Construction and Building Materials 96 (2015) 518–529

Contents lists available at ScienceDirect

Construction and Building Materials

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

Structural performance and serviceability of concrete beams reinforced

with hybrid (GFRP and steel) bars
Ahmed El Refai a,⇑, Farid Abed b, Abdullah Al-Rahmani b
a
Department of Civil and Water Engineering, Laval University, Québec City, Québec G1V 0A6, Canada
b
Civil Engineering Department, American University of Sharjah, Sharjah, United Arab Emirates

h i g h l i g h t s

Flexural response of hybrid-reinforced concrete beams was investigated.

Parameters included the reinforcement ratio and the ratio of steel to GFRP bars.

Code equations were assessed against the experimental test results.

New bond coefficient proposed to predict the crack width of the hybrid beams.

New deformability factor proposed to assess the deformability of the hybrid beams.

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

Article history: This paper reports on the structural performance of concrete beams reinforced with hybrid reinforce-
Received 18 April 2015 ment. Six concrete beams reinforced with a combination of steel and glass fiber-reinforced polymer
Received in revised form 26 July 2015 (GFRP) bars and three other beams reinforced with only GFRP bars were tested in flexure. Over-
Accepted 9 August 2015
reinforced hybrid beams showed higher strength and ductility than their GFRP-reinforced counterparts.
The CSA-S806-12 equation accurately predicted the deflections of the hybrid-reinforced beams with high
effective reinforcement ratios. Based on the test results, a bond coefficient was proposed to predict the
Keywords:
crack width of the hybrid-reinforced beams using the ACI-440.1R-06 equation. A modified deformability
Composite materials
Concrete beams
factor was also utilized to assess the deformability of the hybrid-reinforced beams. Comparison between
Cracking the experimental and predicted results showed the adequacy of the models used in predicting the load-
Deformation carrying capacity, deflection, crack widths, and deformability of hybrid-reinforced concrete beams.
Ductility Ó 2015 Elsevier Ltd. All rights reserved.
Fiber reinforced polymer
Glass fiber
Hybrid methods
Reinforced concrete
Serviceability

1. Introduction
Fiber-reinforced polymers (FRPs) have been widely used as
reinforcing materials in the last decades. Due to their anti-
corrosive characteristics, FRP bars are becoming very promising
alternatives to conventional steel bars in reinforcing concrete
structures. However, one of the main disadvantages of FRP bars
is their brittleness. FRP materials exhibit linear elastic behavior
up to failure, which adversely affects the ductility of the concrete
structure and limits its inelastic response. The failure modes of
⇑ Corresponding author.
E-mail addresses: ahmed.elrefai@gci.ulaval.ca (A. El Refai), fabed@aus.edu
(F. Abed).
FRP-reinforced structures vary widely with the amount of the rein-
forcement used. A low amount of FRP reinforcement leads to the
rupture of the bars prior to concrete crushing. When high rein-
forcement ratios are used, concrete in compression crushes, while
tensile stresses in FRP bars remain below their ultimate strength.
Most design codes and guides call for over-reinforcing FRP-
reinforced structures to ensure plastic deformation of the com-
pressed concrete and to enhance ductility.
In addition to their lack of ductility, FRP bars are known by their
low modulus of elasticity as compared with steel bars. As a result,
the FRP-reinforced structure suffers excessive deflections and wide
cracks that affect its serviceability. In this case, design of FRP-
reinforced structures should be governed by their serviceability
limit state rather than their ultimate limit state. Therefore, the

http://dx.doi.org/10.1016/j.conbuildmat.2015.08.063
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529 519

Nomenclature

Af area of GFRP reinforcement nf ratio of modulus of elasticity of GFRP bars to modulus of

As area of steel reinforcement elasticity of concrete
a distance between the support and the point load (shear ns ratio of modulus of elasticity of steel bars to modulus of
span) elasticity of concrete
b width of cross section Pa applied load
c distance from extreme fiber in compression to neutral Pu ultimate load
axis R ratio of axial stiffness of steel bars to axial stiffness of
d distance from extreme fiber in compression to center of GFRP bars
reinforcement s spacing of reinforcing bars
dc thickness of concrete cover from the tension face to the w maximum crack width
center of the closest bar b ratio of the distance between the neutral axis and the
DF deformability factor tension face to the distance between the neutral axis
DF mod modified deformability factor and the centroid of reinforcement
ff tensile stress in GFRP bars b1 ratio of depth of equivalent rectangular stress block to
ffu ultimate tensile stress in GFRP bars depth of the neutral axis
fs tensile stress in steel reinforcement bd reduction coefficient as given in Eq. (9)
fy yield stress in steel reinforcement Dm maximum deflection at midspan of the beam
Ec modulus of elasticity of concrete ecu maximum concrete compressive strain (0.003 for ACI-
Ef modulus of elasticity of GFRP bars 318-08 provisions)
Es modulus of elasticity of steel bars ef tensile strain in GFRP bars
0
fc concrete compressive strength es tensile strain in steel bars
Icr cracked moment of inertia esu ultimate tensile strain in steel bars
Ie effective moment of inertia ey yield strain in steel bars
Ig gross moment of inertia g coefficient given in Eq. (11)
k coefficient = c/d qeff effective reinforcement ratio in hybrid sections given by
kb bond coefficient Eq. (6)
L beam length qf GFRP reinforcement ratio
Lg distance from the support to the point where M = Mcr qfb balanced reinforcement ratio
Ma applied moment at the critical section qs steel reinforcement ratio
M cr cracking moment ws curvature at service moment
Ms service moment wu curvature at ultimate moment
Mu ultimate moment wy curvature at yield moment
My yielding moment

concept of combining steel bars with FRP bars (hybrid system) in
reinforcing concrete structures seems to be a practical solution to
overcome the ductility and serviceability problems of purely FRP-
reinforced structures. This approach of using hybrid reinforcement
in concrete elements has gained interest in the last decades. In a
hybrid system, the addition of steel reinforcing bars ensures the
ductility of the structure and enhances its serviceability, whereas
the FRP bars maintains its load-carrying capacity. Near-surface-
mounted (NSM) technique is one form of hybrid construction in
which FRP bars are placed near the tensile surface to strengthen
steel-reinforced concrete elements. Research studies conducted
on NSM hybrid reinforcement showed its effectiveness in restoring
the strength and serviceability of the concrete elements [17].
However, the use of hybrid system in reinforcing new concrete
structures is relatively new.
In their experimental work, Aiello and Ombres [1] carried out
flexural tests on hybrid concrete beams reinforced with a combina-
tion of aramid FRP (AFRP) and steel bars. Steel and AFRP bars were
placed either at the same level or at different levels in the tensile
zone. It was reported that the addition of steel reinforcing bars
to heavily AFRP-reinforced concrete sections significantly
enhanced the ductility and reduced the crack widths and spacing.
However, the contribution of added steel reinforcement to the flex-
ural capacity did not exceed 15% in over-reinforced hybrid beams.
An increase in stiffness was reported for hybrid beams with steel
bars placed above the AFRP bars.
Leung and Balendran [14] investigated the flexural response of
hybrid concrete beams reinforced with glass FRP (GFRP) and steel
bars placed at different levels. The authors reported that the
hybrid-reinforced beams had higher flexure strength than the
steel- or GFRP-reinforced beams. Over-reinforced hybrid beams
failed by concrete crushing. The test results showed that the stiff-
ness of hybrid-reinforced beams increased after the steel bars had
yielded, indicating that the GFRP bars became more effective at
this stage.
Qu et al. [15] carried out an experimental and analytical inves-
tigation on six hybrid-reinforced beams. The amount of reinforce-
ment and the ratio of GFRP to steel bars were the main parameters
investigated. The test results showed that the use of steel rein-
forcement in combination with GFRP bars improved the ductility
of the hybrid-reinforced beams. Beams with higher reinforcement
ratios showed higher load-carrying capacity than the other beams.
Lau and Pam [13] reported similar results after testing twelve
steel-, FRP-, and hybrid-reinforced concrete beams. The latter
beams behaved in a more ductile manner when compared with
the flexure behavior of FRP-reinforced beams. Ductility improve-
ment was more pronounced in over-reinforced FRP beams than
in their under-reinforced or balanced-reinforced counterparts.
Recently, Safan [16] investigated both experimentally and ana-
lytically the structural behavior of twelve concrete beams rein-
forced with hybrid (GFRP and steel) bars arranged at different
levels, with the GFRP bars placed at the outer layers of the tensile
zone. All hybrid-reinforced beams failed due to concrete crushing
after yielding of steel reinforcement. The authors reported that
GFRP bars were effective in maintaining the flexure capacity of
the beams and in enhancing their serviceability aspects. This
520 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529

finding was more pronounced in beams having lower steel rein- 2.2. Midspan deflection
forcement ratios with steel bars placed at upper layers of the ten-
sile zone. The concept of using an effective moment of inertia, Ie, after
The current study aims at investigating the flexural behavior of cracking is adopted by most of the codes and guides to reflect
hybrid-reinforced concrete beams. It consists of an analytical part the continuous change in the beam’s stiffness as the applied load
that is verified experimentally through a series of flexural beam increases. ACI-318-08 [2] and CSA-A23.3-04 [10] codes recom-
tests. The main test parameters include the reinforcement ratio mend the use of Eq. (7) [8] to estimate Ie, which is then used to
and the ratio of steel to GFRP bars used to reinforce the test beams. determine the deflection of steel-reinforced beams at service loads.
The study also aims at evaluating the most common design code
3
equations used to predict the response of the hybrid-reinforced M cr
Ie ¼ Icr þ ðIg  Icr Þ 6 Ig ð7Þ
beams. Design models for predicting flexural strength, deflection, Ma
crack width, and deformability of the beams are presented. Previous studies concluded that Branson’s equation overesti-
Serviceability and ultimate aspects of the hybrid beams are the mated the effective moment of inertia of FRP-reinforced beams,
key parameters of this study. especially when the beams were under-reinforced [18].
Therefore, the ACI 440.1R-06 [3] committee introduced a reduction
2. Design approach coefficient, bd, in Branson’s equation to account for the reduced
tension stiffening in FRP-reinforced members. According to ACI
2.1. Flexural strength 440.1R-06 [3], the effective moment of inertia, Ie, is calculated as
follows:
In this study, the design of hybrid sections is based on the
3
Mcr
assumption that steel reinforcing bars yield before the ultimate Ie ¼ Icr þ ðbd Ig  Icr Þ 6 Ig ð8Þ
Ma
tensile stress in GFRP bars is reached. Beams of different reinforce-
ment ratios and having different ratios of steel to GFRP bars were where
designed in accordance to the provisions of the ACI-318-08 code !
[2]. An iterative process was adopted to predict two possible 1 qf
bd ¼ 6 1:0 ð9Þ
modes of failure, namely; (a) yielding of steel bars prior to the rup- 5 qfb
ture of GFRP bars and (b) crushing of concrete at ultimate before
Bischoff [4–6] and Bischoff and Scanlon [7] suggested changes
yielding of steel bars or rupture of GFRP bars.
to Eq. (8) to improve the incorporation of the tension stiffening
In general, the depth of the neutral axis, c, was determined from
in deflection calculations for both steel- and FRP-reinforced beams
the equilibrium of compression and tension forces as given in Eq.
as follows:
(1). Equilibrium of compression and tension forces for both modes
of failure resulted in the quadratic equations (2) and (3), Icr
Ie ¼  2 6 Ig ð10Þ
respectively, which were solved mathematically for the depth of
1g Mcr
the neutral axis, c. In both cases, strains at the reinforcement level Ma

were determined from the stress–strain relationship of each

where
material. The ultimate moment, Mu, was therefore determined
using Eq. (4). Icr
g¼1 ð11Þ
0 Ig
0:85f c b1 cb ¼ As f s þ Af f f ð1Þ
For beams tested under four-point loadings, the midspan
0:85f c bb1 c2 þ ðAf Ef ecu  As f y Þc  ðAf Ef ecu dÞ ¼ 0
0 deflection is determined after cracking using Eq. (12) as follows:
ð2Þ
Pa a
Dm ¼ ð3L2  4a2 Þ ð12Þ
0:85f c bb1 c2 þ ðAf Ef ecu þ As Es ecu Þc  ðAf Ef ecu þ As Es ecu Þd ¼ 0
0
ð3Þ 48Ec Ie

 On the other hand, the CSA-S806-12 [9] code recommends the
b c
M u ¼ ðAs f s þ Af f f Þ d  1 ð4Þ use of the closed-form relationship given in Eq. (13) to determine
2
the maximum deflection of a beam loaded in four-point loading
For FRP-reinforced beams, the balanced reinforcement ratio configuration. The rigidity of the beam is taken as EcIcr.
refers to a simultaneous rupture of FRP bars and concrete crushing. "
3 #
Pa L3 a a3 Lg
The balanced ratio was calculated from the force equilibrium and Dm ¼ 3 4  8g ð13Þ
strain compatibility assumptions as follows: 48Ec Icr L L L

Ef ecu
0
fc The parameter Lg represents the distance from the support to
qfb ¼ 0:85b1 ð5Þ
f fu Ef ecu þ f fu the point where M = Mcr and is determined as given in Eq. (14):

The effective reinforcement ratio, qeff, for hybrid-reinforced M cr

Lg ¼ a ð14Þ
beams was determined by Eq. (6). This ratio was compared to Ma
the balanced ratio of the FRP reinforcement, qfb, to define the In Eqs. (7)–(13), the cracked moment of inertia, Icr, is deter-
expected failure mode in each hybrid beam. If qeff > qfb , the beam mined from the elastic cracked section analysis as given in Eq. (15):
is over-reinforced and the flexural failure is expected to occur due
1
to concrete crushing. If qeff < qfb , the beam is under-reinforced and
3 2 2
Icr ¼ bðkdÞ þ ðnf Af þ ns As Þd ð1  kÞ ð15Þ
3
steel bars are expected to yield prior to concrete crushing.
where
f
As f y þ Af fy qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qeff ¼ fu
¼ qs þ qf ð6Þ k¼ ðnf qf þ ns qs Þ2 þ 2ðnf qf þ ns qs Þ  ðnf qf þ ns qs Þ ð16Þ
bd f fu
 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529 521

Ef Es 3.3. Instrumentation and test setup

nf ¼ ; ns ¼ ð17Þ
Ec Ec
in this study, Eqs. (7)–(17) that represent four different approaches
were used to predict the midspan deflections for the FRP- and
hybrid-reinforced concrete beams.
3. Experimental program
Nine reinforced concrete beams designed as flexural-critical specimens were
tested under a four-point loading setup. Beams were labeled according to the num-
ber, type, and size of their longitudinal tensile reinforcement. The letter G denoted
the GFRP reinforcement used, whereas the letter S referred to the steel reinforce-
ment. Accordingly, beam 2G12-2S10 refers to a hybrid beam reinforced with 2
GFRP bars of 12 mm diameter combined with 2 steel bars of 10 mm diameter.
3.1. Test specimens
Schematics of the test specimen and cross-sections are illustrated in Fig. 1. All
beams were 4000 mm long with rectangular cross sections of 230  300 mm. The
beams were reinforced with GFRP or hybrid (GFRP and steel) bars arranged in
one layer in the tensile region. The beams were also reinforced with 10 mm diam-
eter closed steel stirrups spaced at 100 mm along the shear spans and at 150 mm
between the point loads. Two longitudinal steel bars of 8 mm diameter were placed
in the compression zone and served as hangers for the stirrups. All beams were ini-
tially designed with clear concrete cover of 40 mm measured from the beam soffit.
However, fabrication errors resulted in variations in the final concrete covers of
some specimens. After testing, actual concrete covers were measured and were
used later in all calculations.
Table 1 summarizes the experimental test program. Specimens were divided
into three groups according to their reinforcement arrangement. Group A consisted
of three GFRP-reinforced beams with qf = 0.38%, 0.64%, and 1.12%. Beam 2G12
(qf = 0.38%) served as a control specimen for the hybrid beams of group B. Group
B consisted of three beams reinforced with two 12 mm diameter GFRP bars in com-
bination of one 10 mm, two 10 mm, and two 12 mm diameter steel bars. Specimens
of group C were reinforced with three 16 mm diameter GFRP bars in combination
with two 10 mm, two 12 mm, and two 16 mm diameter steel bars.
The effective reinforcement ratios calculated using Eq. (6) are given in Table 1. It
can be noticed that the addition of steel bars to beam 2G12 caused the beams of
group B to be slightly over-reinforced, with qeff less than 1.4 qfb. The ratio qeff/qfb
for beams of group B ranged between 1.04 (for beam 2G12-1S10) and 1.36 (for
beam 2G12-2S12). On the other hand, the three beams of group C were highly
over-reinforced after the addition of steel bars to the two 16 mm GFRP bars. The
ratio of qeff/qfb for beams of group C ranged between 1.74 (for beam 2G16-2S10)
and 2.32 (for beam 2G16-2S16).
3.2. Material properties
Beams were fabricated and cast using ready-mixed concrete of average cylinder
compressive strength of 40 MPa. Concrete was produced using Type II Ordinary
Portland Cement, dune sand, and limestone coarse aggregate with a nominal max-
imum aggregate size of 20 mm. Silica fume and ground granulated blast furnace
slag were used as partial replacement of cement in the concrete mixture. Yield
and ultimate stresses of the longitudinal steel reinforcement were 520 MPa and
650 MPa, respectively. The GFRP bars had nominal tensile strength of 1000 MPa
and modulus of elasticity of 50 GPa, as provided by the manufacturer.
One side of each beam was instrumented at midspan section with 60 mm long
strain gauges spaced at 25, 75, and 125 mm from the top surface of concrete. Strain
gauges of 5 mm length were also mounted on the longitudinal bars at midspan to
monitor the strain variation during loading. Crack widths were measured by means
of crack transducers of 100 mm gauge length attached on the side of the beams at
the level of the longitudinal reinforcement. For each beam, the load was initiated
until the first crack appeared. A crack transducer was then installed at this location
to measure the crack’s growth throughout the test. However, the visual inspection
of the side surface of the beams prior to testing revealed the existence of vertical
flexural cracks formed at different depths in some specimens, probably during
the beam handling. For these specimens, crack transducers were placed at the loca-
tion of these cracks.
The nine beams were tested under a four-point test configuration with a clear
span of 3700 mm. A steel spreader beam was used to create two point loads spaced
at 1200 mm. The applied load was monitored by means of a load cell located on the
top of the spreader beam. A linear variable displacement transducer (LVDT) was
placed at the soffit of the beams to measure the deflections at midspan. The read-
ings of load cell, strain gauges, and crack and displacement transducers were
recorded and stored in a 30-channel data acquisition system at a rate of 5 read-
ings/s.
4. Test results and analysis
4.1. Flexural response
Cracks were initiated in the maximum moment region below or
between the point loads and propagated upward toward the com-
pressed concrete zone. As the load increased, more cracks appeared
along the beam and the cracks in the shear spans acquired inclina-
tion toward the central zone. Failure occurred when one or more
cracks extended to the upper concrete fibers in the maximum
moment zone, leading to concrete crushing. The under-reinforced
beam 2G12 failed by rupture of the GFRP bars as predicted,
whereas the over-reinforced beams 3G12 and 3G16 failed by con-
crete crushing. The hybrid beams, on the other hand, failed by con-
crete crushing that followed yielding of steel bars. None of the
GFRP bars ruptured in hybrid-reinforced beams except beam
2G12-1S10 that exhibited a catastrophic failure after both steel
and GFRP bars ruptured simultaneously with the concrete crush-
ing. Fig. 2 shows the cracking patterns of the test beams upon
failure.
Fig. 3 shows the moment–deflection curves of the test beams.
Initially, all beams exhibited a linear moment–deflection relation-
ship until cracking occurred. The slope of the uncracked portion of
the curve slightly varied from one beam to another depending on
the cracking condition before testing. After cracking, all beams suf-
fered reduced stiffness and increased deflections as indicated from
the reduced slope of the moment–deflection curves. The three
GFRP-reinforced beams exhibited bilinear moment–deflection
relationships in both uncracked and cracked stages. The reduction

Table 1
Details of test specimens.

Beam As (mm2) Af (mm2) As/Af R qs (%) qf (%) qeff (%) qfb (%) qeff/qfb
Group A: GFRP-reinforced beams
2G12 – 226.19 – – 0.00 0.38 0.38 0.49 0.77
3G12 – 339.29 – – 0.00 0.64 0.64 0.49 1.31
3G16 – 603.19 – – 0.00 1.12 1.12 0.49 2.28
Group B: Hybrid-reinforced beams: 0.51 < qeff < 0.67
2G12-1S10 78.54 226.19 0.35 1.56 0.15 0.43 0.51 0.49 1.04
2G12-2S10 157.08 226.19 0.69 2.25 0.27 0.39 0.55 0.49 1.12
2G12-2S12 226.19 226.19 1.00 4.00 0.43 0.43 0.67 0.49 1.36
Group C: Hybrid-reinforced beams: 0.85 < qeff < 1.13
2G16-2S10 157.08 402.12 0.39 1.39 0.27 0.70 0.85 0.49 1.74
2G16-2S12 226.19 402.12 0.56 2.78 0.41 0.73 0.96 0.49 1.96
2G16-2S16 402.12 402.12 1.00 4.00 0.73 0.73 1.13 0.49 2.32
522 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529
Fig. 1. Reinforcement details and test configuration: (a) group A, (b) group B, and (c) group C (refer to Table 1) (dimensions in mm).
in stiffness after cracking was mainly influenced by the amount of
GFRP reinforcement. Beam 2G12 (q/ = 0.38%) suffered the largest
reduction in stiffness followed by beams 3G12 (q/ = 0.64%) and
3G16 (q/ = 1.12%) as shown in Fig. 3(a).
Fig. 3b and c shows the moment–deflection curves of hybrid-
reinforced beams. The moment–deflection curve of beam 2G12 is
plotted in Fig. 3b for comparison with beams of group B. It can
be noticed that the moment–deflection curves of hybrid-
reinforced beams showed three distinct segments. The first seg-
ment extended from zero to the cracking moment and was almost
identical in all hybrid-reinforced beams. The second segment
extended linearly at the onset of cracking until yielding of steel
reinforcement occurred. The reduced slope of the second segment
demonstrated the loss of stiffness of the beams after cracking. After
yielding, the moment–deflection curves exhibited a non-linear
behavior with a pronounced reduction in slopes as loading contin-
ued to ultimate. This pattern of the moment–deflection curve was
identical in all hybrid-reinforced beams except for beam 2G12-
1S10 that displayed a bilinear curve similar to the under-
reinforced GFRP beam 2G12. As previously mentioned, beam
2G12-1S10 failed by rupture of the GFRP reinforcement
simultaneously with concrete crushing. This mode of failure indi-
cated the necessity of adding a minimum amount of steel rein-
forcement to ensure ductility and to prevent catastrophic failure
of hybrid-reinforced beams.
The results of the hybrid-reinforced beams indicated the effec-
tiveness of adding steel reinforcement on the stiffness of the hybrid
beams in comparison to their GFRP-reinforced counterparts. A con-
siderable increase in stiffness can also be noticed as the effective
reinforcement ratio increased (beams of group C in comparison
to beams of group B). This effect was more pronounced at service
loads prior to yielding. After yielding, the decreasing trend of stiff-
ness was dictated by the low modulus of elasticity of the GFRP
bars. At this stage, the hybrid beams suffered excessive deflections
prior to failure. Comparing the moment–deflection curves of
hybrid beams within each group, it was noticed that hybrid beams
with the smallest qeff (2G12-1S10 from group B and 2G16-2S12
from group C) exhibited the least stiffness before yielding as com-
pared to the other hybrid beams in each group (see Fig. 3b for
group B and Fig. 3c for group C). After yielding, beams of groups
B and C showed almost identical stiffness as evidenced from the
similar slopes of their moment–deflection curves at this stage.
 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529
Beam 2G12
Beam 3G12
Beam 3G16
Beam 2G16-2S10
Beam 2G16-2S12
Beam 2G16-2S16
Fig. 2. Cracking patterns at failure.
4.2. Yielding and ultimate moments
Experimental and analytical moments at which steel bars
yielded, My, for the hybrid-reinforced beams are listed in Table 2.
Using the equations of equilibrium and strain compatibility, the
yielding moments were determined analytically by setting the
strain at the level of reinforcement equal to ey = 0.0026 (yielding
strain of steel bars). In this case, the tensile force in the reinforcing
bars and the depth of the neutral axis were determined. Very good
agreement between the experimental and predicted values was
obtained.
The experimental and design moment capacities are also com-
pared in Table 2. The moment capacities of the test beams, Mu,
were calculated using Eq. (18) as follows:
Pu L
Mu ¼ ð18Þ
4 the identical slopes of the R data plots for groups B and C (Fig. 4)
Evidence of excellent agreement between the obtained capaci-
ties can be demonstrated. Discrepancies in the design moments
of some beams of groups B and C were attributed to the variation
in their effective reinforcement depths. It is important to mention
that actual reinforcement depths were used in the calculations.
These results validated the accuracy of the design model adopted
in this study.
On the other hand, it was important to assess the influence of
the effective reinforcement ratio, qeff, and the axial stiffness ratio,
R, (ratio of the axial stiffness of the steel bars to that of the GFRP
bars = AsEs/AfEf) on the moment capacities of the hybrid-
reinforced beams. Fig. 4 shows the relationship between the
523
Beam 2G12-1S10
Beam 2G12-2S10
Beam 2G12-2S12
experimental moment capacities (shown on the vertical axis) and
both qeff and R (shown on the upper and lower horizontal axes,
respectively). The flattened slope of the axial stiffness ratio, R, in
comparison to the sharp slope of the reinforcement ratio, qeff,
revealed that the influence of qeff on the flexural capacities of the
hybrid-reinforced beams was more significant than that of the
stiffness ratio R. The comparison between beams 2G12-2S12 and
2G16-2S16 confirmed this finding. Both beams had an axial stiff-
ness ratio R = 4, with the reinforcement ratio, qeff, of the former
beam being 0.67% compared to 1.15% for the latter beam. Beam
2G16-2S16 showed a moment capacity approximately 42% more
than that of the beam 2G12-2S12. Similarly, beam 2G16-2S10 with
qeff = 0.85% and R = 1.40 failed at higher moment than that of beam
2G12-2S10 with qeff = 0.55% despite the higher value of R = 2.25 of
the latter beam. These results also confirmed those of Qu et al. [15]
on hybrid (GFRP and steel) reinforced beams. On the other hand,
suggested that the axial stiffness ratio, R, had identical effects on
the moment capacities of both under-reinforced and over-
reinforced GFRP-reinforced beams, irrespective of the reinforce-
ment ratios used. These results indicated that the effective rein-
forcement ratio, qeff, of hybrid-reinforced beams rather than the
axial stiffness ratio, R, is a reasonable parameter to determine the
flexural capacities of hybrid beams.
4.3. Prediction of mid span deflections
Eqs. (8)–(17) were used to predict the midspan deflections of
the test beams at service moments up to 60% of the carrying-
524 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529

moment capacity of each beam. Comparison between the experi-

mental and predicted deflections is shown in Fig. 5.
Fig. 5a compares the predicted and experimental midspan
deflections of GFRP-reinforced beam 2G12 (qf = 0.38%). It was clear
that the prediction models overestimated the deflections at all
stages of loading after cracking, with the CSA-S806-12 model [9]
(Eq. (13)) being the most conservative. Both Bischoff and ACI-
440.1R-06 [3] equations predicted almost similar deflection values
as shown in Fig. 5a. Better prediction of deflections was found for
the GFRP over-reinforced beams 3G12 and 3G16 (qf = 0.64% and
1.12%, respectively). As shown in Fig. 5b and c, CSA-S806-12 [9]
equation overestimated the deflections of beams 3G12 and 3G16
at service moments, whereas ACI-440.1R-06 [3] equation underes-
timated the deflections at low moments. It was obvious that the
prediction of midspan deflections of GFRP-reinforced beams using
the Bischoff and ACI models improved as the reinforcement ratio
increased.
Experimental and predicted deflections of the hybrid-reinforced
beams of group B are shown in Fig. 5d–f. Midspan deflections of
beam 2G12–1S10 (qeff = 0.51%) were overestimated by all equa-
tions up to yielding of steel bars, with CSA-S806-12 [9] equation
being the most conservative. As the reinforcement ratio increased,
Bischoff equation tended to underestimate the midspan deflections
at higher service moments as shown in Fig. 5d–f. For beams 2G12-
2S10 and 2G12-2S12 (qeff = 0.55% and 0.67%, respectively), the ACI-
440.1R-06 [3] and CSA-S806-12 [9] equations accurately predicted
the deflections of the beams at high moments. Similar results were
found for hybrid beams of group C, as illustrated in Fig. 5g–i. From
a design point of view, the CSA-S806-12 [9] equation safely pre-
dicts the deflections up to moment equal to 40% of the nominal
capacity of the beam.
Fig. 6 shows the ratio of predicted to the experimental deflec-
tions for all beams at two moment levels namely, at 20% of the
nominal ultimate capacity (close to the cracking moment) and at
40% of the ultimate capacity (a reasonable service moment for
hybrid-reinforced beams). The experimental and predicted values
at these two moments are also tabulated in Table 3. From Fig. 6a,
it can be seen that all equations overestimated the deflections at
moments close to or immediately after cracking. The deflections
of specimen 2G12-1S10 were highly over predicted and were not
shown in the plots. In Fig. 6b, a consistent trend was depicted from
the plotted data. CSA-S806-12 [9] equation clearly overestimated
the deflections of GFRP-reinforced beams with low reinforcement
ratios. Predictions improved as the reinforcement ratio, qf,
increased from 0.38% (beam 2G12) to 1.1% (beam 3G16). This trend
was also valid for the hybrid-reinforced beams of group B. On the
other hand, an excellent agreement between the predicted and the
experimental deflections was noticed for all beams of group C,
which had higher effective reinforcement ratios. This might be
Fig. 3. Moment–deflection relationships for (a) group A, (b) group B, and (c) group C
attributed to the use of the cracked moment of inertia in the
of the test beams (refer to Table 1).
closed-form equation adopted by the CSA-S806-12 [9] method.

Table 2
Yielding moments, ultimate moments, and modes of failure of the test beams.

Beam Yielding moment, My (kN m) Ultimate moment, Mu (kN m) Mode of failure

Exp. Theo. Exp. Theo.
2G12   49.03 50.28 GFRP rupture
3G12   53.78 51.42 Concrete crushing
3G16   69.55 67.31 Concrete crushing
2G12-1S10   47.62 47.27 Steel & GFRP rupture, concrete crushing
2G12-2S10 25.31 25.88 53.55 58.43 Steel yielding, concrete crushing
2G12-2S12 34.06 31.12 58.94 55.72 Steel yielding, concrete crushing
2G16-2S10 30.06 31.02 68.30 71.41 Steel yielding, concrete crushing
2G16-2S12 40.44 37.38 64.71 70.92 Steel yielding, concrete crushing
2G16-2S16 55.31 56.47 83.53 81.39 Steel yielding, concrete crushing
 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529
Fig. 4. Effective reinforcement ratio, qeff, and axial stiffness ratio, R, versus ultimate
moments in hybrid-reinforced beams.
More data are required to thoroughly assess the CSA-S806-12 [9]
equation for hybrid beams with reinforcement ratios higher than
1.1% (the maximum ratio used in beams of group C). Bischoff and
ACI-440.1R-06 [3] equations predicted well the deflections of
beams with high reinforcement ratios but were not conservative
in some cases.
4.4. Cracking pattern and failure modes
According to the provisions of the Japan Society of Civil
Engineers [12], the crack width limitations can be disregarded for
purely GFRP-reinforced specimens considering the anti-corrosive
properties of the GFRP bars. However, the existence of steel bars
in hybrid-reinforced beams imposes limits to the widths of cracks
that can be formed during the beam service, as a preventive mea-
sure against corrosion. Fig. 7 shows the variation of crack widths
with the applied moments in both GFRP- and hybrid-reinforced
beams. Crack widths of beams 3G16 and 2G12-1S10 are not shown
in Fig. 7, as crack transducers were not accurately installed across
visible cracks, and hence the obtained results were not representa-
tive of the real crack widths.
It can be noticed from Fig. 7 that crack widths were inversely
proportional to the reinforcement ratio of the beam. Increasing
the amount of reinforcement decreased the crack widths in all test
beams. Comparing the crack widths of beam 2G12 and the hybrid-
reinforced beams of group B, it can be noticed that the use of steel
reinforcement significantly decreased the crack widths at and
beyond the service loads. The crack width at the onset of yielding
of steel bars for beam 2G12-2S10 was 0.54 mm (at
My = 25.31 kN m) compared to 1.0 mm for the beam 2G12 at simi-
lar moment. Similarly, the crack width for beam 2G12-2S12 was
0.60 mm at My = 34.06 kN m compared to 1.8 mm for the beam
2G12.
The ACI-440.1R-06 [3] recommends that the maximum crack
width in FRP-reinforced concrete beams be estimated as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ff  s 2
2
w¼2 bkb dc þ ð19Þ
Ef 2
Eq. (19) was used to predict the crack width of the test beams at
service moments. ACI-440.1R-06 [3] also recommends that the
bond coefficient kb be taken equal to 1.00 for steel-reinforced
beams and 1.40 for FRP-reinforced beams to account for the differ-
ent characteristics of FRP bars. Fig. 8a and b compares the crack
widths measured during the tests for beams 2G12 and 3G12,
525
respectively, with those predicted by Eq. (19) at service moments.
Lower and upper limits for the bond coefficient kb, taken as 0.62
and 1.40, respectively, were used to predict the crack width of
the beams. Using kb = 0.62 significantly under-predicted the crack
widths of the beams, whereas kb = 1.40 gave conservative results.
A value of kb = 1.10 showed good agreement between the experi-
mental and predicted values as can be seen in Fig. 8a and b and
was therefore recommended for the GFRP-reinforced beams tested
in this study.
Based on the test results, Eq. (20) is proposed to determine the
bond coefficient for the hybrid-reinforced beams:

15
Af
kb ¼ 1:4a ð20Þ
As
Eq. (20) takes into account the ratio of GFRP and steel reinforc-
ing bars while determining the bond coefficient, kb. The parameter
a accounts for the size effect of GFRP bar on its bond to concrete;
with smaller diameters having better bond. Values of a were taken
as 1.2 and 1.0 for the GFRP bar of diameters 16 mm and 12 mm,
respectively. Since the GFRP bars used in this study were similar,
no data was available to predict the effect of the surface texture
of the FRP bar on the bond coefficient kb in Eq. (20).
The plots in Fig. 8c–g show the variation of the crack widths
with the applied moments using a lower limit of kb = 1.0 and an
upper limit determined from Eq. (20). It can be noticed that using
kb = 1.0 in the ACI-440.1R-06 [3] equation significantly under-
estimated the crack widths at all stages of loading prior to yielding.
The coefficient kb calculated from Eq. (20) accurately predicted the
crack widths, except in the case of beam 2G12-2S10 in which the
equation underestimated the crack widths at high service
moments. Values of kb determined from Eq. (20) are shown in
Fig. 8. In all cases, the ACI-440.1R-06 [3] equation cannot accu-
rately predict the crack widths beyond yielding irrespective of
the value of kb used. Due to the significant influence of the bond
coefficient, kb, on the predicted crack widths, a thorough investiga-
tion is required to accurately determine this parameter for hybrid-
reinforced beams to account for the different characteristics of
commercially available FRP bars.
4.5. Ductility and deformability
The ductility concept is related to the ability of the beam to sus-
tain inelastic deformations and large rotations prior to failure
without compromising its load-carrying capacity. The ductility of
conventional steel-reinforced concrete beams can be determined
by the ductility index, DI, taken as the ratio of the total deformation
at ultimate to the deformation at yielding. This model cannot be
applied to FRP-reinforced beams because of the linear response
of the bars up to failure. Therefore, the Canadian Highway Bridge
Design Code Canada (CSA-S6-10) [11] adopts a concept based on
the deformability of the beam rather than its deflection to ensure
adequate deformation of the FRP-reinforced beams. The deforma-
bility factor, DF, recommended by the CSA-S6-10 code [11] is given
in Eq. (21). The CSA-S6-10 code [11] recommends that the service
moment, Ms, and the curvature of the beam be taken as the
moment and curvature corresponding to a maximum compressive
concrete strain of 0.001. It also recommends that the DF value
should not be less than 4 for GFRP-reinforced beams.
M u wu
DF ¼ ð21Þ
M s ws
In the present study, the deformability approach was adopted
and DF was determined for GFRP- and hybrid-reinforced beams.
Table 4 shows the ultimate and service moments at which DF
was determined for GFRP-reinforced beams. The beams’ curvatures
526 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529

Fig. 5. Experimental versus predicted midspan deflections.

 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529 527

Fig. 7. Variation of crack widths with the applied load.

Yielding and ultimate moments and their corresponding curva-

tures are listed in Table 5. The results showed that the deformabil-
ity of the hybrid-reinforced beams decreased as the amount of
steel bars increased. The highest DFmod was obtained for beam
2G12-1S10 that had the least reinforcement ratio (qeff = 0.51%)
and, conversely, the lowest value of DFmod was obtained for the
highly reinforced beam 2G16-2S16 (qeff = 1.15%).

5. Conclusions and recommendations

The behavior of hybrid (GFRP and steel) reinforced concrete

Fig. 6. Experimental versus predicted midspan deflections at 20% and 40% of
ultimate moments.
at ultimate and at concrete strain of 0.001 are also listed in Table 4.
For hybrid-reinforced beams, the use of service moments, Ms, in Eq.
(21) at concrete strain of 0.001, as recommended by the CSA-S6-10
code [11], neglected the yielding of steel bars. Therefore, the
authors suggested a modified deformation factor, DFmod, taken as
the ratio of the product of moment and curvature at ultimate to
the product of moment and curvature at yielding of steel reinforce-
ment as given in Eq. (22):
Mu wu
DF mod ¼
My wy
beams has been investigated in this study. The structural behavior
in terms of load-carrying capacity, deflection, cracking, deformabil-
ity, and failure mode was evaluated. Based on the test results, the
following conclusions could be drawn:
– The use of steel reinforcement in combination with GFRP rein-
forcement enhanced the flexural behavior of purely GFRP-
reinforced concrete beams in terms of deformability, cracking,
stiffness, and load-carrying capacity. The enhancement in ser-
viceability aspects was more pronounced at service loads prior
to steel yielding.
– Design of hybrid-reinforced beams should consider steel yield-
ing prior to concrete crushing or FRP rupture to ensure adequate
deformation in the beams.
– The effective reinforcement ratio, qeff, as defined in this study,
had more influence than the axial stiffness ratio, R, on the
moment-carrying capacity of the hybrid-reinforced beams.
ð22Þ Increasing qeff increased the moment capacities of the hybrid-
reinforced beams.

Table 3
Midspan deflections at service moments 20% and 40% of ultimate moments of the test beams.

Beam Dm at 20% ultimate, mm Dm at 40% ultimate, mm

Exp. Bischoff model ACI-440.1R-06 [3] CSA-S806-12 [9] Exp. Bischoff model ACI-440.1R-06 [3] CSA-S806-12 [9]
2G12 6.91 6.53 8.03 20.82 29.99 40.29 38.20 48.09
3G12 12.05 9.45 6.35 21.30 35.98 40.45 34.78 46.81
3G16 8.52 12.18 7.95 17.37 26.93 31.31 29.25 34.07
2G12-1S10 1.08 4.90 7.24 12.89 22.60 22.64 23.67 27.26
2G12-2S10 3.31 4.50 7.92 8.42 15.65 15.22 17.15 17.38
2G12-2S12 5.50 5.44 8.28 8.88 17.38 16.23 17.86 18.08
2G16-2S10 4.41 6.45 7.37 9.21 18.87 17.14 17.99 18.57
2G16-2S12 4.84 5.52 6.51 8.01 14.47 14.39 15.21 15.72
2G16-2S16 5.28 6.11 7.16 7.40 14.74 13.74 14.34 14.40
528 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529

Fig. 8. Experimental versus predicted crack widths.

Table 4
Deformability factors, DF, for GFRP-reinforced beams using Eq. (21).

Beam At ultimate At service DF

Mu (kN m) ecu (mm/mm) cu (mm) wu  105 (1/mm) Ms (kN m) ecs (mm/mm) cs (mm) ws  105 (1/mm)
2G12 50.28 0.003 34.25 8.76 23.21 0.0010 28.30 3.53 5.37
3G12 51.42 0.003 40.30 7.44 22.99 0.0010 32.03 3.12 5.33
3G16 67.31 0.003 52.68 5.69 30.48 0.0010 42.15 2.37 5.30
 A. El Refai et al. / Construction and Building Materials 96 (2015) 518–529 529

Table 5
Modified deformability factors, DFmod, for hybrid beams using Eq. (22).

Beam At ultimate At yield DFmod

Mu (kN m) ecu (mm/mm) cu (mm) wu  10 5
(1/mm) My (kN m) ecy (mm/mm) cy (mm) wy  10 5
(1/mm)
2G12-1S10 47.27 0.003 36.82 8.15 15.23 0.0005 39.64 1.37 18.52
2G12-2S10 58.43 0.003 42.01 7.14 25.88 0.0007 50.97 1.31 12.34
2G12-2S12 55.72 0.003 43.95 6.83 31.12 0.0008 54.99 1.49 8.23
2G16-2S10 71.41 0.003 52.20 5.75 31.02 0.0008 55.35 1.35 9.76
2G16-2S12 70.92 0.003 54.41 5.51 37.38 0.0009 59.96 1.44 7.24
2G16-2S16 81.39 0.003 63.43 4.73 56.47 0.0011 71.86 1.55 4.41

– All hybrid-reinforced beams with qeff > qfb failed in a favorable
ductile manner due to concrete crushing after yielding of steel
reinforcement.
– The presence of steel bars in balanced and slightly over-
reinforced hybrid beams did not prevent the catastrophic fail-
ure of the beams due to GFRP bars rupture. The limit of qeff > 1.4
qfb recommended by ACI-440.1R-06 [3] should also be applied
to hybrid-reinforced beams.
– A new bond coefficient, kb, that accounts for the GFRP bar diam-
eter and the ratio of GFRP to steel bars, has been proposed.
Using this coefficient, the ACI-440.1R-06 [3] equation ade-
quately estimated the crack widths for hybrid-reinforced beams
prior to yielding.
– More studies are required to accurately determine the bond
coefficient, kb, to account for the various characteristics of com-
mercially available FRP bars.
– The CSA-S806-12 [9] equation was found to be conservative in
predicting the deflections of hybrid-reinforced beams.
However, it accurately predicted the deflections of the hybrid-
reinforced beams that have high effective reinforcement ratios.
– A new deformability factor that accounts for steel yielding in
hybrid-reinforced beams has been proposed. Increasing the
amount of steel reinforcement decreased the modified deforma-
bility factors of the hybrid-reinforced beams.
It should be emphasized that the above conclusions are based
on the test results for the GFRP- and the hybrid-reinforced beams
presented in this study. The applicability of these conclusions on
beams reinforced with other type of FRP bars or other combina-
tions of FRP and steel bars is unknown.
Acknowledgment
The authors would like to thank Pultron personnel for providing
the GFRP bars used in the tests.
References
[2] American Concrete Institute (ACI) Committee 318, Building code requirements
for structural concrete and commentary, ACI 318-08, American Concrete
Institute, Farmington Hills, MI, USA, 2008.
[3] American Concrete Institute (ACI) Committee 440, Guide for the design and
construction of structural concrete reinforced with FRP bars, ACI 440.1R-06,
American Concrete Institute, Farmington Hills, MI, USA, 2006.
[4] P.H. Bischoff, Reevaluation of deflection prediction for concrete beams
reinforced with steel and fiber reinforced polymer bars, J. Struct. Eng. 131
(5) (2005) 752–767.
[5] P.H. Bischoff, Deflection calculation of FRP reinforced concrete beams based on
modifications to the existing Branson equation, J. Compos. Constr. 11 (1)
(2007) 4–14.
[6] P.H. Bischoff, R. Paixao, Tension stiffening and cracking of concrete reinforced
with glass fiber-reinforced polymer (GFRP) bars, Can. J. Civ. Eng. 31 (4) (2004)
579–588.
[7] P.H. Bischoff, A. Scanlon, Effective moment of inertia for calculating deflections
of concrete members containing steel reinforcement and fiber-reinforced
polymer reinforcement, ACI Struct. J. 104 (1) (2007) 68–75.
[8] D.E. Branson, Design procedures for computing deflections, ACI Struct. J. 65 (8)
(1968) 730–742.
[9] Canadian Standards Association (CSA), Design and construction of building
components with fiber-reinforced polymers, CSA-S806-12, Mississauga, ON,
Canada, 2012.
[10] Canadian Standards Association (CSA), Design of concrete structures standard,
CSA-A23.3-04, Mississauga, ON, Canada, 2004.
[11] Canadian Standards Association (CSA), Canadian highway bridge design code,
CSA-S6-10, Mississauga, ON, Canada, 2010.
[12] Japan Society of Civil Engineers (JSCE), Recommendation for design and
construction of concrete structures using continuous fiber reinforcing
materials, Concrete engineering series No. 23, Research Committee on
Continuous Fiber Reinforced Material, Tokyo, Japan, 1997.
[13] D. Lau, H.J. Pam, Experimental study of hybrid FRP reinforced concrete beams,
Eng. Struct. 32 (12) (2010) 3857–3865.
[14] H.Y. Leung, R.V. Balendran, Flexural behavior of concrete beams internally
reinforced with GFRP rods and steel rebars, Struct. Surv. 21 (4) (2003) 146–
157.
[15] W. Qu, X. Zhang, H. Huang, Flexural behavior of concrete beams reinforced
with hybrid (GFRP and steel) bars, J. Compos. Constr., ASCE 13 (5) (2009) 350–
359.
[16] M.A. Safan, Flexural behavior and design of steel–GFRP reinforced concrete
beams, ACI Mater. J. 110 (6) (2013) 677–685.
[17] W.C. Tang, R.V. Balendran, A. Nadeem, H.Y. Leung, Flexural strengthening of
reinforced lightweight polystyrene aggregate concrete beams with near-
surface mounted GFRP bars, Build. Environ. 41 (2006) 1381–1393.
[18] J.R. Yost, S.P. Gross, D.W. Dinehart, Effective moment of inertia for glass fiber-
reinforced polymer-reinforced concrete beams, ACI Struct. J. 100 (6) (2003)
732–739.

[1] M.A. Aiello, L. Ombres, Structural performances of concrete beams with hybrid
(fiber-reinforced polymer–steel) reinforcements, J. Compos. Constr. 6 (2)
(2002) 133–140.