You are on page 1of 9

Machine Translated by Google ARTICLE IN PRESS

International Journal of Adhesion & Adhesives 29 (2009) 405–413

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives


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

Experimental and numerical analysis of single-lap joints


for the automotive industry
a a
LDR Grant , R.D. Adams , Lucas FM da Silva b,
a
Department of Mechanical Engineering, University of Bristol, Queen's Building, University Walk, Bristol BS8 1TR, UK
b
Departamento de Engenharia Mecaˆnica e Gesta˜o Industrial, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

article info abstract

Article history: Lap joints are used extensively in the manufacture of cars. In order to determine the effect of using a
Accepted 11 September 2008 structural adhesive instead of spot-welding, a detailed series of tests and finite element analyzes were
Available online 17 September 2008 conducted using a range of loadings. The adhesive was a toughened epoxy and the adherend was mild
Keywords: steel typical of that used in the manufacture of car bodyshells. The lap joints were tested in tension (which
Epoxy/epoxides creates shear across the bondline), four-point loading (pure bending) and three-point loading (bending plus
steels shear). Various parameters were investigated such as the overlap length, the bondline thickness and the
lap shear spew fillet. The major finding is that three-point bending and tension loading are very similar in the way in
Finite element stress analysis which they affect the adhesive while the four-point bend test does not cause failure because the steel yields
Four-point bending
before the joint fails. A failure criterion has been proposed based on the tensile load and bending moment
Three-point bending
applied to the joint.
& 2008 Elsevier Ltd. All rights reserved.

1. Introduction (ii) low-strength adhesives, (iii) medium-strength adhesives and (iv)


high-strength adhesives. The high-strength category of adhesives is
The automotive industry has recently implemented what the used where the adhesive plays the primary role in the joining and
aerospace industry has been using for decades, namely that strength of a structure. At present, there is very little structural
adhesives can be used for joining load-bearing components. As the adhesive used in the car bodyshell. The first step towards greater
designers of road vehicles try to produce cheaper and lighter use of structural adhesives is the characterization of those joints
products, more ways are needed for joining new and dissimilar found typically in the automotive industry by a combination of testing
materials together. The main method of joining in the automotive and analysis, in order to improve the knowledge of the behavior of
industry is by means of spotwelds. This has required large investment these joints [3] . The objective of this research is to increase the
in the appropriate technology, such as highly automated production amount of data available to the automotive design engineer. Since
lines and many years' experience of designing. a single-lap joint is widely known and used to characterize bond
However, there are disadvantages with spotwelds as they require strength, the testing program was started with that joint and the
access to both sides of the joint, they cannot join aluminum results obtained are presented here. The single-lap joint may be
effectively, or composites at all, and they generally destroy any loaded in tension or in bending (sometimes both together) and so
coatings used to improve the corrosion resistance of steels. these two loading conditions are considered here.
A good, cheap method that can solve all the above problems is to
use adhesive bonding. In order to use adhesives, they must fulfill Extensive work has been done on the testing and modeling of
the performance requirements. The fundamental problems with single-lap joints under tension [4–6]. The pioneering works of
using adhesives have been overcome, such as bonding directly to Volkersen [7] and Goland and Reissner [8] have been developed
oily steel [1], high-speed application of adhesives using robotic over the years to take into account various parameters such as
technology in high-rate production lines, and the development of adhesive plasticity [9,10], the free stress at the end of the overlap
toughened adhesives that can withstand impact better than the older [11–13 ], composite adherends [14–17], adherend plasticity [4,18]
brittle forms. Adhesives are used today in a variety of places in the and the adhesive fillet [19]. For complex geometry such as adhesive
vehicle [2], and can be split into four categories; (i) sealants, fillets or adherends shaping, the finite element (FE) method is very
useful. Recent studies dealing with the adhesive fillet include those
of Belingardi et al. [20] and Andreassi et al.
Corresponding author. Tel.: +351 22 5081706; fax: +351 22 5081445. [21]. The FE method was also used by Fessel et al. [22] to study
E-mail address: lucas@fe.up.pt (LFM da Silva). single-lap joints with 'wavy adherends' for automotive

0143-7496/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijadhadh.2008.09.001
Machine Translated by Google ARTICLE IN PRESS

406 LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413

applications. When there is yielding of the adherend, Adams et al. [4]


have shown that failure is dictated by the adherend yielding and the
failure load prediction can be found simply by calculating the initial
yielding of the adherend. A similar conclusion was obtained by da
Silva and Adams [23] in the case of T joints. It is known that lap shear
strength decreases as the adhesive thickness increases, but the
range of adhesive thicknesses rarely goes beyond 1 mm. The results
presented here show what happens up to very thick bondlines (3
mm), with and without a spew fillet. The reason for the decrease in
joint strength as the adhesive thickness increases is still controversial.
Some authors have shown that this effect can be explained by the
interface stresses [24,25], which increase as the bondline increases
but there is no clear physical explanation.
Fig. 2. Steel tensile stress–strain curve.

Compared with the case of tensile loading, single-lap joints under


external bending loads are not well documented.
Four-point bending has been addressed by Liu et al. [26] both
experimentally and theoretically. The theory could not predict properly
the joint strength because it did not include adherend yielding that
occurred in the tests. Ozel et al. [27,28] also studied the four-point
bend test but used high-strength steel for the adherends. The effect
of the adherend thickness, adherend overlap and type of adhesive
was studied.
In the present study, the lap joints were tested in tension, four-
point bending (pure bending) and three-point bending (bending with
shear). The behavior of the joints was examined both experimentally
and analytically when various parameters, such as the overlap length,
the glueline thickness and the geometry of the edge of the overlap
region were changed.

2. Experimental program Fig. 3. Single-lap joint geometry.

2.1. materials 2.2. Specimen geometry

A single adhesive was used, ESP110 from Permabond. This was The geometry of the basic lap joint that was tested is given in Fig.
an adhesive that had been developed for use in the structural bonding 3. The single-lap shear test is often used to compare shear strengths
of car bodyshells. It is a toughened, single-part paste epoxy with 30% of adhesives and is detailed in ASTM D 1002.
aluminum powder that is heat cured at 150 1C for 45 min. Uniaxial The overlap length varied between 7.5 and 45 mm and the
testing was done on bulk samples and the stress–strain curve is adhesive thickness between 0.1 and 3.0 mm. For tension testing, the
represented in Fig. 1. The strain was measured using a laser strain overlap was kept at 15 mm. Both types of end geometry, the full 451
gauge developed in-house [29]. spew fillet and the square end, were used.
The adherend was mild steel as used in the manufacture of the
car bodyshell, and made to the specification BS1449 CR1E, and is
2.3. Specimen manufacture
typically 0.95 mm thick. The tensile stress–strain curve, from dog
bone specimens, is presented in Fig. 2.
The lap joints were manufactured with a standard, full-depth spew
fillet of 451 and a square end. The lap joints were produced for the
three-point and four-point bend tests with spew fillets with the
exception of one batch that was manufactured with square ends. For
tension testing, both types of geometry were manufactured and tested.

The adherends were degreased with acetone to remove surface


contamination, especially oil, and then grit blasted to give a surface
finish of 2.5 mm. The grit-blasted surface was again degreased using
acetone.
The joints were manufactured in a jig, which allowed very
consistent bondline thicknesses to be produced. Shims were used to
control the adhesive thickness and the adhesive fillet.

2.4. test procedure

The lap joints in tension were tested at a rate of 5 mm/min.


For the bending tests, the loading described in Fig. 4 was used.
Fig. 1. Adhesive ESP110 tensile stress–strain curve. For the four-point bend tests, the loading points were kept at a
Machine Translated by Google ARTICLE IN PRESS

LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413 407

Fig. 5. Load–displacement curves for lap joints with various overlaps in four-point bending.

Fig. 4. The four- and three-point bending test and the corresponding moment diagram. The
shaded area of the moment diagram represents the moment outside the overlap. Table 1
Experimental results of lap joints under four-point bending (6 tests for each case)

constant distance The outside supports were 100 mm apart, with the inside Overlap Average Average central Description of testing
length (mm) peak load loading point
loading points 40 mm apart, which spanned across the overlap with a
(N) deformation (mm)
constant bending moment. For the three-point bend tests, the distance
between the inner loading point, which was always at the middle of the 15 92 10 No failure of lap joint
overlap, and the outer supports (a in Fig. 4) could be varied from 30 to 75 30 115 8 As above
45 2.5 Cohesive failure
mm. The displacement was measured using a linear variable displacement
one hundred

transducer (LVDT). Six samples were tested for each case.

3.Experimental results

3.1. Four-point bending

Typical load vs. displacement plots are shown in Fig. 5 for different
overlaps and 0.1 mm adhesive thickness. The results are summarised in
Table 1. With the 15 mm overlap, the bondline did not fail at all—the
adherend deformed plastically until the maximum load was reached and
then the test was stopped with a displacement of 10 mm. The steel yields at
the two inner loading points, away from the overlap region, and therefore
prevents any significant further increase in load (see Fig. 6). The 30 mm
overlap showed the same response as the 15 mm overlap, but with a higher
peak load and less deformation. Despite a load– displacement curve similar
to that of 10 and 15 overlaps, the 45 mm overlap showed a completely
different behavior since the bondline failed. It started to fail at a small amount
of deformation, but at approximately the same amount of load as the 30 mm Fig. 6. Deformation of lap joint under four- and three-point bending.
overlap specimen. The main reason for this was observed experimentally.
At an overlap length of 45 mm, the loading point is 20 mm on either side of
the center of the overlap. Therefore, the loading point is very close to the bondline thickness of the lap joint under test. If the adhesive layer did not
spew fillet area. As the steel yields, the adhesive is also submitted to large fail, then the load sustained is dependent on the steel itself. If a section of
distortions around the spew area. This large deformation of the steel around steel, with the same dimensions as the lap joint is tested in four-point
the adhesive causes failure, as the adhesive cannot take the large plastic bending, then the maximum load sustained sample is approximately 100 N
deformations. as shown in Fig. 7, which is what would be expected if the maximum load
supported was determined by the yield of the steel. In the outer sections of
the specimen there is only a single thickness of the steel, while the overlap
region has a much thicker cross-sectional area. The bending moment Me to
Different bondline thicknesses were tested with an overlap of 15 mm. initiate yield is
The joints did not fail in all cases. Irregular load–displacement curves were
obtained due to friction and slipping at loading rollers and the inner and
sy sy bt3
outer scatter bounds are shown in Fig. 7. Me ¼ I¼ (one)
The maximum load sustained does not seem to be affected by the r t=2 12
Machine Translated by Google ARTICLE IN PRESS

408 LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413

Table 2
Experimental results of lap joints under three-point bending (6 tests for each case)

Distance between Overlap Average failure Maximum


loading points (a) length (mm) load or peak load bending moment
(mm) (no failure) (N) (at the middle of
the joint (N m)

Failure load (N)


30 7.5 113.8 1.71
15 108.4 1.63
30 211.9 3.18
45 406.7 6.09

45 15 67.2 1.51
60 15 46.2 1.39

Peak load (N)


75 15 34.8 1.31
90 77.6 2.91

Fig. 7. Load–displacement curves for lap joints in four-point bending.

where I is the second moment of area, r the maximum distance


from the neutral axis, b the width and t the thickness. The yield
stress of steel used (sy) is approximately 170 MPa (from Fig. 2),
which gives a bending moment of 0.639 N m. On the other hand,
with the four-point loading jig, the constant bending moment
across the overlap is (see Fig. 4)
P.
M¼a2 (2)

where P is the load and a the distance between the inner loading
point and the outer supports. Therefore, the load is necessary in the
four-point bend tests to initiate yielding of the steel in bending is
P ¼ (Me 2)/a ¼ 43 N, which compares well with the point Fig. 8. Bending moment at the edge of the overlap at failure etc. overlap length for
corresponding to initial non-linearity in Fig. 7. lap joints in three-point bending.

For a comparison with a brittle adhesive, a batch of joints was


produced with adhesive MY750 from Ciba, with an overlap
of 15 mm and a bondline thickness of 0.1 mm. The loading in
bending of a lap joint made up of MY750 and steel is shown in
Fig. 7. In this case, there is catastrophic failure. The average
maximum load of the batch was 60 N and there was very little
plastic deformation before the joint failed. Since the adhesive fails
at about 1% strain, failure of the adhesive occurred before the
adherends became fully plastic.

3.2. Three-point bending

If the bondline thickness is kept at 0.1 mm, and the joint width
is 25 mm, the effect of increasing the overlap is to increase the
maximum load that can be taken by the joint before the crack
initiates, which is shown in Table 2. Since the bending moment in
the three-point bending test varies linearly from a maximum at
the mid-point of the overlap to zero at the outside supports Fig. 9. Load–displacement for lap joints with various bondline thicknesses with
(see Fig. 4), an increase in the overlap length would effectively 451 fillet under three-point bend loading.
reduce the bending moment at the edge of the overlap where
failure initiates. The moment at the edge of the overlap (Medge)
can be expressed in terms of the maximum moment (Mmax) at the a function of the overlap length (see Fig. 8). This result shows that
center of the joint, the distance between the inner and outer the bending moment is constant at the edge of the overlap when
supports (a) and the overlap (l): Failure occurs in the joint.
l one
Fig. 9 shows that for the range of bondline thicknesses used,
Medge ¼ Mmax a (3) when the crack initiates at the edge of the overlap, the load drops
2 a
appreciably, corresponding to crack propagation. Fig. 9 also shows
If the experimental failure load is used to calculate the that the maximum load sustained does not seem to be affected
maximum bending moment according to Eq. (2) (see Table 2), by the bondline thickness, the different values being explained by
the moment at the edge of the overlap computed with Eq. (3) experimental scatter. There is also a slight increase in the stiffness
gives an approximately constant value of 1.5 N m when plotted as of the joint as it is bent, for increasing adhesive thickness creates a
Machine Translated by Google ARTICLE IN PRESS

LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413 409

stiffer overlap section (in bending). The bending moment at the edge of where M ¼ kPt/2, according to Goland and Reissner [8]. The variable k
the overlap region is determined by the load that is carried by the joint is the bending moment factor that reduces (from unity) as the lap rotates
together with the joint geometry. The load at the center of the overlap under the load. The stress acting in the adherend is the sum of the direct
region causes the overlap to deform (see Fig. 6). If the center section is stress and the bending stress. Thus, the maximum load that can be
stiffer, because it is thicker, then the deformation is less. Increasing the carried which just creates adherend yield (PY) is
bondline thickness causes the joint to be stiffer, but the plastic bending
moment in the steel is still the same at the edge of the overlap.
PY ¼ sybt=ð1 þ 3kÞ (7)

where sy is the yield strength of the adherend. For low loads and short
3.3. Tension loading overlaps, k is approximately 1. Therefore, for such a case

PY ¼ sybt=4 (8)
The tests were carried out with an overlap of 15 mm. End pieces
were used to ensure the joint was aligned at the start of each test, However, for joints that are long compared to the adherend thickness,
particularly because of the large bondline thicknesses that was proposed which is the case here (l ¼ 15 and t ¼ 0.95 mm), the value of k decreases
and it is assumed here that it tends to zero. In this case, the whole of the
to be used. The strength of the lap joint decreases with increasing
bondline thickness as shown in Fig. 10. cross-section yields and
From Fig. 10, it can be seen that the 451 spew fillet creates stronger PY ¼ sybt (9)
joints, especially as the bondline thickness increases. At the thinnest
bondline of 0.1 mm, there is very little difference in the strengths of the Eq. (9) gives a failure load of 4.5 kN. This criterion has been applied
joints but, at a bondline thickness of 3 mm, the strength of the square- before with successful results to thin bondlines and a conservative value
ended joint is less than half of that with a 451 spew fillet. The deterioration has always been obtained [4]. In the present case, the prediction is also
in strength as the bondline thickness increases can be seen clearly. The conservative for thin bondlines (less than 1 mm) but overestimates the
strength of the square-ended joint, with a bondline thickness of 0.1 mm, results of thicker bondlines (more than 1 mm). This could be explained
is approximately 4 times the strength of a joint made with a bondline of by the fact that the model does not take into account strain hardening of
3 mm. the steel. For a more accurate solution, the bending moment factor k
could be computed according to Hart-Smith's [9] equation that takes into
Adams et al. [4] proposed a simple methodology for designing single- account the adhesive thickness. Note, however, that Eq. (7) would now
require a numerical solution because k also depends on the load applied
lap joints under tension. The upper limit is given by the load corresponding
to the total plastic deformation of the adhesive (global yielding), which is P. Moreover, Hart-Smith's bending moment factor is inaccurate when
the adherend deforms plastically, as shown in Section 4.3.

PGY ¼ (4)
tybl where PGY is the failure load of the adhesive due to global yielding,
ty the yield strength of the adhesive, b the joint width and l the overlap
length. Applying this equation to the present case and supposing a von 4. Joint modeling
Mises yield model for the adhesive (ty ¼ sy/O3), PGY gives an
approximate value of 18 kN. This value is well above the experimental The joints have all been analyzed using the FE program ABAQUS to
failure load and proves that the adhesive was far from being plastic include the variety of bondline thicknesses from 0.1 to 3 mm, in both
along the whole overlap. A lower limit is given when the adherends yield three-point bending and tension. Each joint was modeled using a pre-
plastically. The direct tensile stress (st) acting in the adherend due to the processor, PATRAN, with a general mesh refined to examine the
applied load P is adhesive layer. Nonlinear geometry and material were included in the
analysis. A 2D plane strain model was used with 8-noded isoparametric
st ¼ P=bt (5) elements.
where t is the adherend thickness. The stress at the inner adherend The objective of the FE analysis was to investigate the magnitude
surface (ss) due to the bending moment M is and direction of the stresses and strains in the different geometries in
order to understand failure. However, the FE results are extremely
ss ¼ 6M=bt2 (6) sensitive to mesh refinement, especially with respect to singularities at
the ends of the joint.
Any global comparison of stresses must be done on similar meshes.
This was difficult since the change in both bondline thickness and joint
geometry meant that the mesh density could never be the same. It could
be argued that variations across the bondline thickness could be modeled
with the same number of elements. However, this did not work well since
one element was in one case covering 1.5 mm of adhesive, but in
another geometry only 0.005 mm of adhesive. This introduced problems
associated with the aspect ratio of the elements. Therefore, a different
approach was taken. Looking at the joints, there was an area under
investigation where initial failure always occurs. Therefore, if this area
were to be modeled consistently, then a direct stress comparison can be
achieved. In the case of the lap joints, the mesh refinement was used as
shown in Fig. 11. The mesh was consistently the same around the
singularity at the edge of the spew fillet, regardless of the size of the
bondline thickness.

The model was progressively built up from the 0.1 mm bondline model
Fig. 10. Failure load vs. bondline thickness of lap joints under tension. until a 3 mm bondline was achieved, by adding extra
Machine Translated by Google ARTICLE IN PRESS

410 LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413

Fig. 11. Mesh refinement used in lap joint analysis. Fig. 12. Maximum principal stress in the adhesive, one element from the loaded
adherend interface, for lap joints in tension with a spew fillet.

elements. If a small corner of the model is viewed just around the


singularity, then the mesh would be identical in all different bondline
thicknesses. The area of interest was the edge of the overlap region,
be it the edge of the spew fillet or with the square-ended geometry.
The mesh density and the surrounding geometry is the same in
each case for both the spew fillet and the square end. This meant
that the stresses and strains could be compared at the node points
a set distance away from the singularity, which influences their
behavior. To compare the stress and strain results between the
models, the same position in the model was used. For the results,
the values are plotted one element away from the interface region,
at the node points.

Fig. 13. Maximum principal stress in the adhesive, one element from the adherend
4.1. stress distribution interface, for lap joints with square ends under tension.

For lap joints under tension with a 451 spew fillet at their
(experimental) failure load, a graph was plotted of the maximum
principal stresses along the adhesive layer, one element away from
the adhesive/adherend interface as shown in Fig . 12. The range of
failure loads is from 5.7 kN for the 0.1 mm bondline to 3.2 kN for the
3.0 mm bondline specimen. However, the highest maximum principal
stress at failure is approximately 70–75 MPa for each joint.

A comparison of the bondline thickness for lap joints in tension


with square ends is shown in Fig. 13. Here, the maximum principal
stress distribution in the adhesive layer is different, but the results
are surprisingly similar. Near the singularity at the edge of the
overlap, the stress values are approximately the same as those for
the case of the spew fillet.

Fig. 14. Maximum principal strain in the adhesive, one element from the adherend
4.2. Strain distribution interface, for lap joints in tension with a spew fillet.

An important aspect of this analysis has been to assume that the before decreasing, and these peak values are very similar to those
bending moment at the edge of the overlap region is initiating failure of the joints with a fillet, and again are about 3%.
as the steel becomes plastic and, therefore, creates large strains in
the adhesive. These strains are too much for the adhesive to stand
and failure therefore occurs. The maximum principal strains in the 4.3. Critical bending moment for joint failure
adhesive at failure for the spew fillet (see Fig. 14) indicate a failure
at approximately 3% strain. This is consistent with the strain to Lap joints under tension were investigated to see the effect of
failure found in the bulk adhesive changing the bondline thickness on the bending moment at the edge
specimens (Fig. 1). The results of the square-ended lap joints are of the overlap. The bending moment was determined using numerical
shown in Fig. 15. Although these peak at very high strains, this is integration from the stresses available at the node points from the
due to the effect of the mathematical singularity found at the very FE results. As the bondline thickness was increased, the bending
sharp corner in the square-ended model. The fillet reduces the effect moment at the end of the overlap was calculated when a load of 3
of the singularity. However, looking back from the edge of the kN was applied. This allowed the determination of the effect of the
overlap, where the singularity is located, the strains peak bending moment caused by the
Machine Translated by Google ARTICLE IN PRESS

LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413 411

Fig. 15. Maximum principal strain in the adhesive, one element from the adherend
interface, for lap joints in tension with a square end.

Fig. 18. Longitudinal stress distribution across adherend section, including plasticity
effects, for lap joints under three-point bending.

Fig. 16. Bending moment (calculated from finite element analysis) at the edge of overlap
etc. bondline thickness of joint for a constant load tensile load of 3 kN.
Fig. 19. Longitudinal stress distribution across the adherend section, including plasticity
effects, for lap joints under tension.

and the 451 spew fillet lap joints in tension (see Fig. 17) shows that
the decrease in the failure load corresponds clearly to an increase
in the bending moment.
The effect of only a bending moment imposed across the lap
joint on the longitudinal stresses in the joint is shown in Fig. 18. As
the load imposed on the joint increases, the bending moment
imposed at the edge of the overlap increases, which in turn
increases the stress at the edge of the adhesive. When the stress
reaches the yield point of the steel, large plastic strains result.
Although the stresses are limited, the strains associated with the
plastic stress are not, as shown in Fig. 18. As the maximum
adhesive strain is limited (see Fig. 1), it therefore fails when the
maximum adhesive strain is exceeded.
For lap joints under tension, there is a longitudinal stress from
Fig. 17. Failure load vs. bending moment (calculated from finite element analysis) at
the direct load together with an additional bending stress due to the
failure for lap joints under tension.
load offset that is superimposed on the tension stress, as shown in
Fig. 19. To reach the same stress level, as the bending moment
increases, the smaller the stress due to direct load has to be. As
offset loading created by the bondline itself. As the bondline the bondline thickness increases, there is an increase in the bending
thickness increases, the offset increases, resulting in an increase in stress due to the load offset. As a result, the direct load, that is, the
the bending moment, as shown in Fig. 16. There is also a difference strength of the joint, is reduced.
in the bending moment at the edge of the overlap between the joint Most analytical methods do not include any effect of the thickness
with the spew fillet and that with the square edge at a fixed load of of the bondline. Therefore, in order to check their validity, a series
3 kN. This explains the difference in joint strength between those of elastic analyzes was conducted on lap joints without a spew fillet.
joints with a spew fillet, and those without, as shown in Fig. 10. A The bending moment at the ends of the overlap from the FE analysis
plot of the failure load as a function of the corresponding bending was compared with the bending moment obtained using the bending
moment determined with FE for both the square-ended moment factor k of both
Machine Translated by Google ARTICLE IN PRESS

412 LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413

Table 3
Theoretical bending moment for a single-lap joint under a tensile load of 3 kN

Bondline thickness (mm) Elastic FE Plastic FE Hart-Smith Zhao


(N m) (N m) (N m) (N m)

0.1 0.934 0.890 0.925 0.872


3.0 3.309 1.460 3.481 0.872

Fig. 21. Failure envelope for lap joints produced from sheet steel and ESP110.

the bending moment seems to have taken over the behavior of the lap joint,
whereas with a bondline thickness of 0.1 mm, the bending moment is
insignificant compared with the effect of the direct tensile stress.

5. Failure prediction

The main reason why the four-point bend test does not cause failure in
Fig. 20. Bending moment at failure for lap joints under tension and three-point bending.
the lap joint is that the steel yields at the two inner loading points, away
from the overlap region, and therefore prevents any further increases in
load (see Fig. 6 ) .
Three-point bending and tension loadings are very similar in the way in
Zhao [30] which they affect the adhesive. Both induce a large bending moment across
the overlap, and both cause failure.
1k (10) Fig. 21 shows the failure envelope and it can be seen that there is a
¼1þ
clear relationship between the bending moment at the edge of the overlap
xc and Hart-Smith [9] at failure, and the tensile failure load. This forms an envelope of failure,
one
where any combination of load and bending moment outside the envelope
ta k ¼ 1 þ (11th) means that failure occurs, and any combination inside means that the joint
t 1 þ xc þ ð1=6ÞðxcÞ
2
will not fail. The limits are defined by the relationship between the bending
for a constant load of 3 kN. In Eqs. (10) and (11), c is half the overlap length, moment and failure load that can be deduced from the adhesive/adherend
ta is the adhesive thickness, D is the adherend bending stiffness and x2 ¼ combination, and the maximum bending moment that can be sustained by
P¯/D (P¯ is the applied tensile load per unit width of the joint). The results the adherend. Further studies [31,32] show that this failure envelope is also
are presented in Table 3. For thin bondlines, both closed-form methods are applicable to other types of joints (T joints) and environments (temperature).
quite accurate in calculating the bending moment but, for the 3 mm thick
bondline, only Hart-Smith's bending moment is a good approximation to the
imposed bending moment found across the overlap as Zhao does not take
into account the bondline thickness. However, the steel adherends do
exhibit a large amount of plasticity, so a plastic analysis was completed and 6. Conclusions
compared to the elastic analysis in order to see if that affects the bending
moment (see Table 3). The bending moment in the joint with a 3 mm Lap joints typical of those used in the automotive industry were studied
bondline thickness was much lower than the elastic analysis as the plasticity under tension, three-point bending and four-point bending. Various geometric
of the steel reduces the effect of the eccentric load path. Therefore, even parameters were studied such as the overlap length, the adhesive thickness
the closed-form method (Zhao or Hart-Smith) is not good enough for a and the distance between loading points in the case of the bending tests.
realistic approximation of the bending moment. Also, from the experimental The main conclusions are as follows:
results it was obvious that the fillet, especially in the thicker bondline joints,
had a large effect on the strength of the joint. It is therefore necessary also
to model the joints with a 451 spew fillet. 1. There was extensive adherence yielding in all cases.
2. In the four-point bending test, the steel yields at the loading points, away
from the overlap region, and therefore prevents any further increases in
load without causing failure of the joint.
The bending moment results for all bondline thickness with and without
fillets on lap joints in three-point bending and tensile lap shear are shown in 3. The bending moment is constant at the edge of the overlap when failure
Fig. 20. As the bondline thickness increases, the bending moment induced occurs in the three-point bend test and corresponds to the maximum
is larger, and the bending stress is higher. With a bondline thickness of 3 bending moment the steel can sustain. This bending moment creates
mm, the effect of large plastic strains as
Machine Translated by Google ARTICLE IN PRESS

LDR Grant et al. / International Journal of Adhesion & Adhesives 29 (2009) 405–413 413

the steel becomes plastic and this initiates failure in the adhesive. [3] Vicenti W.G. What engineers know and how they know it: analytical studies from
aeronautical history. Baltimore and London: Johns Hopkins University Press; 1990.

4. The joint strength in the three-point bend tests is independent of [4] Adams RD, Comyn J, Wake WC. Structural adhesive joints in engineering. 2nd ed.
the adhesive thickness. An increase in adhesive thickness London: Chapman & Hall; 1997.
causes the joint overlap section to be stiffer, but the bending [5] Kinloch AJ. Adhesion and adhesives: science and technology. London: Chapman &
Hall; 1987.
moment is still the same at the edge of the overlap. [6] Adams RD, editor. Adhesive bonding: science, technology and applications.
5. For a lap joint under tension, the longitudinal stress from the Cambridge: Woodhead Publishing Limited; 2005.
direct load and the bending moment at the edge of the overlap [7] Volkersen O. Luftfahrtforschung 1938;15:41.
[8] Goland M, Reissner E. J Appl Mech 1944;66:A17.
region create plastic strains when the steel becomes plastic and [9] Hart-Smith LJ. NASA Contract Report, NASA CR-112236.
these cause failure in the adhesive. [10] Bigwood DA, Crocombe AD. Int J Adhes Adhes 1990;10:31.
6. The lap joint under tension is very sensitive to adhesive thickness. [11] Allman DJ. Q J Mech Appl Math 1977;30:415.
[12] Adams RD, Mallick V. J Adhes 1992;38:199.
To reach the same stress level, and plasticity of the steel
[13] Renton J, Vinson JR. J Adhes 1975;7:175.
adherend, as the bending moment increases, the smaller the [14] Wah T. ASME J Eng Mater Technol 1973;95:174.
stress due to tension has to be. As the bondline thickness [15] Adams RD, Atkins RW, Harris JA, Kinloch AJ. J Adhes 1986;20:29.
increases, there is an increase in the bending stress since the [16] Yang C, Pang SS. ASME J Eng Mater Technol 1996;118:247.
[17] Mortensen F, Thomsen OT. Compos Sci Technol 2002;62:1011.
bending moment has increased. As a result the strength of the [18] Crocombe AD, Bigwood DA. J Strain Anal Eng Des 1992;27:211.
joint is reduced. [19] Frostig Y, Thomsen OT, Mortensen F. J Eng Mech 1999;125:1298.
7. A 451 fillet creates stronger joints in tension than joints with a [20] Belingardi G, Goglio L, Tarditi A. Int J Adhes Adhes 2002;22:273.
[21] Andreassi L, Baudille R, Biancolini ME. Int J Adhes Adhes 2007;27:458.
square end, especially as the bondline thickness increases. At [22] Fessel G, Broughton JG, Fellows NA, Durodola JF, Hutchinson AR. Int J Adhes
the thinnest bondline of 0.1 mm, there is very little difference in Adhes 2007;27:574.
the strengths of the joints with and without a fillet but, at a [23] da Silva LFM, Adams RD. Int J Adhes Adhes 2002;22:311.
[24] Gleich DM, van Tooren MJL, Beukers A. J Adhes Sci Technol 2001;15:1091. [25]
bondline thickness of 3 mm, the strength of the square-ended da Silva LFM, Rodrigues TNSS, Figueiredo MAV, de Moura MFSF, Chousal JAG.
joint is less than half of that with a 451 fillets. J Adhes 2006;82:1091.
8. A failure criterion has been proposed based on the tensile load [26] Liu J, Sawa T, Toratani H. J Adhes 1999;69:263.
[27] Ozel A, Kadioglu K, Sen S, Sadeler R. J Adhes 2003;79:683.
and bending moment applied to the joint. [28] Ozel A, Aydin MD, Temiz S. J Adhes Sci Technol 2004;18(3):313.
[29] Grant LDR. The characterization of adhesive joints found typically in the automotive
industry. Ph.D. Dissertation, University of Bristol, 1994.
[30] Zhao X. Stress and failure analysis of adhesively bonded lap joints, Ph.D.
Dissertation, University of Bristol, 1991.
[31] Grant LDR, Adams RD, da Silva LFM. Experimental and numerical analysis of T-peel
references joints for the automotive industry. J Adhes Sci Technol 2008 [accepted for publication].

[32] Grant LDR, Adams RD, da Silva LFM. Effect of the temperature on the strength of
[1] Hong SG, Boerio FJ. J Adhes 1990;32:67. adhesively-bonded single lap and T joints for the automotive industry. Int J Adhes
[2] Moody IN, Fay PA, Suthurst GD. Sheet Met Ind 1987;64(7):332. Adhes 2008 [submitted for publication].

You might also like