Vous êtes sur la page 1sur 13

Engineering Structures 31 (2009) 13691381

Contents lists available at ScienceDirect


Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Flexural behaviour of multi-span GFRP-concrete hybrid beams
Joo R. Correia

, Fernando A. Branco, Joo G. Ferreira


Civil Engineering and Architecture Department, Instituto Superior Tcnico - ICIST, Technical University of Lisbon. Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal
a r t i c l e i n f o
Article history:
Received 14 July 2008
Received in revised form
16 September 2008
Accepted 2 February 2009
Available online 28 February 2009
Keywords:
Fibre reinforced polymers
Glass-reinforced plastics
GFRP pultruded profiles
Pultrusion
Concrete
Shear connection
Composite beams
Composite structures
a b s t r a c t
This paper presents results of an experimental and analytical study on the flexural behavior of multi-
span GFRP-concrete hybrid beams. Previous investigations showed the viability and technical advantages
of connecting GFRP pultruded profiles to concrete elements in GFRP-concrete hybrid cross-sections,
used in simply supported structural elements. The objective of the present study was to investigate the
flexural behaviour of the GFRP-concrete hybrid solution in continuous structural elements, in which,
in addition to the lower deformability when compared to isostatic systems, the cross-sectional and
structural redundancies may also allow for a certain pseudo-ductility at failure. With this purpose, a set
of design equations was derived in order to analyse the flexural behaviour of continuous hybrid beams,
subjected to positive and negative bending. Those equations were used to design two continuous 3-span
and 2-span hybrid beams with a bonded shear connection system, which were tested in bending. The
flexural tests allowed us to investigate the serviceability behaviour and the failure mechanisms of multi-
span GFRP-concrete hybrid beams.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Glass fibre reinforced polymer (GFRP) pultruded profiles have
significant potential for use in civil infrastructure, owing to their
several advantageous properties when compared to traditional
materials. These advantages include high strength, lightness,
free formability, ease of installation, good thermal and electro-
magnetic insulation properties, low maintenance requirements
and resistance to corrosion and fatigue [1]. Initially used mainly in
non-structural elements or in secondary structures, in the last few
years, GFRP pultruded profiles have found an increasing number
of applications in primary structures of bridges and buildings,
in both the rehabilitation of degraded structures and in new
constructions [26].
Despite those several advantageous properties over traditional
materials, GFRP profiles also feature some technical disadvantages.
The most important structural constraints are associated with
the lack of material-adapted structural shapes, the relatively low
elasticity modulus, the brittle behaviour and the susceptibility to
instability phenomena, which prevent the full exploitation of the
GFRP material properties.
Previous investigations reported by Deskovic et al. [7], Hall
and Mottram [8], Canning et al. [9], Fam and Rizkalla [10],

Corresponding author. Tel.: +351 218 418 212; fax: +351 218 488 481.
E-mail address: jcorreia@civil.ist.utl.pt (J.R. Correia).
Hullat et al. [11], Seible et al. [12], Correia et al. [13,14] and
Keller et al. [15] showed that one way to make a better
use of the GFRP profiles material is by connecting them to
concrete elements in GFRP-concrete hybrid elements. The several
hybrid concepts proposed by the different authors aimed at
combining the directional behaviour, the lightness and high
mechanical performance of FRP materials, with the most relevant
properties of traditional materials, particularly with concrete, a
less expensive and more massive material. Such a combination,
typically conceived and tested in simply supported structural
elements, led to solutions that are particularly useful for the
rehabilitation of old floors and also for new construction, resulting
in a synergetic increase of both stiffness and strength.
Based on previous investigations carried out at IST on the
use of GFRP-concrete hybrid cross-sections in simply supported
structural elements [13,14], this paper presents results of a further
study on the behaviour of a GFRP-concrete hybrid solution in
continuous structural elements. With this purpose, a set of design
equations was derived in order to analyse the flexural behaviour
of hybrid beams subjected to negative bending. Those equations,
together with a previously presented analytical formulation for
positive bending [13], were used to design two continuous 3-span
and 2-span hybrid beams comprising a bonded shear connection
system, which were tested in bending. The flexural tests allowed
us to investigate the serviceability behaviour and the failure
mechanisms of continuous beams submitted to negative (and
positive) bending moments.
0141-0296/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2009.02.004
1370 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
Notations
b width of the GFRP profile
b
c
width of the concrete layer
d
p
vertical distance between the top surface of the
concrete layer and the centre of the GFRP profiles
web
d
s
vertical distance between the top surface of the
concrete layer and the centre of the steel bars
f
c
compressive strength of concrete
f
cm
average compressive strength of concrete
f
ctm
average tensile strengh of concrete
f
su
steel failure stress
f
suk
characteristic value of steel failure stress
f
sy
steel yielding stress
f
syk
characteristic value of steel yielding stress
h height of the GFRP profile
h
c
thickness of the concrete layer
n
c
concrete homogenization coefficient
n
s
steel homogenization coefficient
r
j
radius of the jth steel bar
t
f
thickness of the GFRP profiles flanges
t
w
thickness of the GFRP profiles web
A
c
cross-sectional area of the concrete layer
A
f
cross-sectional area of the GFRP profiles flange
A
p
cross-sectional area of the GFRP profile
A
s
cross-sectional area of the steel reinforcing bars
A
w
cross-sectional area of the GFRP profiles web
E
c
Youngs modulus of concrete
E
p
full-scale longitudinal flexural modulus of the GFRP
profile
E
s
Youngs modulus of steel
G
p
full-scale shear modulus of the GFRP profile
H total height of the hybrid
I
c
principal moment of inertia of the concrete layer
(horizontal axis)
I
GFRP
eq
cross-sections equivalent inertia homogenized in
GFRP
I
p
principal moment of inertia of the GFRP profile
(horizontal axis)
K
1
, K
2
factors depending on the load position, given by the
elasticity theory
K
s
factor defining the proportionof shear carriedby the
GFRP profile
L beams span
M
u
ultimate bending moment
N number of steel bars
P value of the point load
X
e
position of the neutral axis for an elastic analysis
(measured from top of concrete layer)
X
u
position of the neutral axis at failure (measured
from top of concrete layer)
Z position of the neutral axis (measured from bottom
of GFRP profile)
deflection

f
deflection for complete shear interaction

p
midspan deflection for partial shear interaction

cu
ultimate compressive strain of concrete

f 2
average strain at the bottom flange of the GFRP
profile

s
strain at the steel reinforcing bars

su
ultimate tensile strain of steel reinforcing bars

sy
yielding strain of steel reinforcing bars

crit
local instability critical stress

f 2
stress at the bottom flange of the GFRP profile

s
stress at the steel reinforcing bars
2. The proposed GFRP-concrete hybrid system
2.1. Previous research carried out at IST
Fig. 1 shows the cross-section of the proposed GFRP-concrete
hybrid solution conceived to be used either in beams or in
slabs [16]. The structural unit element is composed of an off-
the-shelf GFRP I-profile connected to a concrete layer on the top
flange. Two alternative shear connection systems between the
GFRP pultruded profile and the concrete layer were developed and
investigated: (i) stainless steel bolts, fixed on both sides of the
profiles top flange and embedded in the concrete layer (bolted
solution); and (ii) an epoxy adhesive layer applied on the interface
between the two materials (bonded solution).
The hybrid structural elements can be produced in a pre-
fabrication plant or, alternatively, the concrete layer can be cast
in situ over the top flange of the GFRP profile, already supported
by vertical structural elements. In the latter solution, the bonded
system requires the use of an epoxy resin suitable for fresh
concrete (wet-in-wetbonding). In any case, both solutions are
relatively economic, since they make use of standard structural
GFRP pultruded profiles, not requiring any additional production
specifications. Both solutions are very easy to erect, due to the light
weight of the GFRP profiles, and can be particularly competitive in
the rehabilitation of degraded timber floors, when compared with
the alternative use of steel profiles. In fact, in such applications,
the additional dead load introduced by the GFRP profiles in the
partition walls is much less than that introduced by the competing
steel solution. Additionally, in construction rehabilitation, the
use of steel profiles generally requires elevation devices which,
besides being expensive, are very difficult to erect and operate
in historical districts where more construction rehabilitation is
needed. Rehabilitation solutions involving the use of GFRP profiles
avoid the need of such elevation devices, since the material can be
manually placed on the construction site.
In terms of structural systems, the proposed hybrid solution
was originally conceived to be used in simply supported structural
elements, particularly in rehabilitation applications. Therefore,
a set of equations for the flexural design of GFRP-concrete
hybrid beams submitted to positive bending were developed
and presented in [13]. Those equations were developed both for
both serviceability and ultimate limit states design, considering
the effect of the interconnection slip at the materials interface.
Shear connection tests were carried out to evaluate the stiffness
and strength of each shear connection system. Results of those
tests, presented in [14], were used to design simply supported
GFRP-concrete hybrid beams, with different spans and load
arrangements, which were tested in bending. The flexural
behaviour of the hybrid beams was analysed, regarding both
serviceability and ultimate behaviour, with particular relevance
to the effect of the interconnection slip. Tests reported in [14]
showed that the bonded shear connection system provided
greater stiffness when compared to the bolted system. However,
concerning strength, the bolted system allowed higher ultimate
loads; in fact, it was shown that the strength of the beams
with bonded systems, was limited by the strength of the
concrete-adhesive interface due to the development of shear
and normal tensile stresses at the bonding extremities. For both
shear connection systems, the overall behaviour of the hybrid
beams was compared with that of a simple GFRP I-profile,
demonstrating the technical advantages of this new constructive
solution: a very significant stiffness and strength increase, a
better use of the GFRP material and the possible occurrence
of a relatively more ductile failure mechanism due to concrete
crushing.
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1371
Fig. 1. The proposed GFRP-concrete hybrid solution: bolted solution (left) and bonded solution (right).
Fig. 2. Cross-sections dimensions and longitudinal strain and stress distributions for a negative bending moment.
2.2. Multi-span GFRP-concrete hybrid elements
In addition to simply supported systems, the proposed GFRP-
concrete hybrid section may also be used in continuous structural
elements. In such structurally redundant systems, the main advan-
tage may arise from the considerably reduced deformability when
compared to isostatic systems. Furthermore, taking into account
the possible occurrence of premature failure mechanisms at the in-
terface of hybrid beams with bonded shear connection, the cross-
sectional and structural redundancies may also allow for a certain
pseudo-ductility, which constitutes an additional advantage when
compared with the typical fragile failure mechanisms of simple
GFRP profiles.
3. Analysis of GFRP-concrete hybrid beams submitted to
negative bending
Fig. 2 illustrates the cross-section of the GFRP-concrete hybrid
beam, with the corresponding dimensions, together with the
strain and stress distributions corresponding to elastic and failure
analysis. The GFRP profile has a cross-sectional area A
p
, principal
moment of inertia I
p
(horizontal axis), webarea A
w
anda flange area
A
f
; the concrete layer has a cross-sectional area A
c
and principal
moment of inertia I
c
(horizontal axis); and the steel reinforcing
bars have a total cross-sectional area A
s
. The properties of the
materials are: E
p
= full-scale longitudinal flexural modulus of
the GFRP profile; G
p
= full-scale shear modulus of the GFRP
profile; E
c
= Youngs modulus of concrete; f
c
= compressive
strength of concrete;
cu
= ultimate compressive strain of concrete;
E
s
= Youngs modulus of steel; f
sy
= steel yielding stress;
f
su
= failure stress of steel;
sy
= yielding strain of steel;

su
=ultimate tensile strain of steel. For both the serviceability and
failure analyses, two alternative positions of the neutral axis were
considered: either in the concrete layer or in the GFRP profile.
3.1. Serviceability analysis
The elastic analysis of a GFRP-concrete hybrid section for
negative bending moments (Fig. 2(a)) can be performed assuming
the following classical hypotheses: Bernoullis hypothesis is valid;
the whole width of the concrete slab is effective; there is complete
shear interaction between the concrete slab and the GFRP profile
(i.e., there is no longitudinal slippage between the materials);
and there is no vertical separation between the concrete slab and
the GFRP profile. Those hypotheses lead to the following bending
equations for the depth of the neutral axis (X
e
) and the sections
equivalent inertia homogenized in GFRP (I
GFRP
eq
), with n
c
= E
c
/E
p
and n
s
= E
s
/E
p
being the homogenization coefficients of concrete
and steel, respectively:
(a) Considering concretes tensile strength (valid before crack-
ing occurs),
X
el
=
A
p
d
p
+n
c
A
c

h
c
2
+n
s
A
s
d
s
A
p
+n
c
A
c
+n
s
A
s
(1)
I
GFRP
eq
= I
p
+A
p

_
h
c
+
h
2
X
e
_
2
+
n
c
A
c
h
2
c
12
+n
c
A
c

_
X
e

h
c
2
_
2
+n
s
A
s
(X
e
d
s
)
2
+n
s

j=1
r
4
j
4
(2)
where d
p
and d
s
are the vertical distances between the top surface
of the concrete layer and the middle of the web and the centre of
the steel reinforcing bars, respectively, and r
j
is the radius of the jth
steel reinforcing bar (of a total of N bars).
(b) Not considering concretes tensile strength (valid after
cracking occurs):
1372 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
(b
1
) Neutral axis lying on the concrete layer (Fig. 2(a1),
_
n
c
b
c
2
_
X
2
e

_
A
p
+n
c
A
c
+n
s
A
s
_
X
e
+
_
A
p
d
p
+n
c
A
c

h
c
2
+n
s
A
s
d
s
_
= 0 (3)
I
GFRP
eq
= I
p
+A
p

_
h
c
+
h
2
X
e
_
2
+n
c

b
c
(h
c
X
e
)
3
3
+n
s

_
N

j=1
r
4
j
4
+A
s
(X
e
d
s
)
2
_
(4)
(b
2
) Neutral axis lying on the GFRP profile (Fig. 2(a2)),
X
el
=
A
p
d
p
+n
s
A
s
d
s
A
p
+n
s
A
s
(5)
I
GFRP
eq
= I
p
+A
p

_
h
c
+
h
2
X
e
_
2
+n
s

_
N

j=1
r
4
j
4
+A
s
(X
e
d
s
)
2
_
. (6)
The global deformation of a GFRP-concrete hybrid beam is the
sum of the deflection due to bending, considering the connection
flexibility, and the deflection due to shear. A shear load acting on
the beam is distributed between the concrete slab and the GFRP
profile but, in the latter, the shear deformation can be significant.
A conservative assumption can then be adopted, namely, that
shear is carried only by the profiles web. Therefore, the maximum
deflection of a GFRP-concrete hybrid beam, subjected to a point
loadP appliedat any point along its spanL, is givenby the following
equation,
=
_

f
_

P L
3
K
1
EI
GFRP
eq
+K
s

P L
K
2
G
p
A
w
(7)
where K
1
and K
2
are factors that depend on the load position
and are given by the elasticity theory (as an example, for a load
applied at the extremity of a cantilever, K
1
= 3 and K
2
=
1), K
s
is the proportion of shear carried by the GFRP profile
(which can be conservatively assumed to be K
s
= 1.0) and
p
/
f
is the ratio between bending deflections considering complete
shear interaction (
f
) and partial shear interaction (
p
), which, as
discussed in [14], can be approximately considered to be 1.0 in
beams with bonded shear connection.
3.2. Failure analysis
The hybrid beam subjected to negative bending moments,
can fail due to several mechanisms, among which, web shear
failure (fracture or buckling), web compressive failure (crushing
or local buckling), concrete shear failure, concrete compressive
failure (flexural), steel bars tensile failure (flexural), bottom flange
compressive failure (crushing or local buckling), lateraltorsional
global buckling failure and interface shear connection failure.
For most of the above mentioned failure mechanisms, design
can be carried out using equations available in the literature
(e.g., [17]). Concrete compressive failure (flexural) mechanism of
GFRP-concrete hybrid beams in negative bending is not likely to
occur, if a rational design is carried out.
The failure analysis of a GFRP-concrete hybridsection(Fig. 2(b)),
assuming full shear connection was developed for two different
bending failure mechanisms: (i) tensile failure of the steel bars; and
(ii) local buckling failure of the GFRP bottom flange. For the tensile
failure of the steel bars, the analysis was developed assuming that
the stress level at the steel bars equals the steel yielding stress
(
s
= f
sy
), and the strain level at the steel bars depth equals the
steel ultimate tensile strain (
s
=
su
= 1%). For the local buckling
failure of the GFRP bottom flange mechanism, results are given
as a function of the local instability critical stress (
f 2
=
crit
,
with
f 2
=
crit
/E
p
), which, in general, depends on the loading
arrangement and also on the supporting conditions, and can be
estimated through analytical and numerical formulations or, in
alternative, based on experimental tests. Regarding the stress state
at the steel bars, two alternative possibilities were considered:
(i) local buckling failure occurs without yielding of the steel bars
(
s
< f
sy
); or (ii) local buckling failure occurs with yielding of the
steel bars (
s
= f
sy
, with
sy
<
s
<
su
and
sy
= f
sy
/E
s
). These two
different failure scenarios correspond to different levels of ductility
of the cross-section, which will increase with the yielding strain
of the steel bars at failure. For each one of the above mentioned
bending failure mechanisms, design equations were developed
allowing to determine the neutral axis depth (X
u
) and the ultimate
bending moment (M
u
).
(a) Tensile failure of the steel bars.
(a1) Neutral axis lying on the concrete layer
_
b
c
2
E
c
_
X
2
u

_
A
p
E
p
+A
c
E
c
+A
s

f
sy

su
_
X
u
+A
f
E
p
(h
c
+H) +A
w
E
p

_
h
c
+
h
2
_
+A
c
E
c

h
c
2
+A
s

f
sy

su
d
s
= 0 (8)
M
u
= A
f
E
p

su

_
h
c
X
u
+
t
f
2
_
2
X
u
d
s
+A
w
E
p

su

_
h
c
X
u
+
h
2
_
2
X
u
d
s
+A
f
E
p

su

_
H X
u

t
f
2
_
2
X
u
d
s
+b
c
(h
c
X
u
) E
c

su
4

(h
c
X
u
)
2
X
u
d
s
+A
s
f
sy
(X
u
d
s
) . (9)
(a2) Neutral axis lying on the GFRP profile
X
u
=
A
f
E
p
(h
c
+H) +A
w
E
p

_
H
h
2
_
+A
s

f
sy

s
d
s
A
p
E
p
+A
s

f
sy

s
(10)
M
u
= A
f
E
p

su

_
X
u
h
c

t
f
2
_
2
X
u
d
s
+A
w
E
p

su

_
H X
u

h
2
_
2
X
u
d
s
+A
f
E
p

su

_
H X
u

t
f
2
_
2
X
u
d
s
+A
s
f
sy
(X
u
d
s
) . (11)
(b) Local buckling failure of the GFRP bottom flange without
yielding of the steel bars.
(b1) Neutral axis lying on the concrete layer
_
b
c
2
E
c
_
X
2
u

_
A
p
E
p
+A
c
E
c
+A
s
E
s
_
X
u
+A
f
E
p
(h
c
+H) +A
w
E
p

_
h
c
+
h
2
_
+A
c
E
c

h
c
2
+A
s
E
s
d
s
= 0 (12)
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1373
X
u
=
A
f
E
p
(h
c
+H) +A
w
E
p

_
H
h
2
_
A
s
E
s

f
sy

crit

_
H
t
f
2
_
A
p
E
p
+A
s
E
s
(18)
M
u
= A
f

crit

_
X
u
h
c

t
f
2
_
2
H X
u

t
f
2
+A
w

crit

_
H X
u

h
2
_
2
H X
u

t
f
2
+A
f

crit

_
H X
u

t
f
2
_
+A
s
f
sy
(X
u
d
s
) (19)
Box I.
Table 1
Mechanical properties of the GFRP profile from coupon testingaverage and standard deviation values.
Mechanical property Test and direction
Flexural longitudinal
(ISO 14125)
Tensile longitudinal
(ISO 527)
Compression longitudinal
(ASTM D695)
Compression transverse
(ASTM D695)

u
(MPa) 624.6 26.9 475.5 25.5 375.8 67.9 122.0 15.4
E (GPa) 26.9 1.3 32.8 0.9 26.4 1.9 7.4 0.4

u
[10
3
] 24.9 1.3 15.4 1.5 17.0 2.5 21.5 1.7

xy
[-] 0.28
Interlaminar shear strength, F
sbs
= 35.0 3.9 MPa (ASTM D2344)
In-plane shear strength,
u
= 47.1 MPa (obtained from tensile tests on double lap joints, [18])
M
u
= A
f

crit

_
h
c
X
u
+
t
f
2
_
2
H X
u

t
f
2
+A
w

crit

_
h
c
X
u
+
h
2
_
2
H X
u

t
f
2
+A
f

crit

_
H X
u

t
f
2
_
+n
c

b
c
(h
c
X
u
)
4

crit

(h
c
X
u
)
2
H X
u

t
f
2
+n
s
A
s

crit

(X
u
d
s
)
2
H X
u

t
f
2
. (13)
(b2) Neutral axis lying on the GFRP profile
X
u
=
A
f
E
p
(h
c
+H) +A
w
E
p

_
H
h
2
_
+A
s
E
s
d
s
A
p
E
p
+A
s
E
s
(14)
M
u
= A
f

crit

_
X
u
h
c

t
f
2
_
2
H X
u

t
f
2
+A
w

crit

_
H X
u

h
2
_
2
H X
u

t
f
2
+A
f

crit

_
H X
u

t
f
2
_
+n
s
A
s

crit

(X
u
d
s
)
2
H X
u

t
f
2
. (15)
(c) Local buckling failure of the GFRP bottom flange with yielding
of the steel.
(c1) Neutral axis lying on the concrete layer
_
b
c
2
E
c
_
X
2
u

_
A
p
E
p
+A
c
E
c
A
s
E
p

f
sy

crit
_
X
u
+A
f
E
p
(h
c
+H) +A
w
E
p

_
h
c
+
h
2
_
+A
c
E
c

h
c
2
A
s
E
p

f
sy

crit
_
H
t
f
2
_
= 0 (16)
M
u
= A
f

crit

_
h
c
X
u
+
t
f
2
_
2
H X
u

t
f
2
+A
w

crit

_
h
c
X
u
+
h
2
_
2
H X
u

t
f
2
+A
f

crit

_
H X
u

t
f
2
_
+n
c

b
c
(h
c
X
u
)
4

crit

(h
c
X
u
)
2
H X
u

t
f
2
+A
s
f
sy
(X
u
d
s
) . (17)
(c2) Neutral axis lying on the GFRP profile (See Box I).
4. Flexural tests on multi-span GFRP-concrete hybrid beams
4.1. Experimental programme
As previously referred, the objective of the experimental
programme was to investigate the behaviour of the GFRP-concrete
hybrid solution in continuous structural elements.
Hybrid beam HB6, with a total length of 6.00 m, presenting a
2.00 m central span and two 2.00 m cantilever spans, was tested
in bending, subjected to two point loads applied at a distance
of 0.10 m from the extremities of both cantilevers. Hybrid beam
HB6 was tested in a very particular structural system (with two
considerably slender cantilevers), not likely tobe usedinstructures
of practical interest. The maininterest of this test was toinvestigate
the behaviour of the GFRP-concrete hybrid solution in a structural
element subjected only to negative bending.
Hybrid beam HB7, also with a total length of 6.00 m and
two 2.80 m spans, was tested in bending, subjected to two point
loads applied at approximately 3/8 of each one of the two spans.
Therefore, the 2-span continuous beamHB7 was subjected to both
positive and negative bending moments.
4.2. Materials
The pultruded GFRP I-profile used in the flexural tests
(produced by Topglass and supplied by STEP) is made of a polyester
matrix reinforced with E-glass fibres (inorganic content of 62%,
by weight) and has the following nominal dimensions: height
h = 200 mm, width b = 100 mm and thickness of the
web and flanges t
w
= t
f
= 10 mm. Tensile, compressive,
flexural and shear properties were determined through extensive
mechanical characterization based on experiments on coupons
cut from the original profile -Table 1. The profiles longitudinal
flexural modulus (E
p
= 38.4 GPa) and shear modulus (G
p
=
3.58 GPa) were determined in a full-scale test performed on a
profile similar to that used in the tested beams [18]. The average
compressive strength (f
cm
= 27.1 MPa) and the Youngs modulus
(E
c
= 29.0 GPa) of concrete were experimentally determined. The
shear connection was provided by a low viscosity two-component
epoxy adhesive (with a Trademark MBrace Resin 220, supplied
by S & P Clever Reinforcement), with a density of 1.75 g/cm
3
,
1374 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
Fig. 3. Test setup of hybrid beam HB6: frontal view (top), section S5 (bottom left) and frontal view of section S6 (bottom right), dimensions in cm.
a modulus of elasticity of 7.5 GPa and an average tensile strength of
33.0 MPa [19]. The steel bars used to reinforce the concrete slab of
the hybrid beams, presented the following nominal characteristic
values of the yielding stress and failure stress: f
syk
= 500 MPa and
f
suk
= 550 MPa, respectively.
4.3. Properties of the hybrid beams
The cross-section of the hybrid beams was selected regarding
the preferential application of this type of structural elements,
namely, in the rehabilitation of degraded constructions, as simply
supported elements, with a reference span of 4.00 m, and based
on design equations for positive bending previously developed by
Correia et al. [13]. Consequently, for such structural system, in
order to obtain a concrete compressive failure with the neutral axis
lying at the concrete slab and simultaneously making a good use of
the GFRP profiles material properties, a concrete slab 0.10 m thick
and 0.40 m wide was adopted. Based on results reported in [14],
in order to maximize the flexural stiffness of the beams, a bonded
shear connection system was adopted, using a 2.0 mm adhesive
thickness.
The longitudinal reinforcements of hybrid beams HB6 and HB7
were defined using the previously presented design equations for
negative bending so that flexural failure occurs with local buckling
of the profiles bottom flange, but with some level of plasticization
of the steel bars placed on the top of the concrete layer (
s
=
0.40%). The steel reinforcement resulted from a compromise
between the ultimate bending strength (that increases with the
steel reinforcement) and the level of plasticization of the steel
bars at failure, which is a measure of the section ductility
(that decreases with the steel reinforcement). Calculations were
performed assuming a critical stress for the local buckling of
the bottom flange of 250 MPa, defined based on a flexural test
(described in [18]) previously performed on a simply supported
4.00 m span GFRP profile (from the same batch of the profiles
used in the fabrication of the hybrid beams), laterally braced in
4 sections along its span. Although the cross-section (without
concrete layer), the loading arrangement and the supporting
conditions were different in hybrid beams HB6 and HB7 (and all
these factors influence the local buckling stresses), the result of
that test was used as a reference due to the present difficulties
in the estimation of local (and global) buckling loads in pultruded
elements [20]. As a result, the concrete layer of hybrid beams HB6
and HB7 was reinforced longitudinally at the top, with 416 mm
(8.04 cm
2
) structural steel bars and, at the bottom, with 26 mm+
28 mm steel bars (1.57 cm
2
). Hybrid beam HB6 was reinforced
transversally with 6 mm//0.10m shear stirrups (5.65 cm
2
/m),
while hybridbeamHB7 was reinforcedwith8 mm//0.10 mshear
stirrups (10.05 cm
2
/m). In both hybrid beams, a 15 mm cover was
used for the steel bars reinforcements.
The flexural properties of hybrid beams HB6 and HB7 (cross-
sections subjected to negative bending moments) were deter-
mined using the previously presented design equations, based
on the experimentally determined mechanical properties: neutral
axis depth before concrete cracking X
e
= 61.6 mm; flexural stiff-
ness for complete shear interaction EI
GFRP
eq
= 3224 kN m
2
and
neutral axis depth X
e
= 105.2 mm (both after concrete crack-
ing); ultimate bending moment for full shear connection M

u
=
96.9 kN m; and neutral axis depth at failure X
u
= 127.8 mm. The
flexural properties of hybrid beam HB7 for the cross-sections sub-
jected to positive bending moments were determined fromthe de-
sign equations presented in [13]: flexural stiffness for complete
shear interaction EI
GFRP+
eq
= 4468 kN m
2
and neutral axis depth
X
e
= 58.8 mm (both after concrete cracking); and ultimate bend-
ing moment for full shear connection M
+
u
= 148.0 kN m.
4.4. Experimental setup and test procedure
Figs. 3 and 4 show the loading arrangement and supporting
conditions of hybrid beams HB6 and HB7, together with the
instrumentation.
The supports of the hybrid beams were materialized by 5 cm
diameter steel rollers with 8 cm long top steel plates. All beam
supports allowed for free rotation and one (in HB7) or two (in
HB6) of themalso allowed for longitudinal sliding. In the supports
sections of both hybrid beams, two steel plates with a C-cross-
section (180 mm32 mm6 mm) and a length of 250 mm were
used as stiffeners, fixed on both sides of the web with an epoxy
adhesive, in order to prevent a premature web compressive failure
mechanism.
Loads were applied at a distance of 10 cm from the cantilever
extremities in hybrid beam HB6 and at two sections 140 cm
distanced from the central support of hybrid beam HB7, using two
Enerpac hydraulic jacks with a load capacity of 300 kN, which were
connected in parallel.
Two transversal loading frames were mounted, aligned with
the beams load application cross-sections. The hydraulic jacks,
positioned below the transversal frames, transmitted the load to
the top surface of the concrete layer, through steel spreading plates
(8 cm wide, 5 cm thick). The load was measured by Novatech load
cells (with a load capacity of 300 kN), placed between the steel
spreading plates and the hydraulic jacks. Metallic spherical rotulas
were placed between the load cells and the spreading plates, in
order to avoid possible transversal loading.
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1375
Fig. 4. Test setup of hybrid beam HB7: frontal view (top), section S3 (bottom left), frontal view of section S5 (bottom centre) and section S5 (bottom right), dimensions
in cm.
In hybrid beam HB6, vertical displacements were measured at
sections S1 and S2 (cantilever extremities) and at midspan (section
S5), using TML displacement transducers, model CDP100, with
a stroke of 100 mm and a precision of 0.01 mm. Longitudinal
strains were measured throughout the depth of central midspan
section S5, using 16 strain gauges. The distortions at section S6
were measured with 3 sets of strain gauges, forming 3 rosettes
at different depths of the GFRP profiles web. In hybrid beam
HB7, vertical displacements were measured at the load application
sections (S1 and S2) and at midspansections (S3 and S4), using TML
displacement transducers, model CDP100, with a stroke of 100 mm
and a precision of 0.01 mm. Longitudinal strains were measured
throughout the depth of midspan section S3 (subjected to positive
bending moments), using 16 strain gauges and at section S5
(subjected to negative bending moments), using 16 additional
strain gauges. The distortions at section S5 were measured with
3 sets of strain gauges, forming 3 rosettes at different depths of
the GFRP profiles web. In order to measure concrete deformations,
strain gauges of the type 1-LY41-50/120 were used together with
a two-component adhesive (reference X60, specific for concrete
applications), both manufactured by HBM. In the GFRP profile,
strain gauges of the type FLK-10-11, manufactured by TML, were
bonded with a single component cyanacrylate adhesive (reference
Z70), produced by HBM.
Since hybrid beam HB7 consisted of a statically indeterminate
system, in order to evaluate the internal forces during the test, it
would not have been sufficient to measure only the applied loads.
Therefore, it was decided to measure also the reaction force in the
central support. Four Novatech load cells (each one with a load
capacity of 100 kN) were then placed between the metallic profile
and the metallic steel supporting plate, allowing measuring the
central support reaction.
Tests were performed under a load controlled rate of approxi-
mately 0.2 kN/s. Loads, displacements and strains during the tests
were registered in a PC, using a data acquisition system with 100
channels from HBM, model UPM100.
4.5. Hybrid beam HB6 - Results and discussion
4.5.1. Deformation and ultimate strength
The loaddeflection curves of section S1 (left cantilever), S2
(right cantilever) and S5 (central midspan) of hybrid beam HB6
(bonded) are shown in Fig. 5 together with the corresponding
deflection predictions obtained from Eq. (7). Displacements of
section S5 are plotted with a negative sign, since they are opposite
to displacements of sections S1 and S2.
Fig. 5. Hybrid beam HB6: load vs. deflection at section S1, S2 and S5.
The pattern of the three experimental loaddeflection curves is
identical, all of them presenting three different segments. There is
an initial segment, up to a total load of about 4 kN, which is linear.
Then, from a total load of about 4 kN to approximately 7 kN, the
behaviour is non-linear, with a loss of stiffness corresponding to
concrete cracking. Assuming a concrete average tensile strength
of f
ctm
= 1.9 MPa (based on the average compressive strength),
the total load corresponding to the cracking of support section S3
and S4 is 4.3 kN, which agrees with experimental results. After this
non-linear segment, the behaviour is again approximately linear
up to failure.
A very good agreement was obtained between experimental
results and models deflection predictions (Eq. (7) with K
s
= 1.0).
Although the consideration that the profile carries 100% of shear is
too conservative (as it will be shown below, it carries only about
50% of shear), in hybrid beam HB6, the shear contribution to total
deformation is reduced (in the central span, shear force is actually
null).
Fig. 6 shows the deformed shape of hybrid beam HB6 prior to
failure. Maximum displacements in both cantilevers extremities
were quite significant (about 90 mm) and concrete cracking was
also rather extensive, especially at the central span, but also with
a certain relevance at the cantilever spans. Failure occurred due to
lateraltorsional buckling of the structural element, at a total load
of 79.0 kN, which corresponded to an average compressive stress
at the GFRP profiles bottom flange of 194.0 MPa. This maximum
stress is 78% of the assumed value for the critical local buckling
1376 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
Fig. 6. Hybrid beam HB6: deformed shape prior to failure.
Fig. 7. Hybrid beam HB6: lateraltorsional buckling failure.
stress (250 MPa), used in the previously referred cross-section
failure analysis. Global instability was followed by shear failure
at the web-top flange interface, along the whole length of hybrid
beamHB6. In the two cantilever spans, the part of the GFRP profile
below the separation line rotated laterally relatively to the upper
flangesee Fig. 7.
4.5.2. Axial strains and composite action
Fig. 8 shows the results from strain gauge measurements
at section S5 (central midspan). It can be seen that the axial
deformations measured with the five strain gauges attached to the
bottomsurface of the bottomflange (12to16) are approximately
similar for a total loaduptoabout 50kN. Then, for increasing values
of the total load, measurements of those strain gauges become
different, with the absolute values of axial deformation increasing
across the bottom flange width (12 < 13 < 14 < 15 <
16). The order of this relationship and the increasing difference
between measurements with the total load confirm the triggering
failure mechanism, i.e., lateraltorsional buckling.
Axial strains measured at section S5 (central midspan) of
hybrid beam HB6 showed that, for different values of the bending
moment, there is no slippage at the interface between the
materials, thereby confirming the previous observation reported
in [14] regarding the complete shear interaction provided by
adhesively bonded shear connection systems.
Fig. 9 shows the neutral axis height (Z = H X) in section
S5 of hybrid beam HB6, as a function of the bending moment.
Experimental values show that the neutral axis height decreases
with increasing bending moment and corresponding concrete
cracking development. Experimental results present a very good
agreement with the values for the neutral axis depth before
and after concrete cracking, determined from Eqs. (1) and (5)
respectively, thereby validating the procedure proposed for the
analysis of cross-sections submitted to negative bending moments.
Fig. 8. Hybrid beam HB6: axial strains vs. total load (section S5).
Fig. 9. Hybrid beam HB6: neutral axis vs. moment (section S5).
Fig. 10. Hybrid beamHB6: distortional deformation and shear carried by the GFRP
profile vs. total load (section S6).
4.5.3. Shear distribution
Distortions at section S6, measured at different depths of the
GFRP profiles web, are presented in Fig. 10, as a function of the
total load. The distortions measured at the three different levels are
not uniform, with the maximum distortional deformation being
measured at the top rosette.
The distribution of shear force carried by the GFRP profile and
the concrete slab in section S6 was estimated by averaging the
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1377
Fig. 11. Hybrid beam HB7: load vs. midspan deflection.
Fig. 12. Hybrid beam HB7: deformed shape prior to final failure.
three rosettes measurementsFig. 10. It can be seen that the shear
proportion increases from about 40% for lower values of the total
load to about 60% prior to failure.
4.6. Hybrid beam HB7 - Results and discussion
4.6.1. Deformation and ultimate strength
The loaddeflection curves of sections S1 (left span) and S2
(right span) of hybrid beam HB7 (bonded) are shown in Fig. 11,
together with the corresponding deflection predictions.
In the first part of the plotted curves, up to a total load of about
220 kN, it can be seen that a reasonable agreement was obtained
between experimental deflections and the models predictions
(Eq. (7) with K
s
= 1.0), with the model overestimating the
measured deflections. The reason for this conservative lack of
accuracy has to do with the consideration that shear is entirely
carried by the profiles web. If, alternatively, a value of K
s
= 0.5
is considered (in fact, as shown next, for service loads the GFRP
profiles web carries about 50% of shear), the agreement between
experimental deflections and models predictions becomes very
good.
Experimental loaddeflection curves of section S1 and S2 are
divided in 3 parts: (i) in the first part, the total load monotonically
increased up to a value of 219.5 kN, for which a first local failure
or damage occurred in the beam, causing a total load reduction
to 175.5 kN; (ii) in the second part of the curves, the total load
increased fromabout 175.5 kN up to a value of 280.9 kN, for which
a second local failure or damage occurred, followed by a total load
reduction to 205.9 kN; (iii) in the third and final part, the total load
increased fromabout 205.9 kN up to a value of 268.2 kN, for which
final failure occurred causing the collapse of the hybrid beam.
Load-deflection curves are very similar in their first part,
presenting a linear behaviour up to a total load value of about
Fig. 13. Hybrid beamHB7: Local buckling of top flange followed by web-top flange
separation in the vicinity of cross-section S1.
100 kN, subsequently starting to present a certain non-linearity.
First local failure or damage occurred for a total load of 219.5 kN
and corresponded to the development of a crack at the adhesive-
concrete interface (a few millimetres inside the concrete layer),
along the first span, starting at the left extremity section and
ending at load section S1see illustration in Fig. 11. After this
first local failure, the load dropped to 175.5 kN and deflection in
the first span (where shear connection failure occurred) increased.
In opposition, deflection in the second span decreased almost
elastically. A considerable slippage at the interface between the
two materials was also observed in the left extremity of the hybrid
beam.
The second part of the experimental loaddeflection curves at
sectionS1andS2were bothnon-linear, but presentedconsiderably
different stiffnesses, with the first span presenting a much higher
deformability (because in almost half of its length, the two
materials of the hybrid beam had absolutely no shear interaction)
and the second span presenting a small stiffness increase (due
to the loss of stiffness of the first span). Second local failure or
damage occurred for a total load of 280.9 kN and corresponded to
the development of a crack at the adhesive-concrete interface (a
few millimetres inside the concrete layer), along the second span,
starting at section S2 and ending at central support section. After
this second local damage, the load dropped to 205.9 kN. Deflection
in the second span, along which the crack developed, increased
and, in opposition, deflection in the first span slightly decreased
following a non-linear unloading path.
In their third part, experimental loaddeflection curves at sec-
tion S1 and S2 were increasingly non-linear and the deformabil-
ity of the first span approached that of the second span (where a
significant loss of flexural stiffness had occurred in almost half of
its length). Fig. 12 shows the deformed shape of the hybrid beam
prior to final failure that occurred for a total load of 268.2 kN
and was triggered by the local buckling of the top flange, in the
vicinity of section S1. This type of failure was enabled because,
following local failure of the interface, the two materials became
separated and the concrete layer no longer prevented the GFRP
profiles top flange from buckling. After local buckling of the top
flange, a crack appeared at the web-top flange junction of the GFRP
profile, which developed along a short length (about 40 cm to
50 cm), in the vicinity of section S1. This crack caused the separa-
tion between the web and the top flange of the GFRP profile along
this length, as shown in Fig. 13. Following this separation, the con-
crete slab started crushing the GFRP profiles web, under load sec-
tion S1. During the crushing development, the web suffered a com-
pressive plate bucking failure, which corresponded to the final fail-
ure of the hybrid beam Fig. 14.
The redundancy provided at both, the cross-section level (al-
lowing for stress redistribution in the cross-section, following local
1378 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
Fig. 14. Hybrid beam HB7: web compressive buckling at cross-section S1.
Fig. 15. Hybrid beamHB7: ratio of load carried by the central support vs. total load.
failure of the shear connection system) and also at the structural
element level (allowing for internal forces redistribution, as dis-
cussed next), led to a relatively ductile structural behaviour. The
experimental loaddeflection curves at sections S1 and S2, pre-
sented in Fig. 11, illustrate this pseudo-ductility, due to the suc-
cessive damages suffered by the hybrid beam and the consequent
considerable deformation increase, which occurred in both spans
prior to collapse.
4.6.2. Reactions and internal forces
The previously described local damages suffered by the hybrid
beam and the corresponding stiffness variation caused an internal
force redistribution.
Fig. 15 shows the ratio of the total load carried by the central
support, as a function of the total load. It can be concluded that
in the first loading path (until the first local damage or failure
occurred), the central support carried about 52% of the total load.
This value is in very good agreement with that provided by the
model, 51.2%. In the second loading path, after first local damage
occurred at the external part of the first span, next to the left
support reaction, the ratio of load carried by the central support
increased to about 57%. Finally, in the third loading path, after
second local damage occurred at the internal part of the second
span (next to the central support), the reaction carried by the
central support decreased to about 53% of the total load, which
remained approximately constant up to final failure.
In parallel, shear force and bending moment distributions
suffered variations. Following the first local damage, for the same
total load, shear decreased at the outer segments of both spans
Fig. 16. Hybrid beam HB7: axial strains measurements at section S3.
(varying from 8% to 13%) and increased at their inner segments
(from 8% to 11%), while the bending moment decreased at load
section S1 and S2 (from 8% to 13%) and significantly increased at
central support section S6 (from35% to 40%). Following the second
local damage, for the same total load, shear increased at the outer
segments of both spans (varying from 5% to 12%) and decreased at
their inner segments (from 7% to 8%), while the bending moment
increased at load section S1 and S2 (from8% to 12%) and decreased
at central support section S6 (from 23% to 24%).
4.6.3. Composite action
4.6.3.1. Section S3. Fig. 16 presents the values of the axial strains
measured at section S3 (left midspan) of hybrid beam HB7 during
the three load phases. In order to facilitate the interpretation of
the results, axial strains are not plotted as a function of the total
load or bending moment, but as a function of time (not indicated).
Measurements during the first load phase reflect a relatively
linear behaviour of the section materials and confirm the previous
observations regarding the complete shear interaction provided
by adhesively bonded shear connection systems (no slippage at
the interface between the materials). As previously described, the
first local damage, which developed fromthe left extremity section
towards section S1 (very close to section S3), caused the separation
of the materials along the above mentioned length and originated
a considerable slippage between the two materials that constitute
the hybrid section. By comparing the axial strains during the first
and second load phases, it is possible to identify the effects of
this first local damage on the axial strains and composite action
of section S3. Fig. 16 shows that strain gauges 14 and 1516
(attached to the bottom of the web and the bottom surface of
the bottom flange, respectively), which were initially in tension,
maintain their sign after the first local damage and their values
continue to increase in the second load phase. However, in the
same figure, it can be seen that strain gauge 13 (attached at
middle depth of the web), although maintains its sign, does not
continue to increase from the first to the second load phase. More
significantly, Fig. 16 shows that strain gauges 811 and 12
(attached to the top flange and the top of the web, respectively),
which were initially under tension in the first load phase, change
sign in the second load phase, being now under compression.
The strain gauges placed in the concrete layer, 13 and 67
(attached to the top surface of the concrete layer and to the bottom
steel reinforcements, respectively), do not change sign. However,
analysis of the measurements showed that, for the same load
level, the magnitude of such axial strains increases after first local
damage.
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1379
Fig. 17. Hybrid beam HB7: axial strains vs. section depth, for increasing bending
moments [kN m] (section S3), comparison of three load phases (P1, P2 and P3).
The previously described changes become clearer when the
axial strains are plotted as a function of the section depth, for
different values of the bending moment and for the different load
phases, as shown in Fig. 17. A comparison of the axial strain
distributions in the first (P1) and second (P2) load phases, for
the same values of the bending moment (25 kN m, 30 kN m
and 35 kN m), shows the effect of the first local damage on
the composite action of section S3: it is clear that from the
first to the second load phase, complete shear interaction in no
longer verified, with very significant slip strains developing at
the interface between the materials in the second load phase. For
bending moments higher than 25 kNm, there are nowtwo neutral
axes, one lying at the concrete layer and the other lying at the GFRP
profile, whose top flange in nowunder compression. Fromthe first
to the second load phase, for the same bending moment, there
is a significant increase of the axial tensile strains in the bottom
flange of the GFRP profile (between 30% and 40%) and also at the
level of the bottom steel reinforcement (between 78% and 109%),
with the curvature of both materials increasing, specially in the
GFRP profile. In the second load phase, when the bending moment
increases, the slippage between the materials also increasesthe
neutral axis height inthe GFRP profile decreases towards its middle
height, the neutral axis in the concrete layer increases towards its
top surface, which means that the shear interaction between the
materials decreases and the materials are now working more and
more separately.
As previously described, the second local damage was very
similar to the first one, but developed from the central support
section towards section S2. Figs. 16 and 17 show that, from the
second (P2) to the third (P3) load phase, there are no considerable
changes in the pattern of the axial strain distribution of section
S3. However, the slippage between the materials suffers a very
considerable increase and materials at section S3 are now working
almost completely separately. With the bending moment increase,
the neutral axis height in the concrete layer continues to increase,
the neutral axis height in the GFRP profile continues to decrease
and the curvature of both materials continues to increase very
significantly. For a bending moment of 45 kN m, the neutral axis
height in the GFRP profile is about 112 mm, thus, very close to the
webs middle depth, whichcorresponds to the neutral axis position
if both materials had absolutely no shear connection.
Slip strain values at the interface between the materials of
section S3 were estimated based on the axial strain measurements
and are plotted in Fig. 18, as a function of the bending moment.
Results confirm that in the first load phase there is no perceptible
slippage at the interface between the materials. The first local
Fig. 18. Hybrid beam HB7: slip strain vs. moment (section S3).
Fig. 19. Hybrid beam HB7: axial strains measurements at section S5.
damage caused a slip strain to develop, which considerably
increased with the bending moment during the second load phase.
The second local damage also caused a slip strain increase, which
continued to augment in the third load phase up to the final failure.
Measurements of strain gauges 811, plotted in Fig. 16, also
allow us to identify the instant when the top flange of the GFRP
profile started to buckle. In fact, this instant is marked by a sudden
increase of the deformations measured by strain gauges 8 and
10 (placed in the top and bottom surfaces of the top flange,
respectively, but both on the left hand side) and a sudden decrease
of the deformations measured by strain gauges 9 and 11 (placed
on the right hand side of the flange).
4.6.3.2. Section S5. Fig. 19 presents the values of the axial strains
measured at section S5 of hybrid beam HB7 during the three load
phases. Again, in order to facilitate the interpretation of the results,
axial strains are not plotted as a function of the total load or
bending moment, but as a function of time. From the first to the
second load phase, there are no significant changes in the pattern
of the axial strains at the different depths throughout cross section
S5. With the exception of strain gauges 1920 (attached to the
bottom steel reinforcements), which suffered a signal variation,
but presented in both first and second load phases very low strain
levels, all strains kept the same direction of variation, with the
absolute values of the measurements increasing from the first to
the second load phase. This deformation increase was due to the
negative bending moment increase which, as previously referred,
occurred after first failure (for the same total load, negative
1380 J.R. Correia et al. / Engineering Structures 31 (2009) 13691381
Fig. 20. Hybrid beam HB7: axial strains vs. section depth, for increasing bending
moments [kN m] (section S5), comparison of three load phases (P1, P2 and P3).
bending moment over the central support increased between 35%
and 40%). On the contrary, fromthe second to the third load phase,
considerable changes occur on the pattern of the axial strains at
section S5. Fig. 19 shows that strain gauges 27 and 2832
(attached to the lower part of the web and the bottom flange,
respectively), which were initially under compression, maintain
their sign. Strain gauge 26 (attached at middle depth of the web),
although still under compression, presents almost no deformation
in the third load phase. More significantly, the same figure shows
that strain gauges 2124 and 25 (attached to the top flange and
the upper part of the web, respectively), which were initially under
compression, are now in tension. Regarding the strain gauges
placedinthe concrete layer, straingauges 1920(attachedto the
bottom steel reinforcement) continues under tension and strain
gauges 1718 (attached to the top steel reinforcement) change
sign and are now under compression.
In Fig. 20 the axial strains are plotted, for the different load
phases, as a function of the section depth, for increasing values
of the bending moment. This figure confirms that the first local
damage had no significant effect on the axial strain distribution of
section S5, nor on its shear connection. Axial strain distributions
in the first (P1) and second (P2) load phases seem to keep the
same proportionality with the value of the bending moment, with
the deformation increase between the two load phases being a
consequence of the bending moment increase. On the contrary,
Fig. 20 shows that the second local damage had a very significant
effect on the axial strain distribution and composite action of
section S5. It is clear that, from the second (P2) load phase to
the third (P3) load phase, very significant slip strains develop at
the interface between the materials. In the third load phase, there
are now two neutral axes, one lying at the concrete layer and the
other lying at the GFRP profile. With the bending moment increase,
the neutral axis lying at the concrete layer raises and the neutral
axis lying at the GFRP profile lowers, reaching almost its middle
height.
4.6.4. Shear distribution (Section S5)
Distortions at different depths of the web of hybrid beam HB7
measured at section S5, are presented in Fig. 21, as a function of
the total load. During the first load phase, it can be observed that
distortions are not uniformover the sectiondepth, withthe highest
distortional deformation being measured at the level of the top
rosette, as in hybrid beam HB6.
The first local damage caused a decrease of the distortion values
measured at the three rosettes that maintained the same relative
Fig. 21. Hybrid beam HB7: total load vs. distortional deformation (section S5).
Fig. 22. Hybrid beam HB7: shear force carried by the GFRP profile vs. total load
(section S5).
order during the second load phase. The second local damage
caused an increase of the distortion values measured by rosettes
2 and 3, but caused a slight increase in the distortion measured by
rosette 1. During the third load phase, the strain gauges suffered an
inversion of their relative order: the highest distortions started to
be measured at the middle height of the GFRP profile. This result is
in agreement with the previous analysis of the composite action
of cross-section S5, following second local damage, namely, the
fact that materials started working almost independently. In fact,
in a homogeneous cross section with horizontal symmetry, the
maximum shear stresses are observed at middle height.
The distribution of shear force carried by the GFRP profile and
the concrete slab was estimated similarly to hybrid beam HB6, by
averaging the three rosettes measurements. Fig. 22 shows that, in
the first load phase, the GFRP profile carried between 40% and 58%
of shear force at section S5. After the occurrence of the first local
damage, comparing to the first load phase, shear force carried by
the GFRP profile slightly increased (about 4% for the same total
load). Subsequently, during the second load phase, shear force
carried by the GFRP profile at section S5 increased from 58% to
68%. After second local damage occurs, comparing to the second
load phase, shear carried by the GFRP increased very significantly
(about 15% to 20% for the same total load). Finally, prior to
failure, the GFRP profile at section S5 carried about 83% of shear
force.
J.R. Correia et al. / Engineering Structures 31 (2009) 13691381 1381
5. Conclusions
This paper presents an analytical and experimental study on
the flexural behaviour of multi-span GFRP-concrete hybrid beams,
with shear connection provided by a continuous epoxy adhesive
layer. The following main conclusions can be drawn:
1. The proposed methods of analysis for negative bending, in
combination with the previously presented formulation for
positive bending [13], allow predicting the flexural behaviour
of multi-span GFRP-concrete hybrid sections and structural
elements precisely.
2. With regard to deflection calculations for serviceability design,
the precision of the proposed design procedure can be
considerably enhanced if the actual shear distribution factor
(K
s
) is taken into account. For beam HB6, calculated deflections
with K
s
= 1.0 match quite well the experimental results,
because shear contribution to overall deformation of this beam
is rather limited. However, for beamHB7, the consideration of a
unitary K
s
factor is too conservative and the beams deflections
are muchbetter describedif taking into account the actual shear
distribution of approximately 50%.
3. Regarding the ultimate strength, the concrete-adhesive inter-
face failure limited the full exploitation of the strength capacity
of hybrid beamHB7. This type of premature failure mode, asso-
ciated withthe development of shear and peeling stresses at the
extremities of the bonding surfaces [14], causes a loss of com-
posite action between the GFRP profile and the concrete slab. As
a consequence, in hybrid beams with bonded shear connection
systems, it is important to develop anchoring systems to be ap-
plied at the extremity sections, similarly to the procedures and
methods developed for reinforced concrete elements strength-
ened with CFRP systems. One possibility, effective and easy-to-
execute, is the use of steel bolts in the vicinity of the supporting
sections (in addition to the adhesive layer), where maximum
shear and normal stresses occur.
4. Concerning the failure behaviour, although the full plastic
moment at the central support section of hybrid beam HB7
could not be reached with the present design, the redundancy
provided at both the cross-section level (allowing for stress
redistribution in the cross-section, following local damage of
the shear connection system) and at the structural element
level (which allowed for internal forces redistribution) led to
a certain degree of ductility of the structural element, which is
advantageous. However, for structural designs aiming at such
pseudo-ductility through the development of local damages at
the interface, it is also necessary to take into consideration and
prevent the local buckling of the top flange, since it is no longer
fully restrained by the concrete slab.
5. In beams subjected to negative bending moments, for which
the bottom flange of the GFRP profile is under compression,
lateraltorsional buckling may also occur, preventing the
full exploitation of the materials strength capacity. This
phenomenon, observed in hybrid beam HB6, was due to a very
particular structural system (with relatively slender cantilever
spans), not likely to be used in real structural systems made
of GFRP-concrete hybrid beams. If such structural systems are
to be used, adequate lateral bracing constructive measures will
have to be adopted.
Acknowledgments
The authors wish to acknowledge FCT and ICIST for funding the
research and also STEP for supplying the GFRP profiles used in the
experimental investigations.
References
[1] Keller T. Fibre reinforced polymer materials in bridge construction. In: IABSE
symposium, towards a better built environment - innovation, sustainability,
information technology. 2002.
[2] Kharbari V, Seible F. Fibre-reinforced polymer composite materials for civil
infrastructure in the USA. Struct Eng Internat 1999;9(4):2747.
[3] Keller T. Towards structural forms for composite fibre materials. Struct Eng
Internat 1999;9(4):297300.
[4] Burgoyne C. Advanced composites in civil engineering in Europe. Struct Eng
Internat 1999;9(4):26773.
[5] Sobrino JA, Pulido MDG. Towards advanced composite material footbridges.
Struct Eng Internat 1999;9(4):26773.
[6] Keller T. Recent all-composite and hybrid fibre-reinforced polymer bridges
and buildings. Prog Struct Eng Mater 2001;3(2):13240.
[7] Deskovic N, Triantafillou T, Meier U. Innovative design of FRP combined with
concrete: Short-term behaviour. J Struct Eng 1995;121(7):106978.
[8] Hall JE, Mottram JT. Combined FRP reinforcement and permanent formwork
for concrete members. J Compos Construct 1998;2(2):7886.
[9] Canning L, Hollaway L, Thorne AM. An investigation of the composite action
of an FRP/concrete prismatic beam. Construct Building Mater 1999;13(8):
41726.
[10] Fam AZ, Rizkalla SH. Flexural behavior of concrete-filled fiber-reinforced
polymer circular tubes. J Compos Construct 2002;6(2):12331.
[11] Hulatt J, Hollaway L, Thorne AM. The use of advanced composites to form an
economic structural unit. Construct Building Mater 2003;17(1):5568.
[12] Seible F, Karbhari VM, Burgueo R. Kings stormwater channel and I-5/Gilman
bridges, USA. Struct Eng Internat 1999;9(4):2503.
[13] Correia JR, Branco FA, Ferreira JG. Flexural behaviour of GFRP-concrete hybrid
beams with interconnection slip. Composite Struct 2007;77(1):6678.
[14] Correia JR, Branco FA, Ferreira JG. GFRPconcrete hybrid cross-sections for
floors of buildings. Eng Struct. in press doi:10.1016/j.engstruct.2008.04.021.
[15] Keller T, Schaumann E, Valle T. Flexural behavior of a hybrid FRP and
lightweight concrete sandwich bridge deck. Composites Part A 2007;38(3):
87989.
[16] Correia JR. GFRP pultruded profiles in civil engineering: Hybrid solutions,
bonded connections and fire behaviour. Ph.D. thesis. Instituto Superior
Tcnico; 2008.
[17] Bank LC. Composites for construction: Structural design with FRP materials.
Hoboken (NJ): Wiley; 2006.
[18] Correia JR. Glass fibre reinforced polymer (GFRP) pultruded profiles. Structural
behaviour of GFRP-concrete hybrid beams. MSc thesis. Instituto Superior
Tcnico; 2004 [in Portuguese].
[19] Bonaldo E, Barros JOA, Loureno PB. Steel fibre reinforced concrete and CFRP
laminate strips for high effective flexural strengthening Report 05-DEC-E-
14. Universidade do Minho; 2005 [in Portuguese].
[20] Mottram JT. Calculation of the critical buckling load in PFRP shapes.
In: Proceedings of the first international conference ACIC Advanced polymer
composites for structural applications in construction. London: Thomas
Telford; 2002. p. 33745.

Vous aimerez peut-être aussi