Materials and Design 31 (2010) 1130–1147

Contents lists available at ScienceDirect

Materials and Design

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

Experimental and analytical investigation of reinforced high strength concrete

continuous beams strengthened with fiber reinforced polymer
H. Akbarzadeh, A.A. Maghsoudi *
Civil Engineering Department, Kerman University, Kerman, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Carbon and glass fiber reinforced polymer (CFRP and GFRP) are two materials suitable for strengthening
Received 29 June 2009 the reinforced concrete (RC) beams. Although many in situ RC beams are of continuous constructions,
Accepted 22 September 2009 there has been very limited research on the behavior of such beams with externally applied FRP laminate.
Available online 26 September 2009
In addition, most design guidelines were developed for simply supported beams with external FRP lam-
inates. This paper presents an experimental program conducted to study the flexural behavior and redis-
Keywords: tribution in moment of reinforced high strength concrete (RHSC) continuous beams strengthened with
Continuous beams
CFRP and GFRP sheets. Test results showed that with increasing the number of CFRP sheet layers, the ulti-
HSC
FRP
mate strength increases, while the ductility, moment redistribution, and ultimate strain of CFRP sheet
Moment redistribution decrease. Also, by using the GFRP sheet in strengthening the continuous beam reduced loss in ductility
Analytical and moment redistribution but it did not significantly increase ultimate strength of beam. The moment
Nonlinear enhancement ratio of the strengthened continuous beams was significantly higher than the ultimate load
enhancement ratio in the same beam. An analytical model for moment–curvature and load capacity are
developed and used for the tested continuous beams in current and other similar studies. The stress–
strain curves of concrete, steel and FRP were considered as integrity model. Stress–strain model of con-
crete is extended from Oztekin et al.’s model by modifying the ultimate strain. Also, new parameters of
equivalent stress block are obtained for flexural calculation of RHSC beams. Good agreement between
experiment and prediction values is achieved.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Externally bonding fiber reinforced polymer (FRP) sheets with
an epoxy resin is an effective technique for strengthening and
repairing the reinforced concrete (RC) beams under flexural loads.
Consequently, a great amount of research, both experimental and
theoretical, has been conducted on the behavior of FRP strength-
ened RC structures including beams, slabs and columns [1–4]. In
particular, their practical implementation in flexural strengthening
has been numerous [5–13] and resulted in tremendous improve-
ment in their application. The most widely used and studied com-
posites for flexural strengthening the RC simply beams are CFRPs.
Researches concluded that externally bonded FRP could increase
the capacity of RC elements efficiently. A large loss in beam ductil-
ity, however, occurs when CFRP are used for flexural strengthening
of RC beams, because these materials have dissimilar behavior to
that of steel, that is, they exhibit a linear stress–strain behavior
up to failure [7,10,11]. Also, behavior of strengthened RC beam
* Corresponding author. Tel.: +98 3413220054; fax: +98 3413220054.
E-mail addresses: h_akbarzadeh_b@yahoo.com (H. Akbarzadeh), maghsou-
di.a.a@mail.uk.ac.ir (A.A. Maghsoudi).
by GFRP has been studied, As a result, the ductility and stiffness
of CFRP strengthened beams are noticeably lower and higher than
those of GFRP strengthened beams, and all the beams reported are
simply supported having an average concrete compressive
strength of 30 MPa [5,6,8,11,13].
The review of the literature on experimental studies on the
response of RC simply supported strengthened beams with
externally bonded FRP reveals that several different failure modes
were observed. The possible failure mechanisms that observed in
experimental tests are summarized by Teng et al. [2]. Premature
failures such as delaminating FRP and laminate separation can
significantly limit the capacity enhancement and prevent the full
ultimate flexural capacity of the retrofitted beams from being at-
tained. Several studies were conducted to identify ways of prevent-
ing premature failures with a view to improve the load capacity
and ductility of strengthened concrete beams. A number of authors
have recommended the use of steel anchor bolts, steel clamps at
the strip ends and mechanical fasteners for preventing premature
failure of RC beams strengthened with FRP Plates [14–17].
Other researchers studied the use of end anchorage techniques
such as U-strap, L-shape jackets and steel clamp for preventing
premature failure of RC beams strengthened with FRP sheets
[18–20].

0261-3069/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2009.09.041
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1131

Nomenclature

A coefficient Mus ultimate moment at mid-span (sagging region) (kN m)

A0 constant Mexp experimental moment at ultimate load (kN m)
Acon area under stress–strain curve of concrete (mm2) n number of plies of FRP
Af cross-section area of FRP reinforcement (mm2) Py yielding total load of tensile steel at central support (kN)
As total cross-section area of tensile steel (mm2) Pu ultimate applied total load (kN)
A0 s total cross-section area of compressive steel (mm2) p applied predicted load at mid-span (kN)
B coefficient tf thickness of each ply of FRP (mm)
B0 constant x constant
b width of beam section (mm) a mean stress factor for resultant concrete compressive
bf width of FRP (mm) force
C coefficient b moment redistribution ratio
C0 constant c centroid factor indicating position of compressive force
c depth of neutral axis (mm) in concrete
D0 constant ec strain of concrete at compressive extreme fiber (mm/
dc distance from extreme compressive fiber to line of ac- mm)
tion of concrete compressive force (mm) ec strain at area centroid of concrete stress–strain curve
df distance from extreme compressive fiber to centroid of (mm/mm)
FRP (mm) eco concrete strain at the peak stress (mm/mm)
ds effective depth of beam section (mm) ect strain of concrete at tensile extreme fiber (mm/mm)
0
ds distance from extreme compressive fiber to centroid of ecu ultimate strain of concrete (mm/mm)
compressive steel (mm) edb debonding strain in FRP (mm/mm)
Ec modulus of elasticity of concrete (MPa) ef FRP strain (mm/mm)
Ef modulus of elasticity of FRP (MPa) eff failure strain of FRP (mm/mm)
Es modulus of elasticity of steel (MPa) efu ultimate tensile strain of FRP (mm/mm)
fc stress in concrete (MPa) es strain in tensile steel (mm/mm)
fc0 concrete cylinder compressive strength (MPa) e0s strain in compressive steel (mm/mm)
fcu concrete cube compressive strength (MPa) esf failure strain of steel (mm/mm)
ffu ultimate tensile strength of FRP (MPa) esh hardening strain of steel (mm/mm)
f0 constant esu ultimate strain of steel (mm/mm)
fr tensile strength of concrete in flexure (MPa) ey yielding strain of steel (mm/mm)
fs stress in tension steel (MPa) Du mid-span deflection at ultimate load (mm)
fsf steel failure stress (MPa) Dy mid-span deflection at yielding load (mm)
fsu steel ultimate stress (MPa) u curvature of section (mm1)
fy steel yielding stress (MPa) k ratio of the ultimate load of strengthened beams to that
h depth of beam section (mm) of the control beam
k constant lD deflection ductility index
km FRP strain reduction factor n ratio of the yielding load of strengthened beams to that
l clear beam span (mm) of control beam
M predicted resistance moment (kN m) v ratio of the ultimate moment of strengthened beams at
Me elastic moment at ultimate load (kN m) failure time to that of control beam
Muh ultimate moment at central support (hogging region)
(kN m)

Although many in situ RC beams are continuous construction,
there has been very limited research into the behavior of such
beams with external reinforcement [21–24]. In addition, most de-
sign guidelines were developed for simply supported beams with
external FRP laminates [25–27]. Ductility is even more important
for statically indeterminate structures, such as continuous beams,
as it allows for moment redistribution through the rotations of
plastic hinges. Moment redistribution permits the utilization of
the full capacity of more segments of the beam.
Ashour et al. and El-Refaie et al. found out that increasing the
CFRP sheet length to cover the entire hogging or sagging zones
did not prevent the premature failure; further research into the
performance of end anchorage techniques is necessary to minimize
the risk of this mode of failure. Also, they suggested that, strength-
ening both the top surface at central support and beam soffit is the
most effective arrangement of the CFRP laminates to enhance the
beam load capacity [21,22]. Grace et al. investigated the effective-
ness of new triaxially braided ductile fabric in providing ductile
behaviors in RC continuous beams strengthened in flexure. They
concluded that, the beams strengthened with the new fabric
showed greater ductility than those strengthened with the carbon
fiber sheet [23]. However, the concrete used by [21–24] was nor-
mal concrete with an average cube compressive strength of about
35 MPa. In other words, no research report was observed on
strengthening the reinforced continuous beams consist of HSC
and RC continuous beams strengthened with GFRP laminate.
The new developments in concrete technology have led to in-
creased applications of HSC all around the globe. HSC offers many
advantages over conventional concrete. When the strength of con-
crete gets higher, some of its characteristics and engineering prop-
erties become different from those of normal-strength concrete
(NSC) [28,29]. These differences in material properties may have
important consequences in terms of the structural behavior and
design of HSC members [30]. Oztekin et al. obtained new stress–
strain parameters of HSC from experimental stress–strain dia-
grams. Also, they obtained new equivalent stress block parameters
at only ultimate strain not for any given concrete strain [30].These
concretes, with very high compressive strength, can result in less
ductile responses of structural members. The displacement ductil-
ity increases as concrete compressive strength increases for the
1132 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147
same amount of reinforced steel up to some limit and thereafter
decreases as compressive strength of concrete increases [31]. In
this paper, the experimental behavior of five RHSC continuous
(two-span) beams strengthened with externally bonded CFRP
and GFRP sheets along their negative and positive moment regions
are investigated. The beams were loaded with a concentrated load
at the middle of each span. Type of FRP, (CFRP or GFRP), thickness
of CFRP laminates, strengthening the both hogging and sagging re-
gions and end anchorage technique were the main parameters
investigated. During the test, applied load, strain on the concrete
compressive regions and tensile steel at central support and mid-
span, strain at different points of the FRP sheets length and also
the deflection at mid-spans were measured up to failure. The re-
sponses of the strengthened continuous beams were examined
and discussed in terms of deflections, failure modes, strains, mo-
ment and load capacity, moment redistribution and ductility.
Various analytical methods are available to analyze beams
strengthened with bonded FRP laminate. An analytical approach
based on a cross-section analysis is easy and accurate to calculate
the failure load of the strengthened beam. Also, this approach is
applicable for design of the strengthened beam by FRP laminate
[2]. An et al., Malek et al., El-Mihilmy and Tedesco, Chahrour and
Soudki, Toutanji et al. and Rasheed et al. presented analytical pro-
cedures based on the cross-section analysis to calculate the flex-
ural strength of RC simply supported beams bonded with FRP
laminate [6,10,14,32–34]. Elastic model is applied for FRP laminate
behavior in the all reported researches. An et al. have established
basic analytical methodology for evaluating the moment–curva-
ture of reinforced rectangular and T concrete beam sections
strengthened with FRP laminates. Elastic–plastic and linear elas-
tic–perfectly plastic model were used for concrete and steel behav-
ior respectively. Debonding of FRP plate in analytical model was
not considered [6]. Malek et al. has proposed an analysis method
based on linear elastic behavior of the materials for calculating
shear and normal stress concentration at the cutoff point of the
plate [32]. El-Mihily and Tedesco have proposed procedure for
obtaining the ultimate capacity of strengthened beam. Therefore,
they have obtained only ultimate capacity of strengthened beam,
however, ultimate state was verified based on FRP rupture or
crushing of concrete [33]. Rasheed et al. developed an analytical
solution for the load–deflection calculation of FRP strengthened
simple beams at any load stage. The solution assumes a trilinear
moment–curvature response characterized by the flexural crack
initiation, steel yielding, and ultimate capacity. The first stage ex-
tends to the onset point of flexural cracking. The second one fol-
lows until the first yielding of the tension steel. The third line
continues until the limit of concrete useful strain (0.003) or FRP
rupture, depending on the flexural mode involved. Therefore, their
analytical model was not considered the debonding of FRP. Rein-
forcing steel and concrete are assumed to have the classical linear
elastic–perfectly plastic and elastic–plastic response respectively
[34]. Chahrour and Soudki presented an inelastic section analysis
to predict the mid-span moment–curvature response of strength-
ened beam with CFRP laminate. Bilinear and linear elastic–per-
fectly plastic model were used for concrete and steel behavior
respectively. They did not apply any model for debonding of CFRP
laminate in inelastic analysis. But, the analysis is based on the
experimentally measured strains in CFRP laminate at the beam
mid-span [14]. Toutanji et al. present an analytical model for the
moment–deflection calculation of FRP strengthened simple beams
at three points. First and second points are moment and deflection
of cracking and yielding of steel at mid-span and third point is
ultimate state. Debonding of FRP laminate was considered in
analytical model at ultimate state. Elastic–plastic and linear elas-
tic–perfectly plastic were used concrete and steel behavior respec-
tively [10]. Integrity model of reinforcing steel was not used in all
the mentioned literature. As regards, debonding failure occur in
strengthened RC beam with more layer of FRP, therefore, consider-
ing the debonding of FRP laminate in analytical models is neces-
sary. As the mentioned, HSC was used in RC continuous beam;
therefore, offer of the new stress block parameters to determine
the equivalent compressive force of concrete at any given concrete
strain is necessary for cross-section analysis.
Finally, a material nonlinear analytical model was developed to
predict the behavior of FRP strengthened continuous beams. The
developed model is a strength model, which uses the equivalent
stress block to approximate the stress in the compression concrete
at any stage of loading and the moment–curvature model. A new
stress block model is constituted to obtain equivalent stress–strain
parameters for HSC. The stress–strain curves of concrete, steel and
FRP were considered as integrity model. Also, the ACI 440 recom-
mendation was used for FRP debonding strain.
2. Experimental program
2.1. Test specimens
Five large-scale continuous (two-span) beams
(150  250  6000 mm) were tested to failure on a control beam
and four RHSC beams strengthened with externally bonded CFRP
and GFRP sheets on the concrete tension faces. Beams geometry
and reinforcement as well as the loading and support arrangements
are illustrated in Fig. 1. The beams were symmetrically reinforced
with two 16 mm diameter bars at the top and bottom and 10 mm
diameter closed stirrups spaced at 100 mm were provided for
shear. The stirrups and reinforcement ratios are in accordance with
the provision of American Concrete Institute [35]. Thickness of CFRP
laminates, strengthening at both the hogging and sagging region
and end anchorage technique were the main parameters investi-
gated as summarized in Table 1. Thickness and width of each layer
of CFRP sheet were 0.11 mm and 145 mm. Thickness and width of
each layer of GFRP sheet were 0.2 mm and 150 mm. The control
beam, CB, has no strengthened and other beams were strengthened
at both their negative and positive moment regions. The SC1 beam
employed only one layer of CFRP sheet. The SC2 and SC3 beams
used two and three layers of CFRP sheet with three-sided CFRP
straps at the ends of the laminates respectively. The SG3 beam used
three layers of GFRP sheet with three-sided CFRP straps at the ends
of the laminates. The beam SG3 was tested in order to overview the
effectiveness of FRP type on ductility and flexural strength of RHSC
continuous beams. The end anchorage system, consist of two or
three plies of CFRP sheets of 150 mm width was wrapped and
bonded around the sides and the soffit of the concrete beams near
the end of longitudinal CFRP sheets (Fig. 1).
2.2. Material properties
The uniaxial compressive tests on produced concrete
(100  100  100 mm concrete cube) were performed with invari-
able loading rate of 0.15 MPa/s. In these tests, electrical strain
gauge (TML Type) was located on the face of cube specimen in or-
der to measure vertical strain under compressive test. Strains for
each 0.566 MPa stress were recorded and were used to draw
stress–strain curves (Fig. 2). As shown in Fig. 2, HSC specimens
are fractured suddenly and brittle when they reach ultimate stress.
The similar trend for HSC was also reported by Hashemi et al. [36].
The average 28 day tested concrete cube strength (fcu) was
91.3 MPa. The assumed relationship of cylinder strength (f0 c) and
cube strength is (f0 c = 0.85 fcu) thus the average cylinder compres-
sive strength (f0 c) was 77.6 MPa. For each beam six concrete cube
specimens were made at the time of casting and were kept with
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1133

Fig. 1. Test setup and strengthened RC continuous beam details: (a) longitudinal profile of beam, (b) typical cross section of beam in sagging region, (c) typical cross section of
beam in hogging region and (d) end anchorage system.

Table 1
Details of the test specimens.

Beam no. f0 c (MPa) Type of FRP Positive moment strengthening Negative moment strengthening End anchorage
No. of layers Strengthened length (m) No. of layers Strengthened length (m)
CB 74.2 – 0 2.2 0 1.8 None
SC1 74.6 CFRP 1 1 None
SC2 74.1 CFRP 2 2 Yes
SC3 74.4 CFRP 3 3 Yes
SG3 79.7 GFRP 3 3 Yes

Specimen 1 Two bars of diameter 16 mm were tested in tensile and the

measured yield strength was 412.5 MPa, and maximum tensile
Specimen 2
strength was 626.4 MPa. The modulus of elasticity of steel bars
120 Specimen 3 was 2  105 MPa. The Young’s modulus (Efu) and ultimate tensile
Specimen 4 stress (ffu) of the CFRP and GFRP sheet and the properties of epoxies
100
Specimen 5 used for bonding the FRP sheets were obtained from the supplier
and given in Tables 2 and 3.
stress (MPa)

80 Average

60
2.3. FRP bonding procedure
40

20 The process of applying FRP sheet to concrete; involved surface

0 and resin over coating. After the beams surface preparation, a
0 500 1000 1500 2000 2500 3000 3500 4000
Strain (microstrain)
Fig. 2. Stress–strain diagrams for high strength concrete.
the beams during curing. The average concrete compressive
strength (f0 c) for each beam is shown in Table 1.
preparation, priming, resin under coating, FRP sheet application,
two-part primer was applied to the prepared concrete surface.
Next, a two-part epoxy resin was applied to the primed concrete
surface, followed by application of the FRP sheet. The FRP sheet
was installed over the concrete surface by starting at one end
and moving along the length of the FRP sheet until completed. Fi-
nally, a resin over coating was applied over the FRP sheet. Concrete
beams strengthened with FRP sheets were cured for at least seven
days at room temperature before testing.
1134 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147

Table 2 along the beam appear in Fig. 3. Hydraulic actuator was used to
Mechanical properties of the FRP sheets. load the beam and the load was measured by using a load cell.
Material Density Thickness Ultimate Young’s Ultimate The reaction of the beam at the central support was also measured
(kg/cm3) (mm) tensile stress, modulus, strain, by using another load cell. Electrical resistance disposable strain
ffu (MPa) Ef (MPa) efu (%) gauges were pasted on the FRP sheets and on internal reinforcing
CFRP 1.81 0.11 3800 242,000 1.55 bars at different locations. The demec (stainless steel disc) and
GFRP 2.57 0.2 2250 73,000 3.1 electrical strain gauges were also attached along the height of
beams at the mid-spans and the central support to measure the
concrete strains. The mid-span deflections were measured using
Table 3 linear variable differential transformers (LVDTs). The load was ap-
Mechanical properties of the bonding adhesive. plied step-by-step up to failure in a load control manner of test
Material Density Compression Tensile Young’s Shear
beams. The strain gauges, LVDTs, and the load cells readings were
(kg/cm3) strength strength modulus strength recorded at each load increment using data-logging equipment. At
(MPa) (MPa) (MPa) (MPa) the end of each load increment, observations, measurements,
Epoxy resin 1.11 97.4 76.1 3600 54.8 cracks development and propagation on the beam surfaces were
adhesive recorded.
Epoxy resin 1.77 >90 >25 12,800 >15
primer
3. Test results and discussions

2.4. Instrumentation and test procedure
Each test beam, which comprised two equal spans of 2850 mm
each, was loaded with a concentrated load at the middle of each
span (Fig. 1). The various monitoring devices and their location
The beams were loaded with a concentrated load at the middle
of each span. The obtained experimental results are presented and
discussed subsequently in terms of the observed mode of failure,
load–deflection, load–strain, moment and load capacity, moment
retribution and ductility.

Fig. 3. Test setup and monitoring devices.

 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1135

3.1. Failure mode and load–deflections responses
Three different failure modes were observed for tested beams
(Table 4) and are described as follows.
The control beam, CB, failed in the RC conventional flexural
manner. The tensile steel yielded (Py = 105 kN) prior to concrete
crushing at both the central support (Pu = 162 kN) and mid-span
section as shown in Fig. 4. The wide flexural cracks were occurred
at mid-span (wcrmax ¼ 5:9 mm) and central support (wcrmax ¼
6:0 mm). These cracks are well extended to the compressive
regions.
The tensile steel at central support of beam SC1 yielded at a load
of 110 kN. The beam failed by tensile rupture of the CFRP sheet
over the central support (190.6 kN), (Fig. 5). The rupture of CFRP
sheet was sudden and accompanied by a loud noise indicating a ra-
pid release of energy and a total loss of load capacity. Fig. 5 also
indicated that by strengthening the RHSC beam using CFRP sheet,
the beam flexural capacity is increased and therefore more number
of flexural cracks are occurred and developed towards the neutral
axis before beam failure.
The tensile steel at central support of beam SC2 yielded at a load
of 124.6 kN. The load was applied further and the beam was failed
suddenly at a load of 219.3 kN by intermediate crack (IC) debond-
ing of CFRP sheet at hogging region and rupture of end strap at
hogging region, as shown in Fig. 6a, and b. The end U-strap proved
to be effective in limiting end debond, but not intermediate crack
debonding, as shown in Fig. 6b.
The tensile steel at central support of beam SC3 yielded at a load
of 136.0 kN. The beam failed by IC debonding at hogging region
(Fig. 7). The end U-straps did not rupture because three layers of
U-wrap were used for end strap.
The tensile steel at central support of beam SG3 yielded at a
load of 157.3 kN. The beam failed by IC debonding at hogging re-
gion (Fig. 8). As the SC3 beam, the end U-straps did not rupture.
The total applied loads versus deflection at mid-span section of
the beams are shown in Fig. 9. All strengthened beams exhibited
three stage responses up to failure; representing the concrete pre-
cracking stage, concrete postcracking to tension steel preyield
stage, tension steel postyield stage to failure. The control beam
showed a rough bilinear precracking and postcracking behavior
with a yield plateau in the postyield stage. In the uncracked elastic
stage, the same behavior was observed for all tested beams, indi-
cating very similar beams stiffness prior to concrete cracking. In
the cracked preyield stage, the stiffness and yield load of the FRP
strengthened beams were moderately larger than that of the con-
trol beam. However, significant decreases in beams stiffness was
observed after yielding the tensile steel at sections of negative
and positive moments, but by increasing the number of CFRP lay-

Table 4
Experimental results of tested beams including yield and ultimate load, deflection and FRPs ultimate strain.

Beam Failure mod Py n Deflection Pu k Deflection Ultimate strain l

no. (kN) at yielding (kN) at failure of CFRP & (Du/Dy)
load (Dy) (mm) load (Du) (mm) GFRP (lm)
CB Flexural failure 105 1 7.5 162 1 77.4 – 10.32
SC1 rupture of top CFRP 110 1.05 7.5 190.6 1.18 26 10,000 3.47
SC2 IC debonding of FRP sheet and rupture 124.6 1.18 8 219.3 1.35 23 7030 2.87
of end strap at hogging region
SC3 IC debonding at hogging region 136 1.29 8.9 259.3 1.6 19.88 8922 2.23
SG3 IC debonding at hogging region 157.3 1.49 9 222.6 1.37 26.6 12,922 2.96

Fig. 4. Conventional ductile failure of control beam indicating: (a) concrete crushing at central support and (b) concrete crushing at mid-span.
1136 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147

Fig. 5. Rupture of CFRP sheet at central support in SC1 beam.

Fig. 6. Failure modes observed in SC2 beam: (a) IC debonding at negative moment and (b) rupture of end U-strap at hogging region.

Fig. 7. IC debonding of CFRP sheet at negative moment in SC3 beam.

Fig. 8. IC debonding of GFRP sheet at negative moment in SG3 beam.

ers the loss in beams stiffness are reduced. As indicated in Fig. 9, CFRP sheets (see Table 2). Therefore, to increase beam stiffness
the stiffness of SG3 is less than SC3 beam after yielding the tensile strengthened by GFRP sheet a larger cross sections of GFRP are
steel, it is because the elastic modulus of GFRP is much lower than needed. Increasing the number of CFRP layers decreased the mid-
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1137

CB the tension strain in FRP sheet at the central support and mid-span
300 and a negative strain value indicates the extreme concrete com-
SC1
Total Applied Load (kN)

250 SC2 pressive strain. Fig. 10 indicates that each curve consists of almost
SC3 three straight lines of different slopes. The first turning point (A)
200 SG3 indicates the cracking of concrete in tension zone. The second turn-
150 ing point (B) refers to the yielding the tension steel and the third
turning point (C) refer to the ultimate load.
100 The increase in concrete strain for the beams strengthened with
50 CFRP at low load is almost the same. But concrete strain for the
beams strengthened with GFRP is larger than beams strengthened
0 with CFRP at the same value of applied load; it is because the elas-
0 10 20 30 40 50 60 70 80 90 100 110 120 130 tic modulus of GFRP is about one third of CFRP sheet. The extreme
Mid-Span Deflection (mm) concrete compressive strain fiber of the strengthened beams, with
Fig. 9. Total applied load deflection responses of test beams.
increasing the number of FRP layers remains more or less linear up
to beams failure and is not significantly affected by concrete crack-
ing or yielding of the tensile steel. These results demonstrate that
span deflection and generally increased the beams stiffness at the the effect of the strengthening laminate is to reduce strain in the
same value of applied load. Comparing the obtained results of concrete compressive fibers. Thus, externally bonded FRP laminate
RHSC continuous beams strengthened with CFRP and GFRP sheets, may also be beneficially used to reduce concrete compressive
it was found that the mid-span deflection was increased while stresses, in addition to acting as additional tensile reinforcement.
using the GFRP sheets. The FRP sheets tensile strains increased significantly after concrete
cracking. In addition, the strain of the FRP sheets rapidly increased
after yielding the internal tensile steel reinforcement. Not only
3.2. Load–strain response increasing the number of CFRP layers reduced the tensile strains
of CFRP sheets at a given value of the applied load, but it also de-
The total applied load versus CFRP, GFRP sheets and extreme creased the maximum tensile strains in the CFRP sheets achieved
concrete compressive strain at mid-span and central support of before beam failure (Table 4). Also, it can be seen that using the
beams are plotted in Fig. 10. The positive strain value represents GFRP sheet for strengthening the continuous beams increased

Fig. 10. Total applied load versus FRP and concrete strain at: (a) central support and (b) mid-span.
1138 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147
Fig. 11. Total applied load versus FRPs sheet strain and adjacent steel bars for beams SC1 and SG3.
the failure strain of FRP. The maximum recorded strain values of
CFRP and GFRP sheets are given in Table 4.
The total applied load versus CFRP and GFRP strain and tensile
steel at mid-span and central support for SC1 and SG3 beams are
plotted in Fig. 11. It gives a comparison between tensile strains
in FRP sheets and the adjacent steel bars for different load levels.
At early stages of loading, it can be observed that the tensile steel
and FRP strains are essentially the same, indicating effective com-
posite action of different materials. As the load is approached the
yielding, the strains in tensile steel were increased more rapidly
than those in FRP sheets at central support. This is because the
FRP sheets had begun to debond from the concrete surface nearby
cracks.
The variations in longitudinal strain of FRPs sheet at different
total load levels of strengthened beams are plotted in Fig. 12. The
strain drops from the maximum value under the load point and
central support to a zero at the end of FRP sheet. As shown in
Fig. 12, before yielding the steel, the FRPs sheet strains for sagging
and hogging region show a linear strain distribution in the bonded
system and, after yielding the steel, the gradient of FRPs sheet
strains are increased by increasing the portion of load on the FRPs
sheets. Therefore, when the FRPs sheet strain in the central support
exceeds about 7000–13,000 l for beams strengthened by CFRP and
GFRP sheet respectively, the debonding failure occurred. In other
words these values are considered as debonding strains. In beams
SC2, SC3 and SG3, after the occurrence of partial debonding of FRPs
sheets, (due to effect of the anchorage system), the beams behavior
was changed. Therefore, the strain in FRPs sheets was rapidly in-
creased at end length of FRPs sheet (Fig. 12b–d).
3.3. Enhancement of failure load and moment capacity
Table 4 summarizes the ultimate failure load, Pu (i.e., the sum of
the two mid-span point loads at failure), the ultimate load
enhancement ratio (k), which is the ratio of the ultimate load of
an externally strengthened beam to that of the control beam, yield-
ing load of tensile steel at central support (Py) and the yielding load
enhancement ratio (n), which is the ratio of yielding load of the
strengthened beam to that of control beam. As indicated in Table 4,
the addition of one, two or three layers of CFRP sheet and three lay-
ers of GFRP sheet causes to increase the ultimate load capacity by
18%, 35%, 60% and 37% respectively for beams SC1, SC2, SC3 and
SG3 compared to the control beam. Comparing the results of
beams strengthened with CFRP and GFRP sheet, it is found that,
while an increase in beam ultimate load is required, larger cross
sections of GFRP must be used. Also, the increase in the number
of CFRP layers, the yielding load of tensile steel at central support
is slightly increased.
Total applied load versus the hogging and sagging bending mo-
ments for tested beams are plotted in Fig. 13. The bending moment
was calculated by equilibrium considerations of the beams using
the measured central support reaction and mid-span applied load.
The behavior of all beams at low load levels was essentially elastic.
For beams CB, SC1 and SG3 as the applied load was increased, the
bending moment tended to become nonlinear. However, the
behavior of beams SC2, SC3 was elastic until those failed at sagging
region. In general, for beams strengthened by CFRP sheet, by
increasing the number of CFRP layers, the experimental bending
moment values are close to the elastic prediction.
Table 5 presents the failure moment, and the ultimate moment
enhancement ratio, v, which is the ratio of ultimate moment of
strengthened sections at time failure (central support and mid-span
sections) to that of unstrengthened sections. As shown in Table 5,
the addition of one, two or three layers of CFRP and GFRP sheet in-
creases the ultimate moment capacity by 28%, 55%, 88% and 57% at
central support and by 12%, 26%, 47% and 28% at mid-span, for
beams SC1, SC2, SC3 and SG3 respectively, while compared to the
control beam, CB. Table 5 indicated that using the more amount
of GFRP did not led to increase the moment capacity of the strength-
ened beams at hogging and sagging region as found when compar-
ing the results of beams SC3 and SG3. In general, all the
strengthened sections resisted a higher moment than the corre-
sponding unstrengthened sections of the control beam. By compar-
ing the ultimate load enhancement ratio of a strengthened beam
and the moment enhancement ratio of a strengthened section at
central support in the same beam (Table 5), it can be concluded that
the latter was significantly higher than the former (such conclusion
was previously reported by Ashour et al. and El-Refaie et al. [21,22].
It is also remained that, such conclusion is not valid for simply sup-
ported beams strengthened with external reinforcement where the
moment and load enhancement ratios are always the same.
3.4. Ductility
Ductility is more important for statically indeterminate struc-
tures, such as continuous beams, as it allows for moment redistri-
bution through the rotations of plastic hinges. Since FRP repair is a
fairly new innovation and also the HSC behavior is like a brittle
material, therefore, understanding the effect of such materials on
the ductility of the RC beams is critical [36,37].
The limited available investigations on continuous beams
strengthened in flexure with external reinforcement, suggested
that the deflection ductility index, lD, seems to be a good measure
H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1139

Fig. 12. FRPs sheet strain distribution at different load levels.

1140 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147

350

Total Applied Load (kN)

300

250
Elastic
200
CB
150 SC1
SC2
100 SC3
SG3
50

0
80 60 40 20 0 20 40 60 80
Hogging Bending Moment (kNm) Sagging Bending Moment (kNm)

Fig. 13. Total applied load versus bending moment of tested beams.

Table 5
Central support reaction, failure moments and moment redistribution of tested beams.

Beam Pu Central support Central support Mid-span

no. (kN) reaction,
Experimental v Failure moment b (%) Experimental v Failure moment b (%)
(R kN)
failure moment based on elastic failure moment based on elastic
(kN m) analysis (kN m) (kN m) analysis (kN m)
CB 162 106.5 36.33 1 43.28 16.06 39.54 1 36.07 9.62
SC1 190.6 128.1 46.74 1.28 50.93 8.22 44.53 1.12 42.44 4.92
SC2 219.3 149.3 56.5 1.55 58.59 3.57 49.87 1.26 48.83 2.13
SC3 259.3 177.4 68.26 1.88 69.28 1.51 58.14 1.47 57.73 0.71
SG3 222.6 151.4 57.2 1.57 59.47 3.81 50.72 1.28 49.56 2.34

of the ductility [21–23]. Therefore, the ductility index is obtained
based on deflection computation, and is defined by
Du
lD ¼
Dy
The mid-span deflection at beam ultimate load (Du) and yield-
ing load (Dy), and the deflection ductility index (lD) are given in
Table 4. As can be seen from Table 4, increasing the number of
CFRP layers led to decrease the mid-span deflection at beam ulti-
mate load, but the mid-span deflection at yielding load is almost
constant. Therefore, increasing the number of CFRP sheet layers
was found to decrease the beam ductility. But use of more amount
of GFRP in strengthening the beam SG3 did not led to decrease the
ductility because modulus of elasticity and ultimate strain of GFRP
is low and high respectively.
For unstrengthened HSC members, displacement ductility in-
dex, lD in the range of 3–5 is considered imperative for adequate
ductility, especially in the areas of seismic design and the redistri-
bution of moments [37]. Therefore, assuming that an index value of
3 represents an acceptable lower bound to ensuring the ductile
behavior of RHSC continuous beams strengthened with FRP sheet,
it appears that, for the tested beams SC2, SC3 would not meet that
requirement (lD < 3).
3.5. Moment redistribution
The moment redistribution ratio (b) given in Table 5 was calcu-
lated for the sagging and the hogging bending moment at mid-
span and at the central support at failure load. The ratio was calcu-
lated by:
M e  M exp
b¼  100%
Me
where Me is the value of the failure moment at central support and
mid-span based on the elastic analysis and Mexp is the experimental
value of bending moment. As indicated in Table 5, beam CB had a
moment redistribution ratio of 16.06% at central support and
9.62% at mid-span. The moment redistribution ratio of strength-
ened beams was significantly decreased due to increase the number
ð1Þ of CFRP layers. The beams SC1, SC2, SC3 and SG3 had the moment
redistribution ratio of 8.22%, 3.57%, 1.51% and 3.81% at central sup-
port and 4.92%, 2.13%, 0.71% and 2.34% at mid-span respec-
tively. By comparing the results of beams SC3 and SG3, it was
found that, using more amount of GFRP did not cause to decrease
the moment redistribution ratio.
4. Analytical model
Various analytical methods are available to analyze beams
strengthened with bonded FRP laminate. An analytical approach
based on a cross-section analysis is easy and accurate to calculate
the failure load of the strengthened beam. Also, this approach is
applicable for design of the strengthened beam by FRP laminate
[2]. Therefore, an iterative analytical model was developed to pre-
dict the flexural behavior of the RHSC beams strengthened with
FRP composites. This analytical model (Fig. 14) uses the principles
of strain compatibility and equilibrium, and the material constitu-
tive relations for the HSC, steel and FRP to predict flexural behav-
ior. The assumptions include: (1) Linear strain distribution
throughout the full depth of the section. (2) No slip between the
longitudinal reinforcing steel and the surrounding concrete. (3)
No slip between the externally bonded FRP sheets and concrete.
(4) Failure of the beam occurs when either the compressive strain
in the concrete reaches ultimate strain or the tensile strain in the
FRP composites reaches its ultimate strain. Fig. 15 presents the
flexural response of strengthened beam, based on the mentioned
ð2Þ model. The moment–curvature response is idealized as a nonlinear
curve divided into three zones.
The elastic–plastic models for concrete and steel reinforcement
and elastic model for FRP are applied. As following, model of con-
crete, reinforced steel and FRP are described.
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1141

Fig. 14. Strain, stresses and forces used in the moment–curvature model.

Fig. 16. FRP stress–strain model.

Fig. 15. Moment–curvature.

4.1. Stress–strain model for FRP

In strengthening RC beams with FRP laminates, different failure

modes have been observed. Specific failures of FRP laminate are
rupture and delamination. In order to predict delamination failure,
Guidelines and recommendations suggested different models
[2,25–27]. ACI 440 is universal guideline for design of externally
strengthened RC structures with FRP laminate. Fig. 17. Tensile steel stress–strain model.
FRP materials are assumed to behave linear elasticity until fail-
ure (Fig. 16). If the beam fails by FRP rupture, failure strain of FRP the stress–strain relationship for the reinforcement is divided into
eff is the FRP ultimate strain efu from the composite tension test. three regions [38]:
eff ¼ efu ð3Þ
1. Elastic
If the failure is not due to FRP rupture but delamination, the fail-
fs ¼ Es es 0 6 es 6 ey ð6Þ
ure strain of FRP eff for determining the ultimate capacity can be
estimated by the ACI 440 equation [25]. This guideline suggest
model of delamination prediction based on thickness, modulus of 1. Yield plateau
elasticity and ultimate tensile strength of FRP laminate.
fs ¼ fy ey 6 es 6 esh ð7Þ
eff ¼ edb ¼ km efu ð4Þ
The reduction factor km for ultimate strain efu is 1. Strain hardening
8   fs ¼ yfo þ fy
> nEf t f
< 601e 1  360;000 6 0:9 nEf t f 6 180; 000 0 0 2
A xþB x
km ¼
fu
  ð5Þ y ¼ 1þC 0
xþD0 x2
esh 6 es < esf
>
: 601e 90;000 6 0:9 nEf t f P 180; 000 es esh ð8Þ
fu nE t
f f x¼ eo
fo ¼ fsu  fy
4.2. Stress–strain model for steel reinforcement
4.3. Stress–strain model for concrete
Different model of stress–strain for steel reinforcement as linear
elastic–perfectly plastic and bilinear are applied in analytical mod- The distribution of concrete stresses in the compression zone is
els. Using the integrity model of steel reinforcement increase the found from the stress–strain curve of concrete. Stress–strain model
accuracy of analytical verifications. Therefore, referring to Fig. 17, produced by Hognestad is one of the most commonly used for or-
1142 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147

Z ec
dinary concrete that the initial part of the concrete compression
Acon ¼ fc ðec Þdec ¼ afc ec ð13Þ
stress–strain relationship is represented by a second order para- 0
bolic expression up to the maximum concrete stress and the falling
R ec
f ðe Þde
branch is linear [39]. Park and Paulay model is similar to the Hog- a¼ 0 c c c ð14Þ
fc ec
nestad model for concrete in compression, except that Park and
Paulay fix eco as 0.002 in their model [40]. Evaluating the right side of Eq. (14), results in the following val-
High performance concrete specimens are fractured suddenly ues for a
and brittle when they reach ultimate stress under uniaxial com-
ec ð2kec þ 2ec þ 3kecu Þ
pression. Thus, to define the falling branch of stress–strain curve a¼ ð15Þ
is very difficult (Fig. 2). It is assumed that ultimate strain equals 6e2cu
strain in the maximum stress (eco = ecu). Oztekin et al. [30] pro- The position of concrete compressive force Cc, measured from
duced Eq. (9) for stress–strain relationship of high performance the top fiber of concrete, is expressed in terms of the parameter
concrete. c, shown in Fig. 14 and calculated as follows:
 2 ! dc ¼ c  c ð16Þ
ec ec
fc ¼ fc0 k  ðk  1Þ ð9Þ
ecu ecu where dc = distance from top concrete fiber to line of action of con-
crete compressive force.
where The first moment of area under the actual concrete stress–strain
curve is given by:
k ¼ 2  ½ðfc0  40Þ=70 60 MPa 6 fc0 6 94 MPa ð10Þ
Z ec
Q¼ fc ðec Þec dec ¼ ec Acon ð17Þ
ecu ¼ ½2:2 þ 0:015ðfc0  40Þ  103 60 MPa 6 fc0 6 94 MPa ð11Þ 0

According to experimental results (Fig. 2), the value of ecu is as- where ec ¼ strain at centroid of area under stress–strain curve.
sumed to be equal to 0.003 (i.e., similar to suggested value by ACI The strain ec can be defined in terms of ec by:
08). Therefore, proposed model for behavior of compression con- ec ¼ ð1  cÞec ð18Þ
crete in this study was established based on Oztekin et al. model
[30], but only ultimate strain of concrete was modified. Park and and therefore
Paulay, Oztekin et al., proposed model and experimental stress– Z ec
strain curves are compared shown in Fig. 18. It is clear that the pro- Q ¼ ec Acon ¼ ð1  cÞec fc ðec Þdec ð19Þ
posed model is too close to the experimental result. 0
It is known that real stress distribution in compressive area at a The parameter c (centroid factor) is obtained by equating Eqs.
cross section is the same as the stress–strain curve in uniaxial com- (17) and (19).
pression. But stress–strain curve is affected by different variables.
R ec
Therefore, it is not easy to suggest a certain stress–strain curve 0
ec fc ðec Þdec
for concrete. Perhaps the area of stress distribution and its centre c¼1 R ec ð20Þ
ec f ðe Þdec
0 c c
of gravity are more important than the geometry of the stress dis-
tribution for equilibrium equation in RC design. Therefore, in this Evaluating Eq. (20), results in the following values for c
study, a new stress block model is constituted to obtain equivalent
0:5kec þ 0:5ec þ kecu
stress–strain parameters for HSC. For any given concrete strain in c¼ ð21Þ
extreme compression fiber ec, the concrete compression force Cc 2kec þ 2ec þ 3kecu
is expressed in terms of a parameter a, defined as follows:
4.4. Calculation of resisting moment and curvature
C c ¼ afc0 bc ð12Þ
The parameter a (stress factor) is used to convert the actual For tested beams, the strain and stress in the FRP laminate, steel
concrete stress into an equivalent rectangular stress block. This is rebar, and concrete at any section are calculated by using an incre-
usually calculated by equating the area of under stress–strain mental deformation technique described in the following. Fig. 14
curve to an equivalent rectangular area; shows the strain, stress and the force diagram for a typical rectan-
gular beam with a composite laminate bonded to the tension face.
The strain in the extreme compressive fiber of concrete at central
90 support or mid-span (ec) is increased until failure is reached. Next,
strains in the tensile and compression steel, composite plate and
80
extreme tensile fiber of concrete are calculated in terms of ec from
70
the following equations:
Stress (MPa)

60
0
50 c  ds
e0s ¼  ec ð22Þ
40
Experimental c
Oztekin d c
30
Park and paulay
es ¼ s  ec ð23Þ
c
20 Proposed Model d c
10 ef ¼ f  ec ð24Þ
c
0 hc
0 500 1000 1500 2000 2500 3000 3500 ect ¼  ec ð25Þ
c
Strain (micron)
Each of the corresponding internal forces can be determined by
Fig. 18. Strain–stress curves of concrete. multiplying the stress by their cross-sectional areas.
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1143

The curvature at central support or mid-span is calculated by A ¼ afc0 b ð36Þ

dividing the concrete strain ec by a distance to the neutral axis B ¼ ec ðA0s Es þ Af Ef Þ  As fy ð37Þ
depth, c; 0
C ¼ ec ðA0s Es ds þ Af Ef df Þ ð38Þ
ec      
/¼ ð26Þ h h 0 h
c M ¼ afc0 bc  cc þ A0s Es e0s  ds þ As fy ds 
2 2 2
 
h
The internal resisting moment (M) of the section can be ob- þ Af Ef ef df  ð39Þ
tained by taking the sum of the moments about the middle height 2
of the section. The location of the neutral axis, c, is obtained from
the equilibrium of internal forces for different state. These states 4.4.4. Strain hardening of steel
in all the mentioned literature in introduction due to use the linear Here, the constant parameters (A, B, C) and the resisting mo-
elastic–perfectly plastic model for steel include (a) before cracking ment are determined by:
of concrete; (b) after cracking of concrete and before yielding of
steel; (c) After yielding [6,10,14,32–34]. But in this paper, because A ¼ afc0 b ð40Þ
the integrity model for steel are used, therefore, state of steel strain B¼e 0
c ðAs Es þ Af Ef Þ  As fs ð41Þ
hardening is added. 0
C¼  c ðA0s Es ds þ Af Ef df Þ
e ð42Þ
     
4.4.1. Before cracking of concrete h h 0 h
Mu ¼ afc0 bc  cc þ A0s Es e0s  d s þ As f s d s 
Before the concrete has cracked or the tensile steel has yielded, 2 2 2
 
the neutral axis (c) is determined by: h
þ Af Ef ef df  ð43Þ
2
2
Ac þ Bc þ C ¼ 0 ð27Þ In this state, the value of fs is determined by Eq. (8), and if com-
pression steel is yielded, the amount of its stress is fy.
Where the constant parameters are as follows:
4.5. Moment–curvature curve of tested beams
b
A ¼ afc0 b  Ec ec ð28Þ
2 For any given strain in the extreme compression fiber of con-
B ¼ ec ðA0s Es þ As Es þ Af Ef þ bhEc Þ ð29Þ crete, the moment and curvature relationship can be found (Eqs.
  (21)–(43)). Therefore the entire moment–curvature relationship
0 b 2
C ¼ ec A0s Es ds þ As Es ds þ Af Ef df þ Ec h ð30Þ is found by incrementally increasing the strain in the extreme
2
compression concrete fiber until the strain in one of the material
components exceeds the ultimate strength of that material (ef P eff
The tensile strength of concrete in flexure
pffiffiffi(MPa)
ffi and modulus
pffiffiffiffiof or ec P ecu).
elasticity of HSC are assumed as fr ¼ 0:6 fc0 and Ec ¼ 3200 fc0 þ A computer program was created to predict the moment–curva-
6900 (MPa) respectively. ture diagram for cross section of RHSC strengthened with FRP. The
The uncracked resisting moment in this state is determined by: plots were generated by increasing the concrete strain in incre-
      ments of 0.00001 mm/mm. Analytical and experimental mo-
h h 0 h ment–curvature response at sagging and hogging sections for
M ¼ afc0 bc  cc þ A0s Es e0s  ds þ As Es es ds 
2 2 2 tested beams are shown in Figs. 19–23. The three zones of pre-
 
h 1 cracking, postcracking preyield, and postyield zones were repro-
þ Af Ef ef df  þ ect Ec bðh  cÞ2 ð31Þ
2 3 duced in the analytical responses of the simulated beams. Good
agreement between the predicted and experimental moment–cur-
vature is clear, which verifies the reported test results.
4.4.2. After cracking of concrete and before yielding of steel
Once the tensile stress in the extreme tensile fiber of concrete
4.6. Implementation of analytical model
exceeds the fr value, the concrete in tension cracks and therefore
all concrete below the neutral axis is neglected. The neutral axis
The objective of implementing the analytical model was to
in this state is determined by the same Eq. (27), but constant
automate the analytical theories to develop first the moment–cur-
parameters (A, B, C) and the cracked resisting moment are deter-
vature diagrams, and then the failure load for continuous beams
mined by:
strengthened with FRP. To calculate the failure load based on mo-
ment–curvature diagrams, the beams are divided in two groups.
A ¼ afc0 b ð32Þ
0 CB
B¼e c ðAs Es þ As E s þ Af E f Þ ð33Þ 45
0 40
C ¼ ec ðA0s Es ds þ As Es ds þ Af Ef df Þ ð34Þ 35
Moment (kN.m)

      30
h h 0 h
M ¼ afc0 bc  cc þ A0s Es e0s  ds þ As Es es ds  25
2 2 2 20 Analytical
  15 Exp
h
þ Af Ef ef df  ð35Þ 10
2 5
0
0 20 40 60 80 100 120
4.4.3. After yielding and before strain hardening of steel Curvature (με/mm)
The constant parameters (A, B, C) and the resisting moment for
this state are determined by: Fig. 19. Analytical versus experimental moment–curvature response for beam CB.
1144 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147

50 SC1 in this group. Therefore, for these beams; only at one section
(mid-span or central support section) reached their flexural capac-
40 ity at failure time. Based on experimental results (Fig. 13), the
Moment (kN.m)

beams behavior is very close to elastic state. Accordingly, from

30
the principle of elasticity, the applied point load, p, at the beam
Analytical mid-span is calculated by:
20
Exp-Hogging
10 16 M uh
Exp-Sagging p¼ ð44Þ
3 l
0
The beams in second group are ductile; i.e., at both or one of the
0 10 20 30 40 50 60 70
mid-span and central support sections have high ductility and
Curvature (με/mm)
reached their flexural capacities at failure time. Therefore, the
Fig. 20. Analytical versus experimental moment–curvature response for beam SC1. beams were strengthened in mid-span or central support and both
the mid-span and central support with low amount of FRP are eval-
uated in this group. Accordingly, from the principle of virtual work,
the applied point load, p, at the beam mid-span is calculated by:
SC2
70
2
60 p¼ ðMuh þ 2M us Þ ð45Þ
l
Moment (kN.m)

50
Muh and Mus are ultimate moment at hogging and sagging re-
40 gion respectively.
Analytical
30 Table 6 presents analytical moment capacity at mid-span and
Exp-Hogging
20 central support sections for all tested beams. The total analytical
Exp-Sagging
10 load capacity (Pu(anal) = 2p) is obtained by Eqs. (44), (45) and the ra-
tio of Pu(anal)/Pu(exp), is between experimental and theoretical load
0
0 10 20 30 40 50 60 70 capacities. Analytical load of the beam SC1 was calculated based
Curvature (με/mm) on Eq. (45) as the moment redistribution value for this beam was
8.22%. It is clear that, both the mid-span and central support sec-
Fig. 21. Analytical versus experimental moment–curvature response for beam SC2. tion reached their ultimate flexural capacities at failure time. The
results of Table 6 are shown that the proposed model predicts
the moment capacity of continuous HSC beams strengthened with
SC3 FRP very well. Also, Table 6 and Fig. 24 show a very good agree-
80 ment between flexural load capacity of proposed method and the
70 experimental results.
Moment (kN.m)

60
50 4.7. Verification of the proposed model using other studies
40
Analytical
30 To testify the proposed model, available experimental data on
Exp-Hogging
20 the flexural behavior of RC continuous beams strengthened by
Exp-Sagging
10
Table 6
0
Comparisons between experimental and predicted moment and load.
0 10 20 30 40 50 60 70
Curvature (με/mm) Beam Muh Mus Group Pu(anal) Pu(exp) Pu(anal)/
no. (kN m) (kN m) (kN) Pu(exp)
Fig. 22. Analytical versus experimental moment–curvature response for beam SC3.
CB 35.18 35.18 2 148.12 162 0.914
SC1 46.56 46.56 2 196.03 190.60 1.026
SC2 58.65 – 1 219.53 219.30 1.002
SC3 69.87 – 1 261.51 259.30 1.009
SG3 SG3 55.78 – 1 215.47 222.60 0.968
70
60
Moment (kN.m)

50
300
40
Analytical Experimental
30 250
Total Applied Load (kN)

Analytical
Exp-Hogging
20
Exp-Sagging 200
10
0 150
0 10 20 30 40 50 60 70
100
Curvature (με//mm)
50
Fig. 23. Analytical versus experimental moment–curvature response for beam SG3.

0
The beams in first group are not ductile, i.e., both the mid-span
CB SC1 SC2 SC3 SG3
and central support sections neither have high ductility nor mo- beams
ment redistribution. Strengthened beams in both the positive and
negative moment regions with high amount of FRP are evaluated Fig. 24. Comparison between experimental and predicted total applied load.
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147 1145

Table 7
Details of RC continuous beams tested by Ashour et al. and Grace et al. [21–23].

Designation Reference Arrangement b l h Main longitudinal f0 c (MPa) n tf bf Ef efu

of strengthening (mm) (mm) (mm) steel (mm) (mm) (MPa) (%)
Top Bottom
H1 Ashour et al. (2004) – 150 3830 250 2u8 2u20 20.4 – – – – –
H2 Hogging region 37 2 0.117 110 240,000 1.625
H3 Hogging region 28.05 6 0.117 110 240,000 1.625
H4 Hogging region 28.22 10 0.117 110 240,000 1.625
H5 Hogging region 39.1 6 0.117 110 240,000 1.625
H6 Hogging and sagging region 37.4 2 0.117 110 240,000 1.625
S1 – 2u20 2u8 22.1 – – – – –
S2 Sagging region 36.46 2 0.117 110 240,000 1.625
S3 Sagging region 28.30 6 0.117 110 240,000 1.625
S4 Sagging region 36.38 6 0.117 110 240,000 1.625
S5 Sagging region 20.74 10 0.117 110 240,000 1.625
E1 – 2u16 2u16 20.4 – – – – –
E2 Hogging region 37.06 1 1.2 100 150,000 1.67
E3 Sagging region 40.63 1 1.2 100 150,000 1.67
E4 Hogging and sagging region 39.2 1 1.2 100 150,000 1.67
E5 Hogging region 38 6 0.117 110 240,000 1.625
CB Grace et al. (2004) – 152 1981 254 2u16 2u16 41.5 – – – – –
F-CTC Hogging and sagging region 2 0.13 152 218,000 1.2
(on both sides as a U-wrap)

Table 8
Comparisons between experimental and predicted moment and load [21–23].

Beam no. Experimental Analytical Pu(anal)/Pu(exp)

Muh (kN m) Mus (kN m) Pu (kN) Muh (kN m) Mus (kN m) Group Pu (kN)
H1 11.21 56.78 138 11.69 50.98 2 118.71 0.860
H2 31.6 61 152.3 32.91 56.69 2 152.79 1.003
H3 46.48 59.56 172.9 48.33 54.32 2 163.94 0.948
H4 53.07 51.32 162.6 47.94 54.37 2 163.65 1.006
H5 35 64.27 162.6 49.98 57.13 2 171.54 1.055
H6 28.26 70.2 172.9 32.94 66.08 2 172.43 0.997
S1 57.77 11.13 83.6 51.90 11.98 2 79.22 0.947
S2 71.28 22.67 121.8 56.58 32.80 2 127.60 1.048
S3 66.9 24.72 121.8 54.40 48.38 2 157.87 1.296
S4 88.97 37.15 170.5 56.56 49.66 2 162.81 0.954
S5 50.18 28.36 111.7 51.18 47.02 2 151.67 1.358
E1 54.49 44.41 149.7 40.55 40.55 2 127.04 0.848
E2 79.78 45.64 178.6 76.22 42.00 2 167.33 0.937
E3 53.56 72.35 207 42.27 76.57 2 204.08 0.986
E4 77 72.29 231.4 76.37 – 1 212.70 0.919
E5 77.42 44.87 174.6 79.47 42.07 2 170.87 0.979
CB 40.61 42.59 254 38.10 38.10 2 230.82 0.909
F-CTC 64.4 59.43 370 57.94 57.94 2 350.99 0.949

400
Total Applied Load (kN)

350 Experimental
300 Analytical
250
200
150
100
50
0
1

E1

E2

E3

E4

E5
S1

S2

S3

S4

S5

TC
B
H

C
F-

Continuous beam s

Fig. 25. Comparison between experimental and predicted total applied load for beams studied in [21–23].

FRP are selected. The details of tested beams are summarized in Ta- (H6 and E4), all of the beams were strengthened at negative or po-
ble 7. Ashour et al. studied the strengthening effect of CFRP sheet sitive moment region. Grace et al. studied the strengthening effect
and plate on RC continuous beams [21,22]. Except two beams of CFRP sheet and new triaxially braided ductile fabric on RC con-
1146 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147
tinuous beams [23]. Beams CB unstrengthened and beam F-CTC
that strengthened along their negative and positive moment re-
gions around the top/bottom face extending 152 mm on both sides
as a U-wrap are selected for verification.
The analytical model was applied to obtain nonlinear moment–
curvature curves at central support and mid-span sections, and
load capacity of tested beams. Forasmuch as normal concrete
was used in mentioned beams [21–23], concrete stress–strain
model produced by Park and Paulay was used.
The comparison between experimental and predicted moment
at central support and mid-span sections and total applied load
are presented in Table 8. Analytical load of beam H6 was calculated
based on Eq. (45) as its ductility value was 3.46. Due to use of U-
wrap for strengthening the beam of F-CTC, its moment redistribu-
tion value was obtained 6.5%, therefore, analytical load of beam F-
CTC was calculated based on Eq. (45). Also, the graphical represen-
tation of the predicted and experimental applied loads, are shown
in Fig. 25. The comparisons indicate that the proposed model
agrees very well with the experimental moment and load capacity,
justifying the use of proposed model, as can be seen in Table 8 and
Fig. 25.
5. Conclusions
The following conclusions can be drawn based on the test re-
sults and verification of proposed analytical model:
1. Increasing the number of CFRP sheet layers will change the fail-
ure mode from tensile rupture to IC debonding of CFRP sheets in
RHSC continuous beams. The end U-strap proved to be effective
in limiting end debonding, but not IC debonding.
2. Increasing the number of CFRP layers reduced loss in stiffness of
strengthened beams after yielding the tensile steel. The stiff-
ness of the beam strengthened with GFRP is less than the beams
strengthened with CFRP after yielding the tensile steel.
3. The extreme compressive strain of concrete fiber in the
strengthened continuous beams remains more or less linear
up to failure, by increasing the number of CFRP layers. The con-
crete strain for the beam strengthened with GFRP is larger than
beams strengthened with CFRP at the same value of applied
load.
4. Increasing the number of CFRP layers increased the ultimate
load capacity, but not significantly the yield load. Larger cross
sections of GFRP must be used for increasing the load capacity
in the beams strengthened with GFRP.
5. The moment enhancement ratio at central support of strength-
ened continuous beam was significantly higher than the ulti-
mate load enhancement ratio in the same beam.
6. Increasing the number of CFRP layers significantly decreased
the moment redistribution ratio from 16.06% to 1.51% in the
unstrengthened and strengthened beams respectively. Using
the more amount of GFRP did not lead to decrease significantly
the moment redistribution ratio.
7. The deflection ductility index was decreased by increasing the
number of CFRP layers. Using the more amount of GFRP in
strengthening the continuous beam did not led to decrease sig-
nificantly the ductility index.
8. A nonlinear moment–curvature model and a method for calcu-
lating the load capacity is developed to evaluated FRP strength-
ened RHSC and RNSC continuous beams by FRP and verified by
results from this study and other similar experiments. Compar-
isons between experimental and predicted moment and load
capacity show that the proposed model agrees very well with
the test results, thus justifying the use of the proposed model
for HSC and NSC in strengthened beams.
Acknowledgment
The authors gratefully acknowledge the support of the Iran Na-
tional Science Foundation (INSF).
References
[1] Meier U, Deuring M, Meier H, Schwegler G. Strengthening of structures with
advanced composites. Alternative materials for reinforcement and prestressing
of concrete. Glasgow, Scotland: J.L. Clarke/Chapman & Hall; 1993.
[2] Teng JG, Chen JF, Smith ST, Lam L. FRP strengthened RC structures. New
York: Wiley; 2002.
[3] Foret G, Limam O. Experimental and numerical analysis of RC two-way slabs
strengthened with NSM CFRP rods. Construct Build Mater 2008;22:2025–30.
[4] Lignola GP, Prota A, Manfredi G, Experimental CosenzaE. Experimental
performance of RC hollow columns confined with CFRP. J Compos Construct,
ASCE 2007;11(1):42–9.
[5] Saadatmanesh H, Ehsani M. RC beams strengthened with GFRP plates. I:
experimental study. J Struct Eng, ASCE 1991;117(11):3417–33.
[6] An W, Saadatmanesh H, Ehsani MR. RC beams strengthened with FRP plates.
Part II: analysis and parametric study. J Struct Eng, ASCE
1991;117(11):3434–55.
[7] Spadea G, Swamy RN, Bencardino F. Strength and ductility of RC beams
repaired with bonded CFRP laminates. J Bridge Eng 2001;6(5):349–55.
[8] Grace NF, Sayed GA, Soliman AK, Saleh KR. Strengthening reinforced concrete
beams using fiber reinforced polymer (FRP) laminates. ACI Struct J
1999;96(5):865–74.
[9] Rabinovitch O, Frostig Y. Experiments and analytical comparison of RC beams
strengthened with CFRP composites. Composites: Part B 2003;34:663–77.
[10] Toutanji H, Zhao L, Zhang Y. Flexural behavior of reinforced concrete beams
externally strengthened with CFRP sheets bonded with an inorganic matrix.
Eng Struct 2006;28:557–66.
[11] Thomsen H, Spacone E, Limkatanyu S, Camata G. Failure mode analyses of
reinforced concrete beams strengthened in flexure with externally bonded
fiber-reinforced polymers. J Compos Construct, ASCE 2004;March/
April:123–31.
[12] Wang YC, Hsu K. Design recommendations for the strengthening of reinforced
concrete beams with externally bonded composite plates. Compos Struct
2009;88(2):323–32.
[13] Li L, Guo Y, Liu F. Test analysis for FRC beams strengthened with externally
bonded FRP sheets. Construct Build Mater 2008;22:315–23.
[14] Chahrour A, Soudki K. Flexural response of reinforced concrete beams
strengthened with end-anchored partially bonded carbon fiber-reinforced
polymer strips. J Compos Construct, ASCE 2005;9(2):170–7.
[15] Garden HN, Hollaway LC. An experimental study of the influence of plate end
anchorage of carbon fibre composite plates used to strengthen reinforced
concrete beams. Compos Struct 1998;42:175–88.
[16] Lamanna AJ, Bank LC, Scott DW. Flexural strengthening of reinforced concrete
beams using fasteners and fiber-reinforced polymer strips. ACI Struct J
2001;98(3):368–76.
[17] Bank LC, Arora D. Analysis of RC beams strengthened with mechanically
fastened FRP (MF-FRP) strips. Compos Struct 2006;79:180–91.
[18] Spadea G, Bencardino F, Swamy RN. Structural behavior of composite RC
beams with externally bonded CFRP. J Compos Construct 1998;2(3):132–7.
[19] Pham H, Al-Mahaidi R. Prediction models for debonding failure loads of carbon
fiber reinforced polymer retrofitted reinforced concrete beams. J Compos
Construct, ASCE 2006;10(1):48–59.
[20] Xiong GJ, Jiang X, Liu JW, Chen L. A way for preventing tension delamination of
concrete cover in midspan of FRP strengthened beams. Construct Build Mater
2007;21:402–8.
[21] Ashour AF, El-Refaie SA, Garrity SW. Flexural strengthening of RC continuous
beams using CFRP laminates. Cem Concr Compos 2004;26:765–75.
[22] El-Refaie SA, Ashour AF, Garrity SW. Sagging and hogging strengthening of
continuous reinforced concrete beams using carbon fiber-reinforced polymer
sheets. ACI Struct J 2003;100(4):446–53.
[23] Grace NF, Ragheb WF, Abdel-Sayed G. Strengthening of cantilever and
continuous beams using new triaxially braided ductile fabric. ACI Struct J
2004;101(2):237–44.
[24] Liu IS T, Oehlers DJ, Seracino R. Moment redistribution in FRP and steel-plated
reinforced concrete beams. J Compos Construct, ASCE 2006;10(2):115–24.
[25] ACI. Guide for the design and construction of externally bonded FRP systems
for strengthening concrete structures. American Concrete Institute, ACI
440.2R-02, Farmington Hills, Mich, 2002.
[26] JSCE. Recommendations for the upgrading of concrete structures with use of
continuous fiber sheets. Jpn Soc Civil Eng, Concr Eng, Series 41, Tokyo; 2001.
[27] fib. Externally bonded FRP reinforcement for RC structures. Fédération
Internacionale du béton, Task Group 9.3 FRP; 2001.
[28] ACI. State-of-the-art report on high-strength concrete. American Concrete
Institute, ACI 363R, Detroit; 1992.
[29] Rashid MA, Mansur MA, Paramasivam P. Correlations between mechanical
properties of high-strength concrete. J Mater Civ Eng 2002;14(3):230–8.
[30] Oztekin E, Pul S, Husem M. Determination of rectangular stress block
parameters for high performance concrete. Eng Struct 2003;25:371–6.
 H. Akbarzadeh, A.A. Maghsoudi / Materials and Design 31 (2010) 1130–1147
[31] Ashour SA. Effect of compressive strength and tensile reinforcement ratio on
flexural behavior of high-strength concrete beams. Eng Struct 22: 413-
423.
[32] Malek AM, Saadatmanesh H Ehsani MR. Prediction of failure load of R/C beams
strengthened with FRP plate due to stress concentration at the plate end. ACI
Struct J 1998;95(2):142–52.
[33] El-Mihilmy MT Tedesco JW. Analysis of reinforced concrete beams
strengthened with FRP laminates. J Struct Eng, ASCE 2000;126(6):684–91.
[34] Rasheed HA, Charkas H Melhem H. Simplified nonlinear analysis of
strengthened concrete beams based on a rigorous approach. J Struct Eng,
ASCE 2004;130(7):1087–96.
1147
[35] ACI. Building code requirements for reinforced concrete. American Concrete
Institute, ACI 318, Farmington Hills, Michigan; 2008.
[36] Hashemi H, Maghsoudi AA, Rahgozar R. Flexural ductility of reinforced HSC
beams strengthened with CFRP sheets. Struct Eng Mech 2008;30(4).
[37] Maghsoudi AA, Akbarzadeh H. Flexural ductility of HSC members. Struct Eng
Mech 2006;24(2).
[38] Wang PI, Shah SP. High strength concrete in ultimate strength design. J Struct
Div, ASCE 1978;104(11):1761–73.
[39] Hognestad E, Hanson N W McHenry D. Concrete stress distribution in ultimate
strength design. ACI J Proc 1955;52(12):455–79.
[40] Park R, Paulay T. Reinf Concr Struct. New York: John Wiley and Sons; 1975.