fig. 3. Sketch of boundary conditions for the analysis of the single lap joint. M ¼ 0:5Ttk ð24aÞ

Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
 ARTICLE IN PRESS
4 X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]

and
1
kGR ¼ pffiffiffi pffiffiffi ð24bÞ
1 þ2 2 tanhðzc=2 2Þ

3. Results

The results of the new formula for the bending moment are
shown in the form of graphs of bending moment vs. load per unit
width. The load per unit width varies from 0 to 700 kN/m. A
typical failure load for an epoxy varies with the type of epoxy, the
type of adherend and geometrical parameters. However, it is
reasonable to say that most practical cases with a width of 25 mm
give a failure load under 15 kN [16], i.e. 600 kN/m. Since the load T
in the theory above is per unit width, the bending moments given
below are all for a joint of unit width.

Fig. 5. Bending moments calculated by using a FE analysis with different lengths

3.1. Identical adherends of adherend (t1 = t2 =1.6 mm, E1 = E2 = 70 GPa).

3.1.1. Different lengths of overlap

To verify and demonstrate the use of the new formula
presented above, numerical examples are presented in this
section. First, single lap joints with identical adherends are used
and calculations are made with different overlap lengths and
loads. To compare with Goland and Reissner’s and Hart-Smith’s
methods, corresponding results from their methods are also
shown in the figures. Comparison was also made with results from
the FE method. In the FE analysis, an adhesive with a Young’s
modulus of 2.8 GPa was used, which represents the real conditions
in a typical lap joint using an epoxy adhesive. A 2D plane strain FE
program was used throughout this study. Eight-noded isopara-
metric elements were used and a large displacement linear elastic
model was included. A typical mesh is shown in Fig. 4. The
boundary conditions used for the FE analysis are shown in Fig. 2.
These are typical boundary conditions used in testing. While
the effects of boundary conditions on the bending moments at the
ends of the overlap are significant with small displacement FE
analysis, the effects are very small with large displacement FE Fig. 6. Bending moment factors and bending moments with 12.7 mm overlap
analysis. Guo et al. showed that the boundary conditions (with (t1 = t2 = 1.6 mm, E1 = E2 =70 GPa).
and without spacers at the grips) are not relevant when the free
length of the adherends is relatively long (l1/c410) [17]. For single
available at the Gauss points from the FE results and then the
lap joints, large displacement analysis is essential to model
bending moments were extrapolated to the edges of the overlap.
the eccentricity of the load. For different lengths of overlap,
The results of the bending moment factors and bending
the adherend length outside the overlap was kept constant at
moments for joints with varying lengths of overlap from 12.7 to
87.3 mm. For different lengths of overlap, it is obvious that the
100 mm are shown in Figs. 6–10, inclusively. For joints with a
ratio of the length of the adherend outside the overlap to that of
12.7 mm overlap, all the three methods yield almost identical
the overlap (l1/l) varies significantly. To analyse this effect,
results for the bending moments within the whole range of the
calculations were made with different lengths of the adherend
load, although there is slight discrepancy in the bending moment
outside the overlap for 12.7 and 100 mm overlaps. It can be seen in
factors as shown in Fig. 6. Fig. 7 shows the bending moment
Fig. 5 that the bending moments are insensitive to the adherend
factors and bending moments for joints with a 25 mm overlap.
lengths outside the overlap, although short adherends outside the
It can be seen that the present method has the best agreement
overlap have some small effect on the bending moments for short
with the FE method for all the methods for bending moment
overlaps. It should be noted that Fig. 5 was obtained for a high
factors and bending moments. Goland and Reissner’s method
load. The adherend length (= 87.3 mm) outside the overlap used in
overestimates both bending moment factors and bending
the present study (see Fig. 2) was long enough not to have any
moments and the larger the load, the larger is the overestimate.
effect on the calculations of the bending moments. The bending
Hart-Smith’s method, however, underestimates both bending
moments were calculated by numerical integration with stresses
moment factors and bending moments, giving large errors at
large load. It may be concluded that either the present method or
Goland and Reissner’s method should be used, because Hart-
Smith’s method does not err on the safe side. For long overlap
joints (50 mm), none of the three methods gives satisfactory
results, as is clearly shown in Fig. 8. Hart-Smith’s method yields
too large an error to be of any practical use. Goland and Reissner’s
Fig. 4. A typical mesh for the finite element analysis of the single lap joint. method and the present method produce the same order of error,

Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
 ARTICLE IN PRESS
X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]
Fig. 7. Bending moment factors and bending moments with 25 mm overlap
(t1 =t2 = 1.6 mm, E1 =E2 = 70 GPa).
Fig. 8. Bending moment factors and bending moments with 50 mm overlap
(t1 =t2 = 1.6 mm, E1 =E2 = 70 GPa).
Fig. 9. Bending moment factors with 100 mm overlap (t1 = t2 = 1.6 mm, E1 = E2 = 70
GPa).
with the former overestimating and the latter underestimating
the true bending moment factors and bending moments.
However, the large error appears only at high loads. For medium
loads, e.g. 10 kN for a joint with a width of 25 mm, the present
method introduces an error of about 10%, which can still be used
for engineering applications. For very long overlap joints
Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
5
Fig. 10. Bending moments with 100 mm overlap (t1 = t2 = 1.6 mm, E1 = E2 =70 GPa).
(100 mm), neither Hart-Smith’s method nor the present method
gives accurate results, although Hart-Smith’s method is worse
than the present method as is shown in Figs. 9 and 10. Goland and
Reissner’s method, however, is the most accurate, even though the
error is quite large for high loads. But it is on the safe side.
As can be seen from the above discussion, the present method
agrees best with the FE method for joints with reasonably long
overlaps (up to 50 mm). The present method is in all circum-
stances superior to Hart-Smith’s method, although the latter
involves much more complicated analysis. For very long overlap
joints, however, only Goland and Reissner’s method gives reason-
ably accurate results. Hart-Smith’s method assumes small
displacements in the overlap region and many assumptions are
made in the process of deriving the bending moments. The
present method ignores any deformation of the overlap region.
Therefore, Hart-Smith’s and the present methods will have large
errors for long overlaps. Large displacements are included in
Goland and Reissner’s method. Consequently, it is superior to
Hart-Smith’s and the present method for very long overlaps. It is
felt that the rigid overlap assumption is acceptable for joints up to
50 mm overlap. However, it should be noted that the employment
of closed-form solutions for very long overlap joints may not be
accurate anyway in that the solution procedure in the second step,
following the determination of bending moments, is based on
small displacement theories. This will therefore introduce errors
in the stress analysis. As a result, the large displacement FE
method is the only tool which should be used for very long
overlap joints.
It should be noted that the cases studied above are all for a
relatively thin bondline (0.125 mm) which is the most common in
practice. However, thick bondlines have a dramatic effect on the
bending moment. Grant et al. [18] have shown that in the case of a
joint with adherends of 1 mm thick and an overlap of 15 mm with
thin bondlines (0.1 mm), the effect of the adhesive thickness is
negligible on the bending moment, but, for thick bondlines
(3 mm), only the method of Hart-Smith, which takes into account
the adhesive thickness, gives satisfactory results, provided the
adherends do not deform plastically. In that case, only a FE
analysis gives realistic values.
3.1.2. Change in thickness of the adherends
All the joints analysed above had 1.6 mm adherends. In joint
testing, thicker adherends are sometimes used, such as in the so-
called thick adherend shear test [19,20] even though the geometry
is different. To assess the accuracy of the present method for
 ARTICLE IN PRESS
6 X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]
thicker adherends, analyses were made with a 12.7 and 50 mm
overlaps for 3.2 and 6.35 mm adherends. Fig. 11 shows the results
with 12.7 mm overlap for 3.2 and 6.35 mm adherends. It can be
seen that all the methods give accurate predictions of the bending
moments for the thick adherends. This is also true for the 50 mm
overlap as shown in Fig. 12. It can be seen that more accurate
results are given with the analytical methods for thick adherends
than for thin ones. The reason is that thick adherends are stiffer
and so bend less than thin ones. As a result, the rigid overlap
assumption is better for thick adherends than for thin ones.
3.1.3. Variation in the stiffness of the adherends
It has been shown that the analytical methods are more
accurate for thick adherends than for thin ones. It may be argued
that the method should therefore be more accurate for stiff
adherends than for flexible ones. This is true, as shown in Fig. 13,
in which the adherend properties are those of a typical steel with
a 1.6 mm thickness and a 50 mm overlap. It clearly shows that
the methods are much more accurate for stiff adherends than
for flexible ones as shown in Fig. 8. Consequently, it may be
concluded that the stiffer the adherends (either high Young’s
modulus or large thickness), the more accurate the analytical
methods.
Fig. 11. Bending moments for 12.7 mm overlap with 3.2 and 6.35 mm adherends
(E1 =E2 = 70 GPa).
Fig. 12. Bending moments for 50 mm overlap with 3.2 and 6.35 mm adherends
(E1 =E2 = 70 GPa).
Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
3.2. Non-identical adherends
3.2.1. Adherends having different material properties
Having discussed results for balanced single lap joints, the
unbalanced single lap joints are calculated below. First, the
geometry of the joints is kept the same as that of balanced joints.
But material properties are changed as shown in Fig. 2 case 2,
which is typical of a steel/aluminium joint. A joint with a 12.7 mm
overlap is used followed by a joint with a 25 mm overlap. Bending
moments are shown in Figs. 14 and 15. Results from the FE
method are also shown, which are used to assess the accuracy of
the analytical methods. It can be seen that the present method is
accurate for the whole range of the load for a 12.7 mm overlap.
The present method overestimates the bending moments at the
stiff end but underestimates the bending moments at the flexible
end. The bending moments at the stiff adherend are higher than
those for the flexible adherend. It should be pointed out that the
error at the flexible adherend is a little larger than at the stiff one.
The reason may be that the flexible adherend end bends more
than the stiff one and therefore the rigid assumption is better at
the stiff end than at the flexible one. However, they are accurate
enough to be used practically. For joints with a 25 mm overlap, the
present method is accurate for up to 400 kN/m load. The errors
become larger as the load is increased as shown in Fig. 15. Again,
Fig. 13. Bending moments for 50 mm overlap and stiff adherends (t1 = t2 =1.6 mm,
E1 = E2 = 207 GPa).
Fig. 14. Bending moments for 12.7 mm overlap for stiff–flexible adherends
(t1 = t2 = 1.6 mm, E1 = 207 GPa, E2 = 70 GPa).
 ARTICLE IN PRESS
X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]
the present method overestimates the bending moments at the
stiff end but underestimates the bending moments at the flexible
end. However, for medium load, typically 10 kN for a joint with a
25 mm width, the error is about 10%, which is still acceptable for
engineering applications.
Strictly speaking, Goland and Reissner’s method is not
applicable to joints with different adherends, but it can be
adapted to calculate the bending moments for joints with
different adherends. The procedure is as follows. When calculat-
ing bending moments for joints with different adherends, the
joint is considered as having two identical adherends. Two
bending moments may then be calculated by inputting the
material properties of either adherend into the Goland and
Reissner equation. Surprisingly, the calculated bending moments
agree well with the FE results as is shown in Fig. 15.
The above discussion concerns 1.6 mm adherends. To model
the effect of the thickness of adherends on the accuracy of the
determination of bending moments, analyses were made with 3.2
and 6.35 mm adherends which had the same material properties.
However, it should be noted that for very thick adherends
(6.35 mm) with stiff materials (E= 207 GPa), the assumption in
Eqs. (15a) and (15b) is not valid as can be seen in Fig. 16. Fig. 16
shows that a large difference exists between the two curves,
Fig. 15. Bending moments for 25 mm overlap for stiff–flexible adherends
(t1 =t2 = 1.6 mm, E1 =207 GPa, E2 =70 GPa).
Fig. 16. Comparison between functions sin h zl and cos h zl for stiff and thick
adherends (t1 = t2 =6.35 mm, E1 =207 GPa, E2 =207 GPa).
Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
7
which, according to the assumption in Eqs. (15a) and (15b), should
be very close to each other. Figs. 17–20 show the bending
moments for 3.2 and 6.35 mm adherends with 12.7 and 25 mm
overlaps. The predictions in Figs. 17–20 are based on Eqs. (13) and
(14). It can be seen that the present method is more accurate for
thick adherends than for thin ones as discussed above. Figs. 21
and 22 show the results obtained by using Eqs. (13) and (14) and
the simplified Eqs. (18) and (19). The results from the FE method
are also shown. They clearly show that there is a big difference
between the results from Eqs. (13) and (14) and the simplified
Eqs. (18) and (19). As a result, Eqs. (13) and (14) should be used
instead of Eqs. (18) and (19) for stiff and thick adherends. The
invalidity of the assumption in Eqs. (15a) and (15b) was neither
addressed by Goland and Reissner [2] nor by Hart-Smith [5]. From
this analysis, it may be concluded that care needs to be taken with
thick and stiff adherends (e.g. a thick steel adherend) when using
Eqs. (18) and (19). The conditions in Eqs. (15a) and (15b) need to
be satisfied to use Eqs. (18) and (19). Otherwise, Eqs. (13) and (14)
should be used. This will be further demonstrated below.
3.2.2. Different thickness of adherends
To model the second type of unbalanced joint, the adherend
materials were kept the same (aluminium), but the thickness of
one of the adherends was increased to five times larger than that
Fig. 17. Bending moments for 12.7 mm overlap with stiff–flexible 3.2 mm
adherends without using simplified equations (E1 = 207 GPa, E2 = 70 GPa).
Fig. 18. Bending moments for 12.7 mm overlap with stiff–flexible 6.35 mm
adherends without using simplified equations (E1 = 207 GPa, E2 = 70 GPa).
 ARTICLE IN PRESS
8 X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]]

Fig. 19. Bending moments for 25 mm overlap with stiff–flexible 3.2 mm adherends Fig. 22. Bending moments with 25 mm overlap for stiff–flexible 6.35 mm
without using simplified equations (E1 = 207 GPa, E2 = 70 GPa). adherends using with and without using simplified equations (E1 =207 GPa,
E2 = 70 GPa).

Fig. 20. Bending moments for 25 mm overlap with stiff–flexible 6.35 mm

adherends without using simplified equations (E1 = 207 GPa, E2 = 70 GPa). Fig. 23. Bending moments with 12.7 mm overlap for thick–thin adherends
without using simplified equations (t1 = 5t2 =8 mm, t2 = 1.6 mm, E1 =E2 = 70 GPa).

Fig. 21. Bending moments with 12.7 mm overlap for stiff–flexible 6.35 mm
adherends using with and without using simplified equations (E1 = 207 GPa, Fig. 24. Bending moments with 25 mm overlap for thick–thin adherends without
E2 =70 GPa). using simplified equations (t1 = 5t2 = 8 mm, t2 =1.6 mm, E1 = E2 = 70 GPa).

of the other one as shown in Fig. 2. Joints with 12.7 and 25 mm present method with Equations 13 and 14 yields accurate results
overlaps were used. Again, comparison is only made with the FE for both thick and thin adherends for the two overlaps. It can also
method as shown in Figs. 23 and 24. It is clearly shown that the be seen that the bending moments in the thick adherend are

Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001
 ARTICLE IN PRESS
X. Zhao et al. / International Journal of Adhesion & Adhesives ] (]]]]) ]]]–]]] 9

Therefore, Eqs. (18) and (19) produce a significant error for the
stiff and thick adherends, as shown in Fig. 25. As a result, care
must be taken when using the simplified Eqs. (18) and (19). In the
case of steel, the simplified equations are applicable up to 4 mm
thick adherends whereas in the case of aluminium, the adherend
thickness can go up to 6 mm because of its lower modulus
for loads needed to cause failure (600 kN/m). Fig. 26 shows a
comparison between functions sinh zl and cosh zl for the case of
steel.

4. Conclusions

A new method for the determination of bending moments for

single lap joints has been proposed and assessed by comparison
with other analytical models and the FE method. The following
conclusions can be drawn:
Fig. 25. Bending moments for a thick–thin adherends joint with 25 mm overlap
and stiff adherends (t1 = 5t2 = 8 mm, t2 = 1.6 mm, E1 = E2 = 207 GPa). 1. The new method is accurate for single lap joints with
reasonably long overlaps, i.e., for a length/thickness ratio up
to 30.
2. It can be used for identical adherends or, in particular, for non-
identical adherends for which the present method is very
simple but accurate.
3. The thicker and stiffer the adherends, the more accurate is the
present method.
4. The bending moments at the ends of the overlap are induced
by the eccentric geometry and have little to do with the
stiffness of the overlap.
5. Care must be taken to use the appropriate equations when
analysing stiff and thick adherends. For steel adherends, the
method is applicable for thicknesses up to 4 mm whereas for
aluminium the adherend thickness can be as thick as 6 mm.

References

[1] Volkersen O. Luftfahrtforschung 1938;15:41.

Fig. 26. Comparison between functions sin h zl and cos h zl for steel with 4 mm [2] Goland M, Reissner E. Journal of Applied MechanicsJ Appl Mech 1944;66:A17.
pffiffiffiffiffiffiffiffiffi
thick adherends (z ¼ T=D, D= Et3/12(1  n2)). [3] Renton J, Vinson JR. J Adhesion 1975;7:175.
[4] Allman DJ. The Quarterly Journal of Mechanics and Applied MathematicsJ
Mech Appl Math 1977;30:415.
[5] Hart-Smith LJ. NASA Contract Report 1973, NASA CR-112236.
[6] Wah T. ASME Journal of Engineering Materials and Technology 1973;95:174.
much higher than in the thin adherend. The bending moments [7] Pirvics J. J Adhesion 1974;6:207.
increase almost linearly with the applied load. The effect of the [8] Srinivas S. NASA Technical Note 1975, NASA TN D-7855.
[9] Cheng S, Chen D, Shi Y. ASCE Journal of Engineering Mechanics 1991;117:605.
stiffness of adherends on the accuracy of the calculations of the [10] Yang C, Pang SS. ASME Journal of Engineering Materials and Technology
bending moments is shown in Fig. 25, in which the stiffness of 1996;118:247.
both the adherends is 207 MPa and the length of the overlap [11] Sawa T, Liu J, Nakano K, Tanaka J. J Adhesion Sci Technol 2000;14:43.
[12] Zou GP, Shahin K, Taheri F. Composite Structures 2004;65:499.
25 mm. It can be seen that the present method (with Eqs. (13) and [13] Lee J, Kim H. J Adhesion 2005;81:443.
(14)) for the stiff adherends gives more accurate results than for [14] Adams RD, Mallick V. J Adhesion 1992;38:199.
flexible ones, as shown in Fig. 24. It can be seen in Fig. 25 that the [15] da Silva LFM, das Neves PJC, Adams RD, Spelt JK. Int J Adhes Adhes
2009;29:319.
present model is closer to the FE points than in Fig. 24. This is in
[16] da Silva LFM, das Neves PJC, Adams RD, Wang A, Spelt JK. Int J Adhes Adhes
line with the above discussions that the stiffer the adherends, the 2009;29:331.
more accurate the present method. Fig. 25 shows results based on [17] Guo S, Dillard DA, Plaut RH. Int J Adhes Adhes 2006;26:629.
Eqs. (13) and (14) and also based on the simplified Eqs. (18) and [18] Grant LDR, Adams RD, da Silva LFM. Int J Adhes Adhes 2009;29:405.
[19] da Silva LFM, Adams RD. J Adhesion Sci Technol 2005;19(2):109.
(19). For the stiff and thick adherends, the conditions for the [20] da Silva LFM, da Silva RAM, Chousal JAG, Pinto AMG. J Adhesion Sci Technol
assumptions in Eqs. (15a) and (15b) are seriously violated. 2008;22:15–29.

Please cite this article as: Zhao X, et al. A new method for the determination of bending moments in single lap joints. Int J Adhes Adhes
(2009), doi:10.1016/j.ijadhadh.2009.09.001