Académique Documents
Professionnel Documents
Culture Documents
+ + + + +
T
T zx yz xy x z z y y x
Y J
Y
F
F
where
T
Y is the yield stress in uniaxial tension. It is only a measure of the deviatoric stresses
and is independent of the hydrostatic stresses. Therefore, it is applicable to flexible aluminium
adherends. However, the polymers and adhesives yield different in tension and compression,
thus, the hydrostatic stresses are important. Therefore, this is to be modified.
b. Modified von Mises yield surface is expressed as
( ) ( ) [ ] 0 12 1 1
2
1
2
1
2
2 2
1 1
1
]
1
+ +
T
Y S J S J S J
S
F
where
T
C
Y
Y
S , and
C
Y is the yield surface in compression. Raghava [35] proposed for the
epoxy adhesives
( )
T C T C
Y Y J Y Y J +
1 2
3
and Crocombe [20] obtained eq (4) by modifying eq (5). When 1 S in the eq (4), it reduces
to eq (3).
3.2 Isotropic Hardening
(2)
(3)
(1)
(4)
(5)
Material generally supports increased load as it seldom yields at a constant stress. It is known
as work hardening. This can be modelled by changing the yield surface and it is necessary to
increase
T
Y . The
T
Y value is obtained from the uniaxial stress-strain curve at an approximate
equivalent strain. The general form of the yield surface is
( ) ( )
T
Y f F
where is some hardening parameter.
3.3 Plastic Stress-Strain Relations
Total strain can be divided into elastic and plastic strain components as
p e
! ! ! +
The elastic strains
e
! can be obtained from the stresses using the elastic-strain relations as
follows
" D !
1
e
where D is the modulus matrix. The plastic strain components are usually determined by
considering increments of strain. Levy [36] and von Mises [37] first suggested that the total
strain increments were proportional to their respective deviatoric stresses and proposed an
incremental relationship. Prandtl [38] and Reuss [39] modified the Levy-Mises equations and
used the plastic strain increments rather than the total strain increments proportional to the
instantaneous deviatoric stress which can be written as
d
d d
y
py
x
px
!
' '
where d is some instantaneous constant. This gives a general flow rule [40]. The rule can be
given as
,
_
f
d d
p
!
This is known as the normality principle because the plastic strain increment to be normal to
the yield surface.
3.4 Application of Plasticity Theory to The Finite Element Method
In the finite element method, a solution is sought for the equation
# " B R $ dV
T
where $ residual vector, R load vector, B strain-displacement matrix and " is stress
matrix. For small displacement, thus, elastic problems
0
B B and % B D "
0
where
0
B is a matrix consist of derivatives of shape functions and eq (11) can be written as
0 % K R $ where
dV
T
0 0
DB B K
is the secant stiffness which is symmetric and constant, therefore, the solution of
displacements is direct. In the large displacement problems
L
B B B +
0
and % B B D "
1
]
1
+
L
2
1
0
(6)
(7)
(8)
(9)
(10)
(11)
(13)
(12a)
(12b)
where
L
B is a matrix being function of derivatives of shape functions and displacements. In
this case, the solution of eq (11) is more difficult because
L
B is a function of displacement
and the secant stiffness is non-symmetric. When plasticity occurs, the modulus matrix D is
no longer constant. The solution is achieved iteratively using Newton-Raphson scheme. In
addition, the full load is applied incrementally and a solution is obtained at each step.
An initial guess to the next load increments is made by
j
i
j
i
j
i
j
i
d
d
1
1 1
1
,
_
%
$
$ % %
where i and j are load step and iteration number. If eq (11) is differentiated
% K $ d d
T
where
[ ]
+ dV
ep
T
T
G S G B D B K
where G and S are matrix of shape function derivatives and stress array, respectively. More
details can be found in refs [41-43]. The only difference in eq (16) is the elastic-plastic
modulus matrix
ep
D which links an increment of stress to an increment of strain. The
tangential stiffness in eq (16) can be computed by replacing the modulus matrix D by the
elasto-plastic modulus matrix
ep
D in eq (13). Since this process achieves convergence in less
iteration, it was used in the analysis.
3.5 Plastic Stress-Strain Behaviour
If we combine the total strain expression eq (7) with the expression of elastic-plastic strain
increments eqs (8,10) yields
a " D ! d d d +
1
where
"
f
a
d
d
. The stresses remain on the yield surface during yielding, and so from eq (1)
( ) ( ) { }
0
d A d dF
Y f d dF
T
T
" a
where
d
d
d
dY
A
T
1
'
,
_
'
After re-arranging eq (17) and substituting eq (19) gives
! D " d d
ep
where
( )
'
Da a
D a a D D
D
T
T
ep
A
.
(14)
(16)
(15)
(17)
(18)
(19)
(20)
Stress increments corresponding to strain increments can be determined in terms of
A ve a
in eq 20. Eq (11) can also be used to obtain the plastic strain increments. More details can be
found in Refs [41-43].
3.5.1 Yield Function Derivatives ( ) a
The derivatives of the yield function can be expressed in terms of the stress invariants:
a) von Mises
2
3J f
xy
xy
x
x x
J
f
J
J
J
f f
,
_
,
_
,
_
,
_
2
'
2
2
2
3
3
2
1
b) Modified von Mises
( ) ( ) [ ]
'
+ +
2
1
2
2 2
1 1
12 1 1
2
1
S J S J S J
S
f
,
_
,
_
,
_
,
_
x x x
J
J
f J
J
f f
2
2
1
1
( )
( ) [ ]
2
1
2
2 2
1
'
1
2
12 1
1
3
2
1
2
1
S J S J
J
S
S
S
S f
x
x
+
1
1
]
1
( ) [ ]
2
1
2
2 2
1
12 1
6
S J S J
f xy
xy
+
3.5.2 Hardening Parameter ( )
A
For uniaxial case,
A is
d
d
d
d
H
d
d
Y
A
up 1 1
1
1
]
1
,
_
1
]
1
,
_
where
up
d
dY
H
which is the slope of the uniaxial stress, plastic strain curve.
(21a,b)
(22a)
(22b)
(23)
3.6 Correction to the Yield Surface
The yield function derivatives a can also be used to restore the stresses to the yield surface if
application of the elastic-plastic modulus
ep
D is sufficiently accurate. In case the stresses do
not lie on the yield surface ( ) 0 F stresses can be corrected so that they are normal to the
yield surface. In case
p d a "
where p is a scalar. If the required stress change is small,
p d F dF
T T
a a " a
Re-arranging and substituting for p in the original equation for the stress change gives
( )
,
_
a a
a
T
F d
3.7 Solution Procedure
The following scheme of procedure assumes that variables have been obtained suitably for the
( )
th
i 1 iteration of a particular load increment. The large displacement solution procedures
can be found in ref [41-43] in detail. Attention is paid to the solution procedure of the
plasticity equations:
a) Corrective displacements, etc. are calculated,
[ ]
i i i i i T i
and % % % $ K % +
1 1
1
1
b) Strains and stresses are updated,
i i i i
% B ! ! +
1
[ ]
i i
i
ep i i
% B D " " +
1
c) Correct plastic stress corresponding to strain increment,
i. Stress is updated to estimated yield
ii. Remainder of increment is divided into steps
i
d!
iii. [ ]
j
i ep
D and
j
i
d are evaluated for the beginning of step j
iv. Calculation of
1. Stress [ ]
i
j
i
ep
j
i
j
i
d! D " " +
1
2. Plastic strain [ ] [ ]
j
i
j
i
j
i
p
j
i
p
d a ! ! +
1
3. Equivalent uniaxial plastic strain [ ] [ ]
j
i
j
i
up
j
i
up
d +
1
! !
v. New yield stress and slope are evaluated,
vi. If stress is too far from yield surface correct stress,
( ) a a
a
" "
T
T
j
i
j
i
F
vii. Go to (ii) and repeat for next step.
(24)
(25)
d) Residual for
th
i iteration is evaluated,
dV
i
T
i i i
" B R $
e) Residual for
th
i iteration is evaluated,
[ ] [ ] ( )
+ dV
i i
T
i i
i
ep
T
i i T
G S G B D B K
f) Convergence is tested. If convergence is satisfied, next load increment is applied, if
not, it is returned to (a) for another iteration.
Figure 1. Dimensions of a) a single lap joint, b) an adhesive specimen, and c) an aluminium
specimen.
4. JOINT CONFIGURATION AND EXPERIMENTAL
The single lap joints are used extensively for the determination of adhesive properties and the
prediction of the performance of the adhesively bonded joints having more complex
geometries. In this study, the elasto-plastic analysis of an adhesively bonded single lap joint
consisting of an adhesive layer and two aluminium plates (adherends) was carried out
experimentally and analytically.
The non-linear finite element analysis requires the mechanical properties of adhesive and
aluminium adherends be known. For this purpose, the uniaxial tensile tests of the adhesive
specimens and the aluminium specimens were carried out. The adhesive specimen has
dimensions as shown in Fig 1b. Araldite
!
! !
!
!
!
!
!
!
! !
!
!
!
!
!
!
!
!
!
!
!
0 0,02 0,04 0,06 0,08 0,1 0,12
0
20
40
60
80
100
120
140
0.0166 mm/sn
0.083 mm/sn
0.33 mm/sn
!
,
Uniaxial strain
U
n
i
a
x
i
a
l
s
t
r
e
s
s
,
N
/
m
m
2
Figure 6. True stress-strain curves for aluminium (ETIAL 5-12 H).
Aluminium adherends were prepared from ETIAL 5-1/2 H sheets in a thickness of 3 mm
produced by Etibank Aluminium Company. After the aluminium sheets were annealed at
345C, the aluminium specimens were produced in the dimensions as shown in Fig 1c. The
true stress-strain curves of the aluminium specimens in Fig 5 were obtained at three different
rates of testing of 0.0166, 0.083 and 0.33 mm/s as shown in Fig 6. These stress-strain curves
were also fitted using the cubic spline fitting technique.
Figure 7. Mesh details of an adhesively bonded single lap joint.
5. NON-LINEAR STRESS ANALYSIS
The elasto-plastic analysis of an aluminium single lap joint was carried out using the non-
linear finite element method. The dimensions of the single lap joint are given in Fig 1a. In the
finite element method, any continuous structure is divided to small regions having the same
properties and with finite size and including nodes at their corners and edges. The elements
are connected through these nodes. The finite element type is also important in modelling
the structure and to achieve a reasonably accurate solution. Since the problem requires a two-
dimensional (plane strain) analysis, eight-noded isoparametric quadratic quadrilateral plane
elements with four integration points were used to model adhesive layer and aluminium
adherends. A commercial finite element software ANSYS (current version 5.3) was used to
solve the problem [44].
The analysis considers material non-linearity and requires the uniaxial stress-strain
behaviours of the epoxy adhesive and aluminium adherends. The experimental data shown
in Figs 4 and 6 were used in the non-linear finite element analysis to model the stress-strain
behaviours of the adhesive and adherends. The adhesive fillets formed around the adhesive
free ends as a result of squeezing the adhesive layer between two aluminium plates were
ignored for the simplicity of the analysis. The single lap joint is subjected to a tensile load of
10 kN at the right free end of the lower adherend and whereas the left free end of the upper
plate is fixed all direction the right free of the lower plate was allowed to move in the
direction of the applied load as shown in Fig 7. Since the adhesive free ends are subjected to
high stress and strain concentrations these critical regions were refined until the reasonable
results are obtained. The free ends of the adhesive layer were divided into 16 elements to
increase computation accuracy as shown in Fig 7.
The full tensile load was applied to the joint incrementally, and a new solution corresponding
to each load step was obtained after the convergence value was satisfied based on forces and
displacements at the nodes of the each finite element forming the model of the single lap
joint. This process is needed in the non-linear finite element analysis to avoid the possibility
of convergence to an incorrect solution. In addition, the geometrical non-linearity (effect of
the large displacements) was also taken into account in the analysis.
Figure 8. Deformed and undeformed joint geometries.
In the analysis, the high stress and strain concentrations occurred around the adhesive free
ends. The large rotations were encountered in the overlap region. These rotations cause high
bending moments around the adhesive free ends; consequently, high peeling stresses arise at
the adhesive free ends. The deformed and undeformed geometries are shown in Fig 8. The
plastic strains in the adhesive layer and in the aluminium adherends occurred when 11% of
the full load was applied. The plastic strains in the adhesive layer initiated from the free ends
of the adhesive-upper and lower aluminium adherend interfaces and propagated into the
middle joint regions. This propagation occurred along the adhesive-adherend interfaces.
When the plastic strain penetration from each free end of the adhesive interfaces reached a
small region in the middle section of the joint this small region was not able to withstand the
applied tensile load, therefore, the single lap joint was failed through this small middle
adhesive layer. The plastic strains in the upper and lower adherends initiated from the free
ends of the adhesive-adherend interfaces. The lower surface of the upper adherend was in
tension whereas its upper surface was in compression.
Figure 9. The distributions of a) von Mises stress, b) elastic von Mises strain, c) plastic von
Mises strain, and d) total von Mises strain in the joint region for a load of 2.8 kN.
The plastic strain propagated along the lower and upper surfaces of the upper adherend to its
fixed end. The lower adherend was also presented similar behaviour, thus the upper surface
of the lower adherend was in tension whereas its lower surface was in compression. The
plastic strains propagated along the lower and upper surfaces of the lower adherend,
especially along the surfaces in tension. However, it was observed that the tensile plastic
strains covered a large region around the adhesive free ends in the higher load steps. Figure
9a shows the distribution of von Mises stresses in the joint region. The small regions at the
free ends of the adhesive-adherend interfaces corresponding to about 11% of the overlap
length is ruptured totally along the adhesive-adherend interface, and the magnitude of the
stresses decreases through the middle joint region. It is also evident that a large region was
subjected to lower stresses. The elastic von Mises strains arise from the region in which the
plastic strain region finishes as shown in Fig 9b, however, these strains would not cause
permanent deformations in the adhesive joints. The plastic von Mises strains are concentrated
around the adhesive free ends as shown in Fig 9c and d, and propagate into the middle
adhesive region and along the surfaces under tension of the aluminium adherends.
Figure 10. Test rig used for the curing period of the single lap joint specimens.
In order to compare the analysis results with the experimental results, a series of specimens
of the single lap joints was prepared. The strength and durability of the single lap joints are
dependent on pretreatment of surfaces to be bonded. Therefore, the bonding surfaces were
degreased with trichlorethylene, etched in chromic acid solution for 20-30 minutes at 60C,
rinsed in cold water and dried in warm air. Later, the epoxy adhesive was applied with a
spatula to bonding surface of the aluminium adherends, and the two aluminium adherends
were bonded by means of the test rig shown in Fig 10. The air bubbles were removed from
the adhesive layer as possible. The single lap joint specimen in the test rig was held in an
oven at 130C for 4 hours for the curing process, and left to cooling at room temperature.
Finally, their tensile tests were carried out under a tensile load of 10 kN. The fracture
surfaces of the failed single lap joints were examined by means of SEM (Scanning Electron
Microscope). The SEM photographs of the fracture surfaces of the single lap joints are
shown in Fig 11. A certain region around the adhesive free ends along the overlap region of
an adherend is blank whereas those of the other adherend include peeled adhesive portions,
so this means that the fracture mechanism along the adhesive layer was interfacial. However,
a large region in which the fracture propagates along the adhesive layer is still exists. This
explains why a small zone adhesive layer in the middle overlap region was a final zone of the
adhesive layer along which the full load was transferred.
The experimental study presents similar deformation and stress states of the adhesive layer
and the aluminium adherends to those of the non-linear stress analysis. However, some
modifications are needed. This study is a part of the research programme in which the
analysis methods are developed to predict the strength of adhesive joints. The non-linear
analysis was carried out for a specific joint configuration.
Figure 11. Fracture surface SEM photographs of a single lap joint specimen.
Therefore, the analysis results can give information about the non-linear behaviour of an
adhesive single lap joint, and can provide a background for the analysis of the adhesive joints
with more complicated geometry. However, the experimental and theoretical studies are
continuing for different metals, alloys and adhesive types, and the load-bearing capabilities
of the adhesive joint are tested.
6. CONCLUSIONS
The non-linear stress analysis and the experimental study of an adhesively bonded single lap
joint showed that the most critical joint regions were the free ends of the adhesive layer, thus
adhesive plastic strains initiated at the free ends of the adhesive-adherend interfaces, and the
plastic zones spread along these interfaces and across the adhesive thickness. Therefore, the
joint failure was interfacial along the adhesive layer. In addition, the plastic zones in the
aluminium adherends initiated the adherend regions neighbouring to the critical adhesive
locations, and spread the adherend surfaces in tension, therefore, the aluminium adherends
may subject to plastic deformations as much as the adhesive layer. Therefore, the aluminium
adherends may yield considerably in case of the rubber modified epoxy adhesives. The
analysis results are limited of a specific adherend material and adhesive type. The effects of
some parameters, such as adhesive fillet, stronger adhesive, overlap length, and adhesive
thickness should be analysed in order to obtain more realistic prediction of the single lap
joints. On account of this, the analysis results should be evaluated for understanding how the
elasto-plastic deformation mechanism in the adhesive layer and the adherends of a single lap
joint develops.
7. REFERENCES
1. Semerdjiev S. (1970), Metal to Metal Adhesive Bonding, Business Books Ltd, London.
2. Adams R.D. and Wake W.C. (1984), Structural Adhesive J oints in Engineering,
Elsevier Applied Science, London.
3. Kinloch A.J . (1987), Adhesion and Adhesives, Chapman and Hall, London.
4. Volkersen O. (1938), Die Nietkraftverteilung in Zugbeanspruchten Nieverbindungen
mit Konstanten Laschenquerschnitten, Luftfahrtforschung, vol 15, pp 41-47.
5. Volkersen O. (1965), Construction Metallique, vol 4, pp 3-13.
6. Goland M. and Reissner E. (1944), Trans. ASME, J . Appl. Mech., vol 11, A17-A27.
7. Hart-Smith L.J . (1972), McDonnell-Douglas Co. Report No: 6059A, California.
8. Hart-Smith L.J . (1973), NASA CR-112235, Langley Research Centre, Virginia.
9. Hart-Smith L.J . (1973), NASA CR-112236, Langley Research Centre, Virginia.
10. Hart-Smith L.J . (1973), NASA CR-112237, Langley Research Centre, Virginia.
11. Hart-Smith L.J . (1973), NASA CR-112238, Langley Research Centre, Virginia.
12. Hart-Smith L.J . (1974), NASA CR-2218, Langley Research Centre, Virginia.
13. Hart-Smith L.J . (1981), in: Developments in Adhesives - 2, A.J . Kinloch (Ed.), pp 1-
44, Elsevier Applied Science, London.
14. ESDU 78042 (1978), Shear Stresses in the Adhesives in Bonded J oints, Single Step
Double Lap J oints Loaded in Tension, Engineering Sciences Data Unit, London.
15. ESDU 79016 (1979), In Elastic Shear Stresses and Strains in the Adhesive Bonding
Lap J oints Loaded in Tension or Shear, Engineering Sciences Data Unit, London.
16. Grant, P. (1976), Strength and Stress Analysis of Bonded J oints, British Aircraft
Corp. Ltd. Rep. 50R(P).
17. Harris, J .A. and Adams, R.D. (1982), The Impact Strength of Adhesive Lap J oints,
Proceedings of the International Symposium on Adhesive J oints, Kansas City, pp 611-
626.
18. Adams, R.D. and Harris, J .A. (1987), The Influence of Local Geometry on the
Strength of Adhesive J oints, Int. J . Adhesion and Adhesives, vol 7, no 2, pp 69-80.
19. Adams, R.D. Atkins, R.W., Harris, J .A. and Kinloch, A.J . (1986), Stress Analysis and
Failure Properties of Carbon-Fibre-Reinforced-Plastic/Steel Double-Lap J oints, J .
Adhesion, vol 20, pp 29-53.
20. Crocombe, A.D. (1989), Global Yielding as a Failure Criterion for Bonded J oints,
Int. J . Adhesion and Adhesives, vol 9, no 3, pp 167-178.
21. Bigwood, D.A. and Crocombe, A.D. (1990), Non-linear Adhesive Bonded J oint
Design Analyses, Int. J . Adhesion and Adhesives, vol 10, no 1, pp 31-41.
22. Crocombe, A.D. and Bigwood, D.A. (1992), Development of a Full Elasto-plastic
Adhesive J oint Design Analysis, J . of Strain Analysis, vol 27, no 4, pp 211-218.
23. Crocombe, A.D., Bigwood, D.A. and Richardson, G. (1990), Analysing Structural
Adhesive J oints for Failure, Int. J . Adhesion and Adhesives, vol 10, no 3, pp 167-178.
24. Czarnocki, P. and Piekarski, P. (1986), Non-linear Numerical Stress Analysis of a
Symmetric Adhesive-Bonded Lap J oint, Int. J . Adhesion and Adhesives, vol 6, no 3
pp 157-160.
25. Reddy, J .N. and Roy, S. (1988), Non-linear Analysis of Adhesively Bonded J oints,
Int. J . Non-linear Mechanics, vol 23, no 2, pp 97-112.
26. Edlund, U. and Klarbling, A. (1990), Analysis of Elastic-plastic Adhesive J oints
Using a Mathematical Programming Approach, Computer Methods in Applied
Mechanics and Engineering, vol 78, pp 19-47.
27. Edlund, U. and Klarbling, A. (1992), A Geometrically Non-linear Model of the
Adhesive J oint Problem and its Numerical Treatment, Computer Methods in Applied
Mechanics and Engineering, vol 96, pp 329-350.
28. Sawa, T., Aoki, M. and Nishikawa, O. (1997), Elasto-plastic Finite Element Analysis
and Strength Evaluation of Adhesive Butt J oints of Similar and Dissimilar Hollow
Shafts Subjected to External Bending Moments, J . of Adhesion, vol 61, no 1-4, pp 55-
69.
29. Kawawaki, M., Nakano, Y. and Sawa, T. (1998), Elasto-plastic Finite Element Stress
Analysis and Strength Evaluation of Adhesive Lap J oints of Hollow Shafts Subjected
to Tensile Loads, J . of Adhesion Science and Technology, vol 12, no 9, pp 907-922.
30. Pandey, P.C., Shankaragouda, H. and Singh, A.K. (1999), Non-linear Analysis of
Adhesively Bonded Lap J oints Considering Viscoplasticity in Adhesives, Computers
and Structures, vol 70, no 4, pp 387-413.
31. Sato, C. and Ikegami, K. (1999), Strength of Adhesively-Bonded Butt J oints of Tubes
Subjected to Combined High-Rate Loads, J . of Adhesion, vol 70, no 1-2, pp 57-73.
32. Yang, Q.D., Thoules, M.D. and Ward, S.M. (1999), Numerical Simulations of
Adhesively-Bonded Beams Failing with Extensive Plastic-Deformation, J . of The
Mechanics and Physics of Solids, vol 47, no 6, pp 1337-1353.
33. Malwern, L.E. (1969), Introduction to the Mechanics of a Continuous Medium,
Prentice-Hall.
34. J ohnson, W. and Mellor, P.B. (1973), Engineering Plasticity, Van Nostrand Reinhold.
35. Raghava, R.S. and Cadell, R.M. (1973), The Macroscopic Yield Behaviour of
Polymers, J . Material Science, vol 8, pp 225.
36. Lvy, M. (1870), Mmare sur les equations gnrales des mouvements intrieurs des
corps solides ductiles au del des limites ou lelasticit pourrait les ramener leur
premier elat, Comptes. Rendus, Paris, vol 70, pp 1323.
37. Mises, R. von (1928), Mechanik der festen Krper in plastisch deformablem Zustant,
Gttingen Nachr. Maths. Phys. Vol 1, pp 582.
38. Prandtl, L. (1924), Spannungsverteilung in Plastichen Krpem, Proc. 1
st
Int. Cong.
Appl. Mech. Delft, pp 43.
39. Reuss, A. (1930), Berucksichtigung der Elastichen Formanderungen in der
Plastizitatststherie, Zeits. Ang. Math. Mech. vol 10, pp 266.
40. Zienkiewicz, O.C. and Taylor, R.C. (1991), The Finite Element Method (Solid and
Fluid Mechanics, Dynamics and Non-linearity), vol. 2, McGraw-Hill Company, UK.
41. Crisfield, M.A. (1993), Non-linear Finite Element Analysis of Solids and Structures,
vol I, J ohn Wiley.
42. M. Kleiber (1989), Incremental Finite Element Modelling in Non-linear Solid
Mechanics, Ellis Horwood.
43. Ishai, O. (1967), Delayed Yielding of Epoxy Resin Under Tension, Compression or
Flexure I. Behaviour Under Constant Strain Rate, J . Appl. Poly. Sci. vol 11, pp 963.
44. ANSYS