Flexural behaviour of glue-laminated bre composite sandwich beams

A.C. Manalo, T. Aravinthan

*
, W. Karunasena
Centre of Excellence in Engineered Fibre Composites (CEEFC), University of Southern Queensland, Toowoomba 4350, Australia
a r t i c l e i n f o
Article history:
Available online 11 March 2010
Keywords:
Composite sandwich beams
Glue-laminated beams
Fibre composites
Flexure
Flatwise
Edgewise
a b s t r a c t
This study involved experimental investigation onto the exural behaviour of glue-laminated bre com-
posite sandwich beams with a view of using this material for structural beams. Composite sandwich
beams with 1, 2, 3, and 4 composite sandwich panels glued together were subjected to 4-point static
bending test in the atwise and edgewise positions to evaluate their stiffness and strength properties.
The results showed that the composite sandwich beams in the edgewise position failed with 25% higher
bending strength but have 7% lower bending stiffness than beams in the atwise position. The results
however indicated that the bending stiffness of atwise specimens converges to that of the edgewise
specimens with increasing laminations. More importantly, the specimens in the edgewise position failed
with greater ductility due to progressive failure of the bre composite skins while the specimens in the
atwise position failed in a brittle manner due to debonding between the skin and core. Wrapping the
glue-laminated sandwich beams with one layer of tri-axial glass bres did not prove to be effective. Over-
all, it has been demonstrated that the glue-laminated sandwich beams exhibited better performance than
the individual composite sandwich beams.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
Composite sandwich structures fabricated by attaching two
thin but stiff skins to the lightweight but thick core have been
widely used in the automotive, aerospace, marine and other indus-
trial applications. This material has also been identied as a very
interesting alternative to traditional construction materials be-
cause of its high bending stiffness and high strength to weight ra-
tios [1]. The exibility of composite sandwich construction allows
innovative structural developments from this material. This com-
posite material can also be combined with traditional construction
materials or be shaped and formed to carry loads that cannot be
carried by the individual sandwich structure. In addition, sandwich
structure can be designed with the desired stiffness and strength
with no additional weight to suit various structural applications.
Recent applications have demonstrated that composite sand-
wich construction can be effectively and economically used in
the civil infrastructure. Composite sandwich materials are now
commonly used as structural panel for roofs, oors, walls and
bridge decks [2]. However, very limited attempt has been made
so far to use these materials for structural beam applications
although engineers have a wide range of composite sandwich pan-
els. The main reason could be that most of the currently used core
materials are not appropriate for this type of structural application.
The commonly used foam core and balsa wood are soft and will
crush under high compressive loads [3,4]. Honeycomb and
trussed-core structure have high compressive strength [5,6] but
the presence of cavities in these core materials reduces their capac-
ity to hold mechanical connections. The evolution of a composite
sandwich structure with lightweight, high strength core material
and with good holding capacity for mechanical connections pro-
vides an opportunity to develop this material for structural beam
applications.
A new generation bre composite sandwich panel made up of
glass bre-reinforced polymer skins and modied phenolic core
material has now been developed in Australia [7]. The exural
behaviour of this innovative composite sandwich material was
investigated by Manalo et al. [8]. The results of their study sug-
gested that the strength and stiffness of this innovative composite
sandwich structure are suitable for structural beam applications.
As these composite sandwich panels are produced in limited thick-
nesses, a structural beam section could be produced by gluing a
number of sandwich panels together either in the atwise (hori-
zontal) or edgewise (vertical) positions. This concept is similar to
glue-laminated (glulam) lumber used in timber engineering where
several smaller pieces of wood are horizontally or vertically lami-
nated (either by nailing or gluing) to produce a single large, struc-
tural member to support a greater load [9]. Similarly, several
bridge decks have been constructed by nailing together timber
placed in the edgewise position [10]. Moreover, Lopez-Anido and
Xu [11] developed a structural system based on the concept of
sandwich construction with strong and stiff FRP composite skins
0263-8223/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2010.03.006
* Corresponding author. Tel.: +61 7 4631 1385; fax: +61 7 4631 2110.
E-mail addresses: manalo@usq.edu.au (A.C. Manalo), aravinthant@usq.edu.au
(T. Aravinthan), karunasa@usq.edu.au (W. Karunasena).
Composite Structures 92 (2010) 27032711
Contents lists available at ScienceDirect
Composite Structures
j our nal homepage: www. el sevi er . com/ l ocat e/ compst r uct
bonded to an inner glulam panel. The glulam panels were fabri-
cated with bonded eastern hemlock laminations. In addition, most
currently available commercial FRP decks are constructed using
assemblies of adhesively bonded bre composite pultruded shapes
[2,12,13]. These examples show that the concept of gluing a num-
ber of composite sandwich panels to form a structural beam is
highly practical.
This paper presents the results of an experimental investigation
on the behaviour of glue-laminated composite sandwich beams in
the atwise and edgewise positions to determine their application
for structural beams. A number of composite sandwich beams
were glued together and subjected to static 4-point bending test
to determine their exural behaviour. The loaddeection behav-
iour, strength and failure mechanisms of the glue-laminated com-
posite sandwich beams are reported. The effects of the number and
the orientation of laminations as well as the glass bre wrapping
on the strength and stiffness of glue-laminated composite sand-
wich beams are also discussed.
2. Experimental program
2.1. Material properties
The structural composite sandwich beams tested in this study
are made up of glass bre composite skins co-cured onto the mod-
ied phenolic core material using a toughened phenol formalde-
hyde resin [7]. The bre composite skin is made up of two plies
of stitched bi-axial (0/90) E-CR glass bre fabrics manufactured
by Fiberex Corporation. The 0 glass bre weighs 400 g/m
2
and
has a thickness of 0.512 mm while the 90 glass bre weighs
300 g/m
2
and has a thickness of 0.384 mm. The modied phenolic
foam core is a proprietary formulation by LOC Composites Pty. Ltd.,
Australia. The composite sandwich panel has a nominal thickness
of 20.0 mm and has an overall density of approximately 1000 kg/
m
3
. The mechanical properties of the bre composite skin and
the modied phenolic core material of this innovative composite
sandwich panel were determined in earlier study by Manalo
et al. [8] and are listed in Table 1.
2.2. Test specimen and preparation
Fig. 1 shows the preparation of glue-laminated composite sand-
wich beam specimens. The specimens for the characterisation of
individual sandwich beam behaviour were cut directly from the
composite sandwich panels provided by the manufacturer. A num-
ber of composite sandwich panels were assembled and glued to-
gether in 2, 3 and 4 layers using Techniglue-HP R5 structural
epoxy resin supplied by ATL Composites (Fig. 1a). The glued sand-
wich panels were then clamped for 24 h to initially cure the epoxy
and were removed from clamping to post-cure at 90 C for 8 h to
attain good bonding between the composite sandwich laminations.
After curing, the glue-laminated sandwich panels were cut into the
required specimen width (Fig. 1b). The glue-laminated composite
sandwich beams with 3 and 4 laminations were prepared with
similar width and depth to eliminate the geometrical effects. The
descriptions of the test specimens are listed in Table 2.
The last four test conguration shown in Table 2 were prepared
by wrapping the specimens with one layer of 750 g/m
2
tri-axial
glass bre composite (0/+45/45). In the preparation of these
specimens, the corners of the glue-laminated composite sandwich
beams were rounded to a 15 mm radius in order to bend and wrap
the bres without causing damage. The bre wraps were provided
through hand lay-up process in two different stages covering the
top and bottom surfaces with one layer of tri-axial glass bres
and the sides with two layers of bres. Hyrex 201 epoxy resin
[14] was used to impregnate and bond the bre wraps to the
glue-laminated composite sandwich beam specimens. The mix ra-
tio, by weight, of the epoxy and hardener was 100:20. After apply-
ing the bre wraps, the specimen were pre-cured for 24 h at
ambient temperature and were post-cured at 90 C for 8 h before
they are tested.
2.3. Test set-up and procedure
The 4-point static bending test on composite sandwich beams
was performed in accordance with the ASTM C393-00 standard
[15]. The load was applied at 0.4 and at 0.6 of the span through a
100 kN universal testing machine with a loading rate of 3 mm/
min. The actual test set-up and instrumentation for the static ex-
ural test of composite sandwiches are shown in Fig. 2. Before each
test, the loading pins were set to almost touch the top surface of
the composite sandwich specimen. The applied load, displacement
and strains were recorded using a data logger. The failure mecha-
nisms were also monitored and recorded.
3. Experimental results
3.1. Loaddeection behaviour of composite sandwich beams
The load and midspan deection behaviour of individual com-
posite sandwiches under 4-point static bending is shown in
Fig. 3. The gure shows that the deection of specimen 1LSW-F in-
creased almost linearly with load up to nal failure. The specimen
failed at an applied load of 4550 N with a midspan deection of
24 mm. The load of specimen 1LSW-E increased linearly with
deection but showed a reduction in stiffness at a load of around
5000 N due to tensile cracking of the core. The specimen then con-
tinued to carry load until 5500 N. Before the nal failure, there was
an increase in deection even without an increase in the applied
load due to progressive failure of the non-horizontal skins. The g-
ure also shows that the load of specimen 2LSW-F increased linearly
with deection. This linear behaviour was observed until a load of
6000 N and a deection of 7 mm. After this load, a non-linear re-
sponse was observed until failure. When the tensile cracks oc-
curred in specimen 2LSW-E at an applied load of 11,000 N, a
slight drop in the load was observed. As the loading continued,
there is a gradual decrease in bending stiffness due to the progres-
sive failure of the bre composite skins. After compressive failure
of the skins, the beams were still able to carry load but showed
large deection until failure.
A similar load deection behaviour was observed in specimens
with 3 and 4 laminations (Fig. 4). The load deection curves for
specimen 3LSW-F and 4LSW-F are almost linear until the develop-
ment of exural tensile cracks in the core material. A decrease in
stiffness was then observed until failure of the specimen. A
decrease in stiffness was also observed in specimen 3LSW-E and
4LSW-E when tensile cracks of the core developed. As the loading
continues, there is a gradual decrease in the bending stiffness due
to progressive failure of the bre composite skins. Specimen
Table 1
Effective mechanical properties of bre composite skin and core material.
Property Skin Core
Modulus of elasticity (MPa) 14,280 1350
Maximum tensile stress (MPa) 246 4.25
Maximum tensile strain (%) 1.60 0.57
Maximum compressive stress (MPa) 201 21.35
Maximum compressive strain (%) 1.24 1.94
Thickness (mm) 1.8 16.4
2704 A.C. Manalo et al. / Composite Structures 92 (2010) 27032711
Fig. 1. Preparation of glue-laminated composite sandwich beams.
Table 2
Description of specimen for exural test of composite sandwich beams.
Specimen Illustration Number of specimens D (mm) B (mm) Length, L
T
(mm) Support span (L) Orientation of testing
1LSW-F
D
B
5 20 50 500 400 Flatwise
1LSW-E
D
B
5 50 20 500 400 Edgewise
2LSW-F
D
B
2 40 50 500 400 Flatwise
2LSW-E
D
B
2 50 40 500 400 Edgewise
3LSW-F
B
D
2 60 60 1400 1200 Flatwise
3LSW-E
D
B
2 60 60 1400 1200 Edgewise
4LSW-F
D
B
2 80 80 1400 1200 Flatwise
4LSW-E
D
B
2 80 80 1400 1200 Edgewise
3LSW-WF
B
D
1 60 60 1400 1200 Flatwise
3LSW-WE
D
B
1 60 60 1400 1200 Edgewise
4LSW-WF
D
B
1 80 80 1400 1200 Flatwise
4LSW-WE
D
B
1 80 80 1400 1200 Edgewise
A.C. Manalo et al. / Composite Structures 92 (2010) 27032711 2705
3LSW-E and 4LSW-E continued to carry load even after compres-
sive failure of the outer bre composite skins. The specimen then
behaved non-linearly with a reduced stiffness up to failure.
In both composite sandwich beams with 3 and 4 laminations,
the specimens in the atwise positions behaved slightly stiffer
than specimens in the edgewise position. However, the composite
sandwich beams in the edgewise position failed at a higher load
than the specimens in the atwise position. Finally, the load
deection curve indicated that the composite sandwich beams
tested in the atwise position failed in a brittle manner while the
composite sandwiches in the edgewise beams failed in a ductile
exural mode. This could be due to the difference in the failure
mode which is discussed in the next section.
3.2. Failure behaviour of composite sandwich beams
Experimental investigation showed that the bre composite
sandwich beams exhibited different failure behaviours in the at-
wise and in the edgewise positions. These different failure modes
of the composite sandwich beams are shown in Fig. 5. Tensile
cracks of the core were observed at the bottom of specimen
1LSW-F but these cracks did not cause immediate failure. The bre
composite skins on the tensile face of the specimen bridged the
cracked core together to prevent failure. The specimen 1LSW-F
failed due debonding between the core and the skins followed by
compressive failure of the bre composite skins as shown in
Fig. 5a. Tensile cracks of the core were also observed in specimen
1LSW-E. However, the presence of the non-horizontal skins in
the edgewise position prevented the premature failure and made
the composite sandwich beams fail in a ductile failure mode. The
specimen 1LSW-E failed due to progressive compressive failure
of the bre composite skin followed by tensile failure of the skin
(Fig. 5b).
Flexural cracks were observed on the core of the bottommost
sandwich layers for specimens with 2, 3 and 4 laminations. These
cracks originated at the top of the tensile skin and progressed with
the application of load. The bre composite skins, however,
bridged the cracked core together to prevent the immediate failure
of the specimen. When the depth of the exural cracks on the core
reached the level of the next bre composite skin, the crack width
increased and a signicant drop in the load was observed. The
presence of the bre composite skins, however, prevented the
extension of the cracking of the core to the core of the next com-
posite sandwich laminations. The glue-laminated sandwich beams
in the atwise positions failed due to compressive buckling of the
bre composite skins followed by the debonding between the bot-
tom skin and the core material as shown in Fig. 5c and e. Tensile
cracks of the core were also observed in the edgewise specimen
at the early application of load. The non-horizontal skins, however,
prevented the tensile cracks in the core from widening to cause
immediate failure. As more cracks developed on the core, the dam-
age on the specimen increased, thereby, decreasing the stiffness,
and subsequently increasing the deection. The specimen contin-
ued to carry load even after compressive failure of the outermost
bre composite skin as shown by the small cracks which devel-
oped near the loading point. The continuous application of load
caused the outermost compressive skins to delaminate from the
core material and caused the cracks to propagate horizontally at
the region of constant moment. This failure resulted in decrease
in lateral stability and eventually compression buckling of the de-
tached bre composite skins. Splitting of the tensile bre compos-
Fig. 2. Test set-up and instrumentation of exural test for composite sandwich beams.
Fig. 3. Load and midspan deection relationship of specimen 1LSW and 2LSW.
Fig. 4. Load and midspan deection relationship of specimens 3LSW and 4LSW.
2706 A.C. Manalo et al. / Composite Structures 92 (2010) 27032711
ite skins were also then observed. Final failure of the glue-lami-
nated sandwich beams in the edgewise positions was due to simul-
taneous compressive failure of the bre composite skins and
crushing of the core material followed by tensile failure of the skins
(Fig. 5d and f). These different failure modes have affected the
strength capacity of the composite sandwich beams. Furthermore,
the results of the experimental investigation showed neither
delamination nor slipping occurred on the glue lines. These results
suggest that the structural epoxy adhesives used in this study has
provided a highly efcient glue joint between the composite sand-
wich laminations and the full capacity of the glue-laminated com-
posite sandwich beams was attained.
3.3. Load and longitudinal strain relationship
The load and longitudinal strain relationship of specimens
3LSW and 4LSW are shown in Fig. 6. In these gures, the longitu-
dinal tensile strain is designated with (T) and the longitudinal
compressive strain with (C). The results showed that the strains
in both tension and compression increased linearly with load at
the early stage of load application for all the tested specimens.
However, a stiffer loadstrain relation curve can be noticed for
specimen 3LSW-F than 3LSW-E but became almost similar for
specimen 4LSW.
At a tensile strain of around 6000 micro strains, the strain gauge
at the tension side of the specimen broke, indicating the develop-
ment of cracks in the core. It is noteworthy that this level of strain
is comparable with the failure strain of the core in tension estab-
lished from the test of coupons (see Table 1). The bre composite
skins prevented the sudden failure of the composite sandwich
beams. Non-linearity in the compressive strain was then observed
indicating the further development of cracks in the core material.
The strain gauge on top of the specimen broke at a compressive
strain of around 12,500 microstrains which indicated the compres-
sive failure of the bre composite skins. Again, this value of strain
represents the strain at which the bre composite skins failed in
Fig. 5. Failure of composite sandwich beams.
A.C. Manalo et al. / Composite Structures 92 (2010) 27032711 2707
compression determined from the coupon tests. In this level of
strain, the specimen tested in the atwise position failed instantly
while the specimen in the edgewise position continued to carry
load and failed at a higher load than expected. Thus, the maximum
level of strain at which the specimen in the edgewise position
failed was not captured by the strain gauges. Final failure of the
edgewise specimen occurred only when the core crushed in com-
pression followed by buckling of the inner bre composite skins.
4. Discussion
4.1. Effect of glue laminations on the stiffness of composite sandwich
beams
Evaluation of the effect of gluing on the stiffness of the compos-
ite sandwich beams was conducted. The exural stiffness, EI of the
glue-laminated sandwich beams was calculated using the elastic
properties of the bre composite skins and core material in Table
1 and simple sandwich beam theory. Calculations were made
assuming that no interlayer slips occurred and the laminated sand-
wiches acted as a solid section with perfect bonding. The contribu-
tion of the epoxy adhesives in the exural stiffness is also
neglected. The exural stiffness in the atwise position was esti-
mated using Eq. (1) and in the edgewise position using Eq. (2).
EI
flat

n
i1
Bt
3
s
12
Bt
s
d
2
s
_ _
E
s

Bt
3
c
12
Bt
c
d
2
c
_ _
E
c
_ _
1
EI
edge

nD
3
6
t
s
E
s

t
c
2
E
c
_ _
2
where B is the width of the sandwich beam, t
s
is the thickness of the
skin, t
c
is the thickness of the core, d
s
and d
c
are the distances from
the centre of the skins and the core to the neutral axis of the glued
section, respectively, D is the depth or thickness of the sandwich
beam while E
s
and E
c
are the modulus of elasticity of the skin and
core, respectively, and n is the number of glue-laminated composite
sandwiches.
Eq. (3) was obtained based on the deection formula of a uni-
form, static composite sandwich beam with loading conguration
shown in Fig. 2a using classical beam theory. The effective bending
stiffness, (EI)
eff
of the composite sandwich beams (which considers
the combined effect of bending and shear deformations) was deter-
mined fromthe results of the experimental investigation. Using the
initial linear elastic portion of the loadmidspan deection curve
(Figs. 3 and 4), (EI)
eff
was calculated using the relation:
EI
eff

59
3000
L
3
DP
Dv
_ _
3
where (DP/Dv) is the slope of the loaddeection curve. The appar-
ent bending modulus of elasticity, E
app
of the composite sandwiches
was then computed by dividing (EI)
eff
by the second moment of area
of the homogenised cross-section of the composite sandwich beam.
The predicted EI and the calculated stiffness (EI)
eff
, apparent bend-
ing modulus and the maximum load, P
max
and bending strength,
r
b,max
of the composite sandwich beams obtained from the experi-
ment are reported in Table 3.
The results show that for individual composite sandwiches in
the atwise position, shear deformation has no signicant effect
on the bending stiffness as the predicted EI is almost equal to the
result of the experimental investigation. On the other hand, the
(EI)
eff
for the two sandwich beams bonded together is 37% lower
than the predicted values. This suggests that shear deformation
could have contributed to the total deformation of the composite
sandwich in the edgewise position due to the decreased span to
depth ratio. For longer beams with 3 and 4 laminations, the (EI)
eff
is slightly higher than the predicted EI. The increased bending stiff-
ness observed in the test is attributed to the friction effects be-
tween the laminations provided by the epoxy adhesives. The
difference between the predicted EI and (EI)
eff
is higher in speci-
mens with four laminations as these specimens have lower span
Fig. 6. Load and strain relationship of specimens 3LSW and 4LSW.
Table 3
Predicted and calculated EI, E
app
and bending strength of the sandwich beams.
Specimen EI (10
6
) (N mm
2
) (EI)
eff
(10
6
) (N mm
2
) E
app
(N/mm
2
) P
max
(N) r
b,max
(MPa)
1LSW-F 256 247 8073 4554 202
1LSW-E 844 819 3957 5589 194
2LSW-F 1287 1207 4188 9472 186
2LSW-E 1411 1373 3723 13,788 259
3LSW-F 4753 4851 4253 9318 195
3LSW-E 4107 4270 3969 11,247 259
4LSW-F 13,997 14,811 4047 20,869 200
4LSW-E 12,581 13,196 3988 26,086 257
2708 A.C. Manalo et al. / Composite Structures 92 (2010) 27032711
to depth ratio than the specimens with three laminations. In gen-
eral, both EI and (EI)
eff
in the atwise position is higher than that in
the edgewise position for specimens with 3 and 4 laminations.
Fig. 7 shows the relationship between the apparent bending
modulus of the glue-laminated composite sandwich beams and
the number of laminations. The results show that the E
app
of the
glue-laminated composite sandwich beams in the atwise position
decreases with increasing number of laminations. This decrease in
E
app
when the composite sandwich beams were laminated in the
atwise positions is expected as the bre composite skins near
the neutral axis of the section did not contribute as much stiffness
as the outermost skins. For both 3 and 4 laminations, the bending
stiffness in the atwise position is 7% higher than in the edgewise
position. Moreover, the E
app
of the individual composite sandwich
beams in the edgewise position is almost equal to that of the glued
composite sandwich beams. This clearly shows that the modulus of
elasticity in the edgewise position is not affected by the number of
laminations as the shear stresses induced by the exure are not
carried across the glue lines. Most importantly, the results showed
that the E
app
of the composite sandwich beams in the atwise and
edgewise positions converges with increasing laminations. This
information suggested that in the construction of beams with high-
er depth of the same width, it is better to glue together composite
sandwich structures in the edgewise position as these beams will
result in higher strength and with fewer glue lines.
4.2. Effect of glue laminations on bending strength of composite
sandwich beams
The relative performance of the glue-laminated composite
sandwich beams tested in the atwise and edgewise positions
was determined by calculating the apparent bending strength
based on the results of the experimental investigation. Similar
assumptions that no interlayer slips occurred and the laminated
sandwiches acted as a solid section with perfect bonding were
made. The bending strength of the composite sandwich beams
was determined from the maximum load measured in the experi-
ment and was calculated using Eq. (4).
r
b;exp

PLD
10EI
eff
E
s
4
Fig. 8 shows the maximum bending strength of the glue-lami-
nated composite sandwich beams for the different number of lam-
inations. The results show that for individual composite sandwich
beams, the bending strength of specimens in the atwise position
is similar to that of the specimen in the edgewise position. In these
sandwich beams, the specimens failed due to compressive failure
of the bre composite skins at a stress of around 200 MPa. It is
noteworthy that this stress value represents the level at which
the bre composite skins failed in compression as determined from
the coupon test.
It is seen from Fig. 8 that, in the glue-laminated composite
sandwich beams tested in the atwise position, the bending
strength capacity is not affected by the presence of glue lines.
There was no difference on the bending strength observed for
the different number of laminations. For all the composite sand-
wich beams tested in the atwise position, failure occurred in a
brittle manner due to compressive failure of the topmost skin fol-
lowed immediately by debonding between the core and the bre
composite skins. This result suggests that the bending strength of
glue-laminated sandwich beams in the atwise position depends
largely on the compressive properties of the bre composite skins.
Gluing the composite sandwich beams together in the edgewise
position resulted in an increase of at least 25% in bending strength.
The results also show that the bending strength of sandwich beams
in the edgewise position increases with increasing number of lam-
inations. The bending strength of glue-laminated specimens is
higher than the individual laminations. In the edgewise position,
the non-horizontal skins prevented the widening of the tensile
cracks in the core, thus preventing premature failure. Even after
compressive failure of the bre composite skins at the outermost
sandwich laminations, the beam continued to carry load until fail-
ure as the load was shared to the inner bonded sandwich lamina-
tions. The structural epoxy adhesives has also provided some
reinforcing effect which prevented the compressive failure and
buckling of the bonded non-horizontal bre composite skins,
delaying its failure, thereby, increasing its strength. This shows
that the glue lines acted as a load-distributing element and hold
the composite sandwich beams together. This load sharing mecha-
nisms led to the increased performance of the glue-laminated
sandwich beams. Furthermore, the loaddeection behaviour of
the composite sandwiches tested in the edgewise positions sug-
gests that, in this position, the specimen will fail in a ductile man-
ner due to progressive failure of the skin. Finally, the results
suggest that in multiple composite sandwich beams in the edge-
wise position, the defect in individual sandwiches is compensated
by the stronger, adjacent sandwich beams.
4.3. Effect of bre wrapping on stiffness and strength
The comparison of the loaddeection relation of glue-lami-
nated composite sandwich beams with and without bre wraps
are shown in Figs. 9 and 10. The gures show that the initial
loaddeection behaviour of all specimens was linear and became
non-linear with a reduced stiffness up to failure. As expected, the
specimen with bre wraps behaved slightly stiffer and failed at a
higher load compared to specimen without wraps. The higher Fig. 7. Apparent bending modulus of glue-laminated sandwich beams.
Fig. 8. Bending strength of glue-laminated sandwich beams.
A.C. Manalo et al. / Composite Structures 92 (2010) 27032711 2709
stiffness of the wrapped specimen became more apparent when
cracking of the core occurs and until nal failure. This could be
due to the bridging effect provided by the bre wraps on the local
defects in the specimen. Moreover, the composite sandwich beams
with bre wraps tested in atwise position behaved slightly stiffer
than the specimens in edgewise position.
The behaviour of specimen with bre wraps tested in the at-
wise position is similar to the specimen without wraps before
cracking of the core. In specimens 3LSW-F and 4LSW-F, a big drop
in the load is observed when compressive failure of the skin oc-
curred. On the other hand, compressive failure of the skins and
cracking of the core in specimens 3LSW-WF and 4LSW-WF are rep-
resented by smaller load drops. Interestingly, the rst load drop oc-
curred at almost the same level of applied load and deection for
both wrapped and unwrapped specimens. As loading continues,
the load starts to rise again but with a reduced stiffness as shown
in the loaddeection curve. The wrapped specimen failed at 18
35% higher load and a higher deection than the unwrapped spec-
imens, thus showing more ductile behaviour. However, the results
also show that the bre wraps could not prevent or delay the com-
pressive failure of the bre composite skin. The bre wraps will
only held the composite sandwich laminations together thereby
preventing the separation between the skins and the core and
increasing the failure strength.
In the edgewise position, the bre composite wraps acted as a
load-distributing element which resulted in a more ductile load
deection behaviour. For specimens 3LSW-E and 4LSW-E, the pro-
gressive failure of the bre composite skin is represented by small
load drops similar to a saw-tooth pattern. As a consequence of
wrapping, the progressive failure of the bre composite skins in
specimens 3LSW-WE and 4LSW-WE is characterised by a decreas-
ing capacity but with a smoother non-linear loaddeection curve.
Similarly, the progressive failure of the bre wraps did not create
visible load drops due to small percentage of additional bres (sin-
gle wrap). The non-linear loaddeection response was terminated
by a sudden drop in the load as a result of the composite sandwich
beam failure. The increase in failure load of specimen with bre
wraps is in the order of 1015% compared to specimens without
wraps. Noticeably, both the specimen with and without bre wraps
failed at almost the same amount of deection. This result shows
that the strength and ductility of the glue-laminated composite
sandwich beams are controlled primarily by the composite sand-
wich beams and not that of the bre wraps. Nevertheless, the bre
wraps provided additional load sharing mechanism amongst the
bonded composite sandwich beams. The presence of bre wraps
prevented the compressive buckling and debonding of the outer-
most bre composite skins. This resulted in a wrapped glue-lami-
nated composite sandwich beams tested in the edgewise position
to fail in a more ductile manner than the unwrapped specimens.
In general, the maximum load recorded for specimens with -
bre wraps is almost the same in the atwise and edgewise posi-
tions. However, if the rst initiation of damage (represented by a
drop in the load) is considered as the failure of the specimen, the
composite sandwich beams in the atwise position has a 2025%
lower capacity than the specimen in the edgewise position. This
difference in strength between wrapped specimens in the edge-
wise and atwise positions is similar to the unwrapped specimens.
Finally, the increase in stiffness and strength is due to additional
reinforcement provided by the bres and not the conning effect
of bre wrapping.
4.4. Effect of wrapping on failure behaviour
The failure modes of glue-laminated composite sandwich
beams with bre wraps are shown in Fig. 11. The results show that
bre wrapping has some signicant effects on the failure behav-
iour of the composite sandwich beams. The presence of bre wraps
prevented the immediate failure of the specimen as it held up the
laminated sandwich beams together which resulted in a higher
strength before failure.
At the early stage of loading, the noise related to micro-cracking
of the core was evident. Prior to failure, the cracking noise are more
frequently heard. In all specimens, the failure of the glue-lami-
nated composite sandwich beams with bre wraps was initiated
at the compressive part at the constant moment region. This failure
mechanism is similar to what was observed in the specimen with-
out bre wraps. Also, the failure mechanisms of wrapped speci-
mens indicated that very little conning effect was provided by
the bre wrap. After compressive failure of the skin in the compos-
ite sandwich beams was detected, progressive failure of the bre
wraps immediately followed. Several points of debonding failure
were observed between the bre wrap and the specimen at the
compressive side followed by splitting of the bre wraps in ten-
sion. Thus, it was concluded that the failure of glue-laminated
composite sandwich beams is governed by the strength of the
composite sandwich beams and not that of the bre wraps. How-
ever, the wrapped specimen failed with more ductility than the
specimen without wrap.
The test results have shown some positive effects of bre wrap-
ping on the exural behaviour of glue-laminated composite sand-
wich beams. In all of the experimental cases, there was
considerable amount of increase in the strength and stiffness of
the wrapped specimens compared to specimens without bre
wraps. However, its overall effect on the behaviour of glue-lami-
nated composite sandwich beams cannot be justied because of
the additional cost of preparation and bres for wrapping. The fail-
ure initiation of glue-laminated composite sandwich beams with
bre wraps is almost the same as that for specimen without wraps.
More effective results could have been obtained with specimens
with more layers of bre wrapping but increasing the number of
bre wraps would denitely entail higher costs.
Fig. 10. Load and midspan deection relationship of specimens 4LSW and 4LSW-W.
Fig. 9. Load and midspan deection relationship of specimens 3LSW and 3LSW-W.
2710 A.C. Manalo et al. / Composite Structures 92 (2010) 27032711
5. Conclusions
The exural behaviour of glue-laminated composite sandwich
beams was determined through experimental investigation. The
results showed that gluing the composite sandwich beams to-
gether resulted in a stronger and more stable section than individ-
ual sandwich beams alone. The results also suggest that using the
same amount of material, glue-laminated composite sandwich
beams in the edgewise position could offer up to 25% increase in
strength compared to beams in atwise position but with a slightly
lower bending stiffness. The presence of non-horizontal skins in
the edgewise position increases the load carrying capacity and re-
sult in a more ductile failure behaviour. The nal failure of the
specimen in the edgewise position is due to simultaneous com-
pressive failure of the inner skins, crushing of the core and tensile
failure of the bre composite skins. In the atwise position, the
failure is governed by the compressive failure of the skin followed
by debonding between the skin and the core, thus resulting to brit-
tle failure. Furthermore, the overall effect of one layer of tri-axial
glass wrapping on the exural behaviour of glue-laminated com-
posite sandwich beams cannot be justied with the additional cost
of bres and preparation. Even with bre wraps, the failure initia-
tion of glue-laminated composite sandwich beams is almost the
same as that for specimen without bre wraps. Finally, the results
of this study demonstrated the high possibility of developing a
structural beam from glue-laminated composite sandwich struc-
tures. Currently, research is being conducted to optimise the
glue-laminated composite sandwich beams with the objective of
exploring the practical application of this for composite structures.
Acknowledgements
The authors gratefully acknowledge the technical and materials
support provided by Dr. Gerard Van Erp and the staff of LOC Com-
posites Ltd., Pty., Australia. The support of Mr. Atul Sakhiya and Mr.
Christopher Pickford in the preparation and testing of the compos-
ite sandwich beams are greatly acknowledged.
References
[1] Belouettar S, Abbadi A, Azari Z, Belouettar R, Freres P. Experimental
investigation of static and fatigue behaviour of composite honeycomb
materials using four point bending tests. Compos Struct
2008;87(3):26573.
[2] Bakis CE, Bank LC, Brown VL, Cosenza E, Davalos JF, Lesko JJ, et al. FRP
composites in construction state of the art review. ASCE Compos Constr J
2002;6(2):7887.
[3] Dai J, Thomas Hahn H. Flexural behaviour of sandwich beams fabricated by
vacuum-assisted resin transfer moulding. Compos Struct 2003;61:24753.
[4] Borsellino C, Calabrese L, Valenza A. Experimental and numerical evaluation of
sandwich composite structures. Compos Sci Technol 2004;64:170915.
[5] Wicks N, Hutchinson JW. Optimal truss plates. Int J Solids Struct
2001;38:516583.
[6] Kooistra GW, Wadley HNG. Lattice truss structures from expanded metal
sheet. Mater Des 2007;28:50714.
[7] Van Erp G, Rogers D. A highly sustainable bre composite building panel. In:
Proceedings of the international workshop on bre composites in civil
infrastructure past, present and future, 12 December 2008. Toowoomba,
Queensland, Australia: University of Southern Queensland; 2008.
[8] Manalo AC, Aravinthan T, Karunasena W, Islam MM. Flexural behaviour of
structural bre composite sandwich beams in atwise and edgewise positions.
Compos Struct 2009;92(4):98495.
[9] Boughton GN, Crews KI. Timber design handbook. Standards Australia HB 108-
1998.
[10] Johnson KA. Timber bridge decks: comparison of types and evolution of
design. St Louis Park (MN): Wheeler Consolidated Inc.; 2002.
[11] Lopez-Anido R, Xu H. Structural characterization of hybrid bre-reinforced
polymer-glulam panels for bridge decks. J Compos Constr
2002;6(3):194203.
[12] Kumar P, Chandrashekhara K, Nanni A. Structural performance of a FRP bridge
deck. Constr Build Mater 2004;18:3547.
[13] Williams B, Shehata E, Rizkalla S. Filament-wound glass bre reinforced
polymer bridge deck modules. ASCE J Compos Constr 2003;7(3):26676.
[14] Rogers D. Characterisation of Hyrez 201 laminating resin. Toowoomba,
Queensland: Polymer Testing Laboratory, University of Southern
Queensland; 2004.
[15] American Society of Testing and Materials. Standard test method for exural
properties of sandwich constructions (C393-00). West Coshohocken (PA):
ASTM International; 2000.
Fig. 11. Failure of glue-laminated composite sandwich beams with bre wraps.
A.C. Manalo et al. / Composite Structures 92 (2010) 27032711 2711