Available online at w w w . r i l e m .

n e t

Materials a n d Structures 38 (July 2005) 627-637

Performance of RC frames with hybrid reinforcement

under reversed cyclic loading

M. Nehdi and A. Said

Dept, o f Civil & Env. Eng., The University o f Western Ontario, London, Ontario, Canada, N 6 A 5B9

Received." 5 May 2004; accepted." 13 October 2004

ABSTRACT
The use of FRP as reinforcement in concrete structures has been growing rapidly. A potential application of FRP reinforcement is in
reinforced concrete (RC) frames. However, due to FRP's predominantly elastic behaviour, FRP-RC members exhibit low ductility and
energy dissipation. Hybrid steel-FRP reinforcement can be a viable solution to the lack of ductility of FRP-RC members. Using two layers
of reinforcement in a section, FRP rebars can be placed in the outer layer and steel rebars in the inner layer away from the effects of
carbonation and chloride intrusion. Combined with the use of FRP stirrups, this approach can enhance the corrosion resistance of RC
members. However, current design standards and detailing criteria for FRP-RC structures do not provide detailed seismic provisions. In
particular, the design and detailing of beam-column joints is a key issue in seismic design. During recent earthquakes, many structural
collapses were initiated or caused by beam-column joint failures. Thus, research is needed to gain a better understanding of the behaviour
of FRP and hybrid FRP-steel-RC under seismic loading. In this study, three full-scale beam-column joint specimens reinforced with steel,
GFRP and a hybrid GFRP-steel configuration, respectively were tested in order to investigate their performance in the event of an
earthquake.
1359-5997 9 2004 RILEM. All rights reserved.

RI~SUMI~
L 'utilisation de barres en polymkres renforcOs defibres (PRF) comme armature clans les structures en b~ton esten train d'augmenter
rapidement. Une application prometteuse des armatures en P R F est dans les structures en cadres de b~ton arm~ d multi-ktages.
Cependant, ~ cause de leur comportement ~lastique, les membres renforcOs avee des armatures en P R F d~montrent une ductilitO limit~e et
une faible capacitO de dissipation d Unergie. Les systbmes hybrides (combinaison d'armatures en aeier et en polymOres renforc~s de fibres
de verre (PRFV)) se pr~sentent comme une solution pratique pour rem~dier aux inconv~nients des systOmes renforc~s uniquement avec
P R F K En utilisant deux couches d'armatures, les barres en P R F V peuvent Otre plac~es h l'ext~rieur et ceux en acier h l'int~rieur, loin des
effets de la carbonatation et 1 'intrusion des ions chlores. En plus, l'utilisation des ktriers en P R F V peut amOliorer la durabilit~ de ces
structures. Cependant, les normes de conception actuelles pour P R F n'offrent pas assez de d~tails sur les provisions sismiques, en
partieulier sur la conception des joints de poutres-eolonnes. De rOcents tremblements de terre ont d~montrO sans Oquivoque que les joints
de poutres-eolonnes sont critiques pour assurer I'intOgrit~ structurale. Alors, il est important d'~tudier le comportement de tels
assemblages utilisant des armatures en P R F V ou des armatures hybrides (aeier-PRFV). Dans eette ktude, trois assemblages poutre-
colonne en grandeur nature renforc~s avec des armatures en acier, PRFV, et hybride (acier-PRFV) ont ~tO construits et testOs sous charges
cycliques pour examiner leur comportement durant des tremblements de terre.

1. INTRODUCTION measurement devices for structural health monitoring purposes.

Corrosion of reinforcing steel has been the primary cause o f
deterioration o f reinforced (RC) structures, requiring substantial
annual repair costs around the world. Furthermore, modem
equipments that employ magnetic interferometers, such as in
hospitals, require a nonmagnetic environment with no metallic
reinforcement. This has led to an increasing interest in fibre-
reinforced polymers (FRP) reinforcement, which is inherently
nonmagnetic and resistant to corrosion [1]. FRP reinforcement
also provides the option of easily embedding fibre optic strain
1359-5997 9 2004 RILEM. All rights reserved.
doi:10.1617/t4221
However, FRP materials often exhibit weaker bond to concrete
and lower ductility compared to that of conventional steel
reinforcement. The bond of FRP to concrete can be improved
by means of mechanical anchorages such as surface
deformations and sand coating, but its lower ductility remains a
major concern, especially in structures subjected to seismic
and/or impact loading.
Brown and Bartholomew [2] observed that FRP-RC beams
behaved in a similar manner to that o f steel-RC beams. They
argued that strength design methods for steel-RC beams can
predict the ultimate moment capacity of FRP-RC beams.
628 M. Nehdi, A. SaM/Materials and Structures' 38 (2005) 627-63 7
However, in the design process, two criteria that are not
usually problematic in the case of steel reinforcement can
govern the design in the case of FRP reinforcement: deflection
and ductility. Most FRP materials usually have a significantly
lower modulus of elasticity compared to that of steel (except
for some new CFRP products) and thus, often generate higher
deflections, Furthermore, the predominantly elastic behaviour
of FRP results in little warning before a usually sudden and
brittle failure. Satisfying deflection and ductility requirements
is a challenge in designing FRP-RC structures. Thus, it is
recommended that flexural design of FRP-reinforced slabs and
beams should aim at over-reinforced sections in order to
achieve a concrete compression failure, which usually allows
FRP-RC flexural members to exhibit some plastic behaviour
before failure [3, 4].
In recent years, there has been a growing interest to
investigate the performance of mixed steel-FRP as well as
steel-free FRP-RC structures. However, research in this area
has been generally limited to some beam and column testing.
Most of the newly adopted specifications for the design of
FRP-reinforced concrete [4-8] are not comprehensive, often do
not include detailed seismic provisions, and do not cover
hybrid FRP-steel RC systems. Therefore, research is needed to
investigate the performance of FlIP and hybrid FRP-steel-
reinforced concrete frames under reversed cyclic loading in
order to form the basis for future design code provisions for
FRP-reinforced concrete in seismic zones. In this study, full-
scale steel-reinforced, steel-free GFRP-reinforced, and hybrid
GFRP-steel-reinforced beam-column joints were tested under
reversed cyclic loading; Their behaviour including load-storey
drift envelope relationship and energy dissipation were
compared and discussed.
2. SCOPE OF PREVIOUS W O R K
The use of FRP as reinforcement in RC beams was
investigated by various researchers. Different permutations of
FlIP and steel as longitudinal and transverse reinforcement,
respectively were studied [9-11]. Although transverse
reinforcement is usually closer to the concrete surface and is
therefore more vulnerable to corrosion, limited investigations
have been performed on the use of FRP stirrups. The use of
FlIP stirrups has been hindered by their limited availability and
the fact that a 60% strength reduction factor at bends for
various types of FRP is recommended [6]. Also, bending FRP
bars to make stirrups typically needs to be performed in
production plants with special care and equipment. The use of
FRP NEFMAC (New Fiber Composite Material for
Reinforcing Concrete) grids can provide a solution to such a
problem; a four-ceU unit taken from a NEFMAC grid provides
a three-branched stirrup as shown in Fig. 1.
Grira and Saatcioglu [12] investigated the use of both steel
grids and CFRP grids as stirrups for confinement of concrete
columns having longitudinal steel reinforcement. Several grid
configurations were used and column specimens were tested
under cyclic loading. They concluded that the performance of
columns reinforced with CFRP stirrups was comparable to that
of columns reinforced with steel stirrups. They also argued that
the use of grids whether made of steel or CFRP provides ease
Fig. 1 - Typical four cell GFRP stirrup used in this study.
of construction and a near-uniform distribution of the
confinement pressure along the column, without congesting
the reinforcement cage. They reported that the NEFMAC grid-
based stirrups failed at their nodes, which is usually the
common weakness of FRP stirrups.
Fukuyama et al. [13] tested a half-scale three-storey
AFRP-reinforced concrete frame under quasi-static loading.
RA11S aramid-bars were used for the longitudinal
reinforcement of columns, RA7S bars were used as flexural
reinforcement for beams and slabs, while RA5 bars were used
as shear reinforcement. RA11S, RA7S and RA5 are braided
bars with cross-sectional areas of 90, 45 and 23 mm 2,
respectively. It was argued that frame deformations governed
the design. The frame remained elastic up to a drift angle of
1/50 rad, and no substantial decrease in strength took place
after rupture of some main beam rebars owing to the high
degree of structural indeterminacy of the frame. It was also
noted that the rehabilitation of such a frame was easier than
that of conventional RC flames since residual deformations
were smaller. However, the flame was not tested to collapse
and its behaviour under excessive deformations was not
reported.
Limited research has been performed on hybrid FRP-steel
reinforced concrete. Aiello and Ombres [14] tested 6 beams
with different configurations of longitudinal reinforcement
including steel only, AFRP only, and hybrid AFRP-steel
beams, all with steel stirrups. For some of the hybrid
specimens, steel was placed with a larger concrete cover to
provide extra protection against corrosion. Experimental
results showed that such a hybrid system can have lower
service deflection and higher ductility at failure than that of
the AFRP steel-flee system. Leung and Balendran [15] tested
seven RC beams under four point bending. Concrete strength
and reinforcement ratios for both steel and GFRP were varied
to produce under-reinforced and over-reinforced sections.
Steel rebars were placed at 30 mm higher concrete cover
compared to that for GFRP rebars. The study showed that for
hybrid beams, steel contributed more effectively to the
overall behaviour up to yield. Afterwards, the stiffness of
yielded steel dropped drastically and the GFRP rebars started
to contribute more efficiently to the section resistance. For
high strength concrete beams, the increased flexural capacity
resulted in shifting the flexural failure into a shear failure.
 M. Nehdi, A. Said / Materials and Structures 38 (2005) 62 7-63 7
Some research focused on providing ductility to FRP
rebars that are manufactured by filament winding or
pultrusion. For instance, Tamuzs and Tepfers [16] investigated
the properties of a hybrid FRP rod. They used multiple fibre
types along with braiding fibre strands around a soft porous
core to achieve a more ductile behaviour. The hybrid rods they
produced could provide a ductile behaviour, but the difference
between the moduli of different fibre strands seemed to cause
uneven load transfer, while the compression of the core
material caused a reduction of cross-section. A similar study
was performed by Bakis et al. [17] who developed pseudo-
ductile FRP rods using different types of fibres. The rods
behaved in a pseudo-ductile manner when tested under
tension, but premature failure took place due to local stress
concentrations. Another study performed by Harris et al. [18]
developed a ductile hybrid FRP rebar through braiding of
various fibres followed by a pultrusion process. Belarbi et al.
[19] were also successful in developing composite reinforcing
rebars with a relatively more stable stress-strain behaviour in
tension and better load-deflection behaviour under four-point
bending. However, such rebars are still in early experimental
stages and there is not enough data on their field performance,
especially under seismic loading.
3. EXPERIMENTAL PROGRAM
Beam-column joints can be isolated from plane frames
at the points of contraflexure. The beam of the current test
unit is taken to the mid-span of the bay, while the column is
taken from the mid-height of one storey to the mid-height
of the next storey.
3.1 Steel-reinforced s p e c i m e n (J1)
The first specimen constructed in this study is shown in
Fig. 2. It is a standard beam-column joint (dl) designed to
satisfy both CSA A23.3-94 [20] and ACI 352R-02 [21]
requirements. It has sufficient shear reinforcement in the joint
area, in the column hinging area, and in the beam hinging area.
The column was 3000 mm high with cross-section dimensions
of 250x400 ram. The beam's length was 1750 mm from the
face of the column to its free end with a cross-section of
250x400 ram. The longitudinal reinforcement used in the
column was 14 M15 bars corresponding to a 2.8%
reinforcement ratio (M15 is equivalent to a 16.0 mm diameter
bar). The transverse reinforcement in the column was made of
M10 closed rectangular ties. The column ties were spaced at
80 mm inside the joint and along 500 mm above and below it
(one sixth of the floor's height) and then spaced at 125 mm for
the rest of the column's height. The top and bottom
longitudinal reinforcements of the beam were 6 M15 bars
each, corresponding to a 1.2% reinforcement ratio. The
transverse reinforcement of the beam was made of M10
rectangular ties starting at 50 mm from the face of the column.
The ties were spaced at 80 mm for the 800 mm adjacent to the
column (equivalent to twice the beam's depth) and then spaced
at 120 mm for the remaining 840 ram, ending at 60 mm from
the free end of the beam.
629
no 10 @ 125mm
Strain gauge locations
no 10 @ 80mm
4750~
I
I-
I 4
~80mm ~ --no 10 @ 120ram
Column stirrups detail
Dimensions in mm.
Fig. 2 - Reinforcement details and strain gauges' locations for
the standard steel-reinforced specimen (Jl).
3.2 G F R P - r e i n f o r c e d s p e c i m e n (J4)
The second beam-column joint specimen constructed in
this study (d4), shown in Fig. 3, had identical dimensions to
specimen (J1) but was made with GFRP grid reinforcement
and a different reinforcement configuration. A view of the
GFRP reinforcement cage is shown in Fig. 4. The
longitudinal GFRP rebars used were NEFMAC G16
(201 mm 2 of cross-sectional area) with 5 nodes per meter in
the beam, but this was increased to 10 nodes per meter in
the joint area to provide extra mechanical anchorage and
avoid premature slippage of the beam reinforcement from
the joint. The rebars used in the column had 5 nodes per
meter. Nodes in all GFRP longitudinal rebars are 50 mm
wide from end to end. The longitudinal reinforcement
configuration aimed at providing a similar bending moment
capacity to that of the standard steel-reinforced specimen,
thus inducing a comparable level of joint shear input. The
transverse reinforcement in specimen (J4) consisted of 3-
branched (both vertically and horizontally) G10 (77 mm 2 of
cross-sectional area) GFRP stirrups. This provides built in
redundancy since the failure of a branch is not complete
until both of its two vertical portions fail. Some properties
of the NEFMAC grids are listed in Table 1. NEFMAC
made of GFRP was chosen instead of that made of CFRP
since its strain at failure is typically 67% higher, which
should give a better indication of imminent failure owing to
larger ultimate deformations.
630 M. Nehdi, A. Said~Materials and Structures 38 (2005) 627-637

3.3 H y b r i d - r e i n f o r c e d s p e c i m e n (J5)
) ~4e,0- ; The choice of the reinforcement configuration for the
hybrid GFRP-steel-reinforced specimen (J5) was aimed at
jj - providing a balance between the behaviour of traditional
~ .... 1310 @ 125ram
steel RC and that of steel-free FRP-RC. Fig. 5 shows the
,J moment-curvature diagram for several beam section
I configurations which served for selecting the hybrid
reinforcement scheme. While using 6 steel rebars or 6
D,s
G10 @ 80mm GFRP rebars provided either remarkably ductile and stiff or
brittle and flexible sections, respectively a mixed
1750 !. . . . . . . . (
\ configuration could strike a reasonable balance between the

1
-
lkir ,l' I I
1

GI0 @ 80ram..... :
........

GI0 @ 120nun
,
Iiio,6!!:
"
4G16 'J:!:::!:~ 4a~

25o ~
two criteria. Moreover, placing the steel rebars in a second
reinforcement layer with a thicker concrete cover provides
extra protection against corrosion. The section will initially
have high stresses in the steel rebars (which are stiffer) up
r to yielding at which point the stiffness of steel drastically
{
[
I I I I I I I I I I I drops and the GFRP rebars are further mobilized. As
~oo~5 Typical G 16 bar for beam reinforcement loading progresses, the GFRP rebars reach their capacity
! ,, ~2
and fail in tension, leaving the steel rebars to supply the
I
1
,L5~
I reserve strength of the section. It can be noted in Fig. 5 that
a hybrid section has two distinct post-cracking points: steel
yield and FRP failure.
Typical G 10 Stirrup Dimensions in ram.
.I.

Fig. 3 - Reinforcement details and strain gauges' locations for

the GFRP-reinforced specimen (.14). '8~
160 F
~" 140 I ~ ~,r

'2~
100 I }t ,
..,'7~ . .

steel yield
. .

1
Igg/" I'-'~ 6steol
40 II~ I ' - ' - 6GFRP I
I[ I " .a.,.. 3SteeI+3GFRP /

0.02 0.04 0.06 0.08 0.10 0.12 0.14

C u r v a t u r e (rad/m x 10 "3)

Fig. 5 - Typical moment curvature for different beam

reinforcement configurations.

The hybrid specimen (J5) had identical dimensions to that

of the two other specimens (Fig. 6) but was made using
Fig. 4 - View of the GFRP reinforcement cage for specimen (J4) ASLAN 100 GFRP #5 rods (#5 rod is equivalent to a 16.0 mm
lifted on the forms before casting.
diameter bar with 217 mm 2 cross-sectional area) as
longitudinal reinforcement together with steel rebars; some
Table 1 - Properties of GFRP NEFMAC grids [25] properties of ASLAN rebars are listed in Table 1. In specimen
and ASLAN 100 rebars [26] J5, GFRP NEFMAC grids were solely used for transverse
reinforcement. A view of the hybrid-reinforcement cage is
Sectional Max Tensile Modulus of
Bar Type and No. Area Load Strength Elasticity shown in Fig. 7. For the longitudinal reinforcement of the
(mm2) (kN) (MPa) (GPa) beam, three ASLAN 100 GFRP rebars and three M15 steel
rebars (in top and bottom) were placed in two layers with the
NEFMAC G10 77 46.7 600 30 GFRP rebars in the outer layer, thus providing extra cover for
NEFMAC G16 201 119.2 600 30 the steel rebars. The column also had a total of 12 rebars
equally divided between M15 steel and #5 ASLAN GFRP
GFRP rebars. The longitudinal reinforcement configuration of the
217 142.5 655 40
ASLAN 100 #5
hybrid specimen aimed at providing a similar bending moment
capacity to that of the standard steel-reinforced specimen, thus
inducing a comparable level of joint shear input. The
transverse reinforcement configuration in specimen (J5) was
identical to that of the steel-free specimen (J4).
 M. Nehdi, A. Said/Materials and Structures 38 (2005) 627-637 631

3.4 Test setup and procedure

+ +
i
+ The beam-column joint specimens were tested under a
constant axial load of 600 kN applied on the column, and
reversed cyclic load (quasi-static) applied at the beam tip.
The constant axial load on the column was applied through
a hydraulic jack before applying the reversed cyclic load on
L25m the beam to simulate a typical state of stress in service. The
-iHlllll [l[lll[ selected loading pattern applied at the beam tip was
ZF
intended to cause forces that induce high levels of
Strain gauges locations
deformations usually experienced by structural frames

~ G10 @ 80mm

1750
i-i
during severe earthquakes.
For the steel-reinforced specimen, the selected load
history consisted of two phases. The first one was load-
controlled in which two load cycles at approximately 10%
of the estimated strength of the specimen were applied to
check the test setup and ensure that all data acquisition
I channels were functioning properly. This was followed by
gOmm ~ -G10 @ 120mm- - ~ L two load cycles reaching the concrete flexural cracking load
in the beam at the column face. These in turn were followed
M15 are steel rebars
# 5 are ASLAN GFRP Rebars by two cycles at the load causing initial yield in the beam
I ~00w25
' ~ %25
GI0 are NEFMAC GFRP grids measured through the load-displacement trace on the data
acquisition monitor. The displacement at initial yield of the
beam section adjacent to the column face, dy, was recorded
i-- and used in the subsequent displacement-controlled phase
of loading in which multiples of the yield displacement, 8,,
! ~ Typical G10 Stirrup Dimensionsin mm.
were applied to the specimen. For each load increment, two
consecutive cycles were applied at the same loading level to
Fig. 6 - Reinforcement details and strain gauges' locations for verify the stability of the specimen. The load routine is
the hybrid-reinforced specimen (J5).
shown in Fig. 8.
A different loading routine was selected for the GFRP-
reinforced (J4) and hybrid-reinforced (J5) specimens since
unlike conventional steel-reinforced sections, those
reinforced with GFRP do not undergo a distinct yielding. A
displacement-controlled load history similar to the one used
by Fukuyama et al. [13] was applied in which incremental
values of drift were imposed on the specimen. Drift was
applied starting at 1/2000rad, then increased to pre-
specified values (1/1000, 1/800, . ..... 1/33, 1/22, 1/20 rad)
in both directions as shown in Fig. 9. The very first drift
was applied in one cycle, while all other subsequent drifts
were applied in two cycles.
The specimens were placed in the test rig as shown in
Fig. 10 to mimic a hinge support at the base of the column
Fig. 7 - View of the hybrid reinforcement cage for specimen (J5) and a roller support at the top part of the column. The roller
shown in the formwork before casting. support was created using a 2 cm vertical slot, which

The assembly of the steel-flee GFRP-reinforced (J4) and

the hybrid-reinforced (J5) specimens was performed at a much
5 - Loadcontrol Displacementcontrol
faster pace than that of the steel-reinforced specimen (J1). The 6 4 ,I
stirrups, being taken from a manufactured grid, were 3
2
dimensionally identical, thus, the longitudinal reinforcement L~ t
needed very little remrangement. The much lighter weight of
the GFRP rebars allowed easier manipulation of the ~-2
reinforcement cage. For the steel-reinforced specimen (J1), -3
extra work was required to fit steel rebars in place, especially
in the congested joint area where the steel stirrups
configuration was more complicated. Loading cycles

Fig. 8 - Load history of the reversed cyclic load test used for the
steel-reinforced specimen J1.
632 M. Nehdi, A. Said/Materials and Structures 38 (2005) 627-63 7

3/50 HYDRAULIC~
JACK ~_
3/75
LOAD
1/50 CELL a:~
e~o 4 8 12
0
2 cmSLOT i
0/50
-2/75
-2/50 LVDTT
Loading cycles
TESTSPECIMENLVDT~
Fig. 9 - Load history of the reversed cyclic load test used for the LVDT ] L~CI
LOAD - 1670 '/
GFRP-reinforced specimen J4 and hybrid specimen J5.
CELL
ACTUAT~
SWIVELJOINT

Fig. 11 - Test setup.

4.1 Behaviour of steel-reinforced specimen (J1)

Fig. 10 - View of the test rig.
allowed vertical deformation in the column as well as the
transmission of the column's axial load from the hydraulic
jack to the lower hinge support. The cyclic load was applied
at the beam tip using a loading ram through a greased pin
connection at an arm length of 1670 mm measured from the
column face. A view and a schematic of the test setup are
shown in Figs. 10 and 11, respectively.
4. ANALYSIS OF RESULTS
The analysis of test results includes a comparison of
load-storey drift plots, response envelopes, energy
dissipation, and beam moment-rotation relationships for the
three full-scale beam-column joint specimens tested in this
study.
The beam tip load-storey drift angle plot for the
standard steel-reinforced specimen is shown in Fig. 12.
First flexural cracking of the beam section subjected to
maximum bending moment appeared at a beam tip load of
15 kN, corresponding to a drift of 0.0017 rad. The onset of
diagonal cracks in the joint area took place at a beam tip
load of 50 kN, corresponding to a drift of 0.006 rad.
Additional cracks in the joint area appeared thereafter as
loading progressed, but remained within a very fine width
throughout the test. The yield of the beam's longitudinal
steel was reached at an average beam tip load of 107 kN
and the corresponding average yield displacement, dy, was
28 mm (drift of 0.0150 rad, based on the average of push up
and pull down values). At a deformation level equal to 2dy
(0.03 rad drift angle), the beam became extensively cracked
along a distance equal to its depth from the face of the
column. At a deformation equal to 4dy (0.06 rad drift
angle), wide cracks developed in the hinge area of the beam
and rubble started falling. At 6@ (0.09 rad drift angle), the
flexural hinge area of the beam lost most of its concrete.
The test was stopped as the beam capacity dropped but the
axial load in the column was maintained and the joint areas
remained still intact, except the presence of fine cracks. The
final crack pattern of the standard specimen (J1) is shown
in Fig. 13. After test termination, two longitudinal rebars
(one top and one bottom) were detected failing in tension.
4.2 Behaviour of GFRP-reinforced specimen
(J4)
The beam tip load-storey drift plot for the GFRP-
reinforced specimen is shown in Fig. 14. The first flexural
crack at the beam bottom adjacent to the column face was
detected at a drift angle of 1/1000 rad, corresponding to a
beam tip load of about 10.5 kN. A permanent beam-tip
 M. Nehdi, A. Said~Materials and Structures 38 (2005) 627-637
1 l:',q
e~
Storey drift (rad x10-2)
Fig. 12 - Beam tip load-storey drift relationship for the standard
steel-reinforced specimen J1.
Fig. 13 - Final crack pattern for the standard steel-reinforced
specimen J1.
deformation of 1.6 mm was measured after the two 1/400
rad drift cycles. As the test progressed, several distinct
cracks extended through the depth of the beam section at
specific locations corresponding to grid nodes in the
longitudinal reinforcement, while several smaller cracks
formed along the beam. This took place since GFRP bars,
which are originally cut from grids, are not deformed and
the bond with concrete is predominantly supplied by the
nodes. The onset of diagonal cracks in the joint area took
place at a beam tip load of 42 kN during the 1/50 rad drift
cycle. Additional cracks in the joint area appeared
thereafter as loading progressed, but remained within a very
fine width throughout the test. At a drift angle of 1/22 rad,
the beam became excessively cracked and rubble started
falling. Failure took place at the 1/20 rad drift angle in a
sudden and brittle manner when two of the beam's bottom
GFRP bars snapped in tension. The final crack pattern of
the GFRP-reinforced specimen (J4) is shown in Fig. 15.
4.3 Behaviour of hybrid-reinforced specimen
(JS)
The beam tip load-storey drift plot for the hybrid-
reinforced specimen is shown in Fig. 16. The first flexural
crack at the beam bottom adjacent to the column face was
detected at a drift angle of 1/1000 rad corresponding to a
633
o
Storey drift (rad xl0 -2)
Fig. 14 - Beam tip load-storey drift relationship for the GFRP-
reinforced specimen J4.
Fig. 15 - Final crack pattern for the GFRP-reinforced specimen
,14.
9.
e~
0)
Storey drift (rad xl 0-2)
Fig. 16 - Beam tip load-storey drift relationship for the hybrid-
reinforced specimen J5.
beam tip load of about 12.4 kN. Afterwards, several fine
flexura! cracks appeared along the beam and increased in
width as testing progressed. The onset of joint cracking
took place at a beam tip load of 40.2 kN pushing down.
Further cracks occurred in the joint area during the test and
were equally spaced across the joint panel. At the 1/67 rad
cycle, inclined cracks of shear-flexure were clear in the
beam. At the 1/33 rad cycle, apparent crushing of the
concrete was observed at the top fibre of the beam adjacent
634 M. Nehdi, A. Said/Materials
to the column face during pushing up. The load-drift angle
plot shows a considerable strength degradation occurring at
the 1/22 rad drift angle, corresponding to the start of the
GFRP beam bars slippage from the joint. After termination
of the test, clearing the failure area in the beam showed that
one out of three GFRP rebars failed in tension, whereas the
other two slipped out of the joint. This could be a motive to
study anchorage techniques for FlIP rebars to improve their
performance in similar applications. The final crack pattern
of the hybrid-reinforced specimen ( J 5 ) is shown in Fig. 17.
4.4 Load-storey drift angle envelope
relationship
For the tested beam-column joint specimens, the
envelopes of the beam tip load-storey drift angle
relationships are plotted in Fig. 18. The envelopes started at
comparable stiffness, but as soon as cracking took place a
distinct difference between the behaviour of the specimens
appeared and was significant for the remainder of the tests.
The three envelopes show a comparable ultimate load
capacity for all specimens, but the GFRP-reinforced
specimen exhibited lower stiffness, which is due to the
lower stiffness of GFRP compared to that of steel. The
GFRP-reinforced specimen (J4) had more than 10% a n d
23% lower total drift compared to that of the steel-
Fig. l 7 - Final crack pattern for the hybrid-reinforcedspecimenJ5.
-i Jr
Storeydrift(radxl0 -2)
Fig. 18 - Beam tip load-storey drift envelopes for the tested
specimens.
and Structures 38 (2005) 627-637
reinforced ( J 1 ) and the hybrid-reinforced ( J 5 ) specimens,
respectively. The steel reinforced specimen was able to
maintain a more stable post-yield load carrying capacity
compared to that of the other two specimens. The load drift
envelope of the hybrid-reinforced specimen ( J 5 ) showed a
typical bond slip characteristic as 2 out of the 3 GFRP
rebars slipped out of the joint, but the specimen maintained
the residual capacity intended in its design as the steel
reinforcement was mobilized. The GFRP-reinforced
specimen ( J 4 ) had an essentially elastic envelope, the steel-
reinforced specimen ( J 1 ) had a typical elastic-plastic
envelope, and the hybrid specimen ( J 5 ) had distinctive
changes in stiffness corresponding to the beam's steel
yielding and slip of GFRP rebars, respectively.
4.5 Cumulative dissipated energy
The capability of a structure to survive an earthquake
depends on its ability to dissipate the energy input from
ground motion. Despite the fact that energy input during a
ground movement event is difficult to estimate, a
satisfactory design should ensure a larger energy dissipation
capability of the structure than the demand. The cumulative
energy dissipated by the beam-column joint specimens
during the reversed cyclic load tests was calculated by
summing up the energy dissipated in consecutive load-
displacement loops throughout the test. The energy
dissipated in a cycle is calculated as the area that the
hysteretic loop encloses in the corresponding beam tip load-
displacement plot.
Fig. 19 shows plots of the cumulative energy dissipation
versus storey drift for the tested specimens. It can be
observed that the standard steel-reinforced specimen (./1)
had about 4 times higher cumulative energy dissipation
capacity at failure than that of the GFRP-reinforced
specimen (,]4). However, the cumulative energy dissipation
capacity at failure of the steel reinforced specimen was only
1.67 times that of the hybrid-reinforced specimen,
indicating that the steel reinforcement component of the
hybrid-reinforced specimen was able to supply a substantial
increase in energy dissipation. This is also clear from the
shape of the individual hysteretic loops of the tested
specimens (Figs. 12, 14 and 16) which are much wider for
the steel and hybrid-reinforced specimens. The ductility of
steel reinforcement allowed higher plastic deformations to
occur in the beam, thus increasing the area of each
"~. 250
----*---. J1 ( S t e e l ) i
200
.................... ----~---. J4 (GFRP) .......................................................................
.-~ i
~ .................................................
.... 9 -.- J 5 ( H y b r i d ) i ~'"'"
m
150
......................................................................................................
i .............. Y : ..............................................................................
[ ,o~ ,,..,A
100 .................................................
~.............................................
:~.~i................................
= ~,4~..:=::'2~.............................
-~ 50 .................................................
!...................::...............~zi...i..................
~ ........................
~....................................................
J 7 ~,~', ....... i ''''''~ i
c..)
3 6 9 12
Storeydrift(radxl0"2)
Fig. 19 - Cumulative energy dissipated for the tested specimens.
 M. Nehdi, A. Said~Materials and Structures 38 (2005) 627-637 635

individual hysteretic loop. The damage levels that the

specimens sustained at failure, shown in Figs. 13, 15 and
17, respectively indicate that while for the steel-reinforced
specimen extensive cracking in the beam hinge area helped
the specimen to dissipate energy, the GFRP-reinforced
specimen sustained severe but localized damage. Yielding
of steel is a major mechanism for RC structures to dissipate
energy, whereas plastic deformations and friction along 0
cracks in concrete usually have lower contribution to the
total energy dissipated.
Beam rotation (rad x 102)
4.6 B e a m m o m e n t - r o t a t i o n relationship
The rotation of the beam at a distance of 175 mm from Fig. 20 - Beam moment-rotation plot at 175 mm from the column
face for the steel-reinforced specimen J1.
the column face was calculated from measurements of two
LVDT's mounted on top and bottom of the beam,
respectively. This area of the beam adjacent to the column ~=.u ) ) ( i ( i ( ~ i

is subjected to the highest bending moment and thus

..~o...) .), ) ..)., .i,
....) ) ) ).
undergoes the highest level of deformation during cyclic ,.+so,--~........... )............~.......- - [ ...........i............T............)..
~. ..+oo...~ ..............i............... . . .............i..................t.................i..................i..................-..................i.................
loading. The rotation angle, (9, was calculated using the
following expression:
....
5o...
~~iiiii. i.................
i...................
i..................
)........
) i i ,-)-
0 - + (i)
d

where ~ is the elongation on the tensile face of the beam, ~;2 " ~ .... i i i i i i ! i i
is the shortening on the compressive face of the beam, and d Beam rotation (rad x 10-2)
is the vertical distance between the transducers. The beam
rotation angle at 175 mm from the column face, in rad, is Fig. 21 - Beam moment-rotation plot at 175 mm from the
plotted against the applied moment for the specimens ( J 1 ) , column face for the GFRP-reinforced specimen J4.
( J 4 ) and ( J 5 ) as shown in Figs. 20, 21 and 22, respectively.
A comparison of the beam moment-rotation plots shows
that for the 175 mm long segment of the beam from the 2 0 0 .... iiii!!!!!!iill !:.........
column face, specimen J 1 had a significantly lower rotation
--t50"
~. "-]' 0"~
0.... i i l ~ .......................................
before beam yielding (at similar bending moments)
compared to that of the beam in specimen J 4 . The lower
stiffness of GFRP rebars caused higher rotations in the
GFRP-reinforced beam at similar moments. However, the "-50"' =" ~................................................... § .............. i .................. [............... . ................ i"

predominantly elastic behaviour of GFRP resulted in very

9 4.50 ....
low residual deformations in the beam. Such lower residual
9~ 0 0 .... ,
deformations were viewed as an advantage by Fukuyama e t
al. [ 13] since it would make the rehabilitation of a damaged Beam rotation (rad x 10"2)
frame easier. For the hybrid-reinforced specimen, residual
deformations were very small until yielding started in the Fig. 22 - Beam moment-rotation plot at 175 mm from the
beam's steel reinforcement. As expected, the hybrid- column face for the hybrid-reinforced specimen J5.
reinforced specimen had a mixed behaviour manifested
through relatively higher rotations compared to that of the
steel-reinforced specimen, but lower rotations than that of 5. DISCUSSION
the GFRP-reinforced specimen (at similar bending
moments). The extent of damage in the beam for specimens
The use of FRP as reinforcement in concrete structures
( J 1 ) , ( J 4 ) a n d ( J 5 ) as shown in Figs. 13, 15 and 17,
has been increasing in popularity, yet various design
respectively indicates less spalling in the beam of the
guidelines and provisions still need to be developed for its
GFRP-reinforced specimen despite the large number of
safe implementation in large-scale field applications. For
cracks. Also, no GFRP stirrup failure was apparent in the
instance, the ACI 440.1R-01 identified a wide variety of
GFRP and hybrid-reinforced specimens, while extensive
research issues pertaining to FRP that need to be addressed,
deformations in the steel stirrups were observed in the steel-
some of which are as simple as establishing the statistical
reinforced specimen. This indicates that the amount of
variation of the tensile capacity of FRP rebars. Moreover, the
transverse reinforcement for the GFRP and hybrid-
contribution of FRP transverse reinforcement to the shear
reinforced specimens, which conformed to CSA $806-02,
capacity of RC elements needs to be properly evaluated. The
was sufficient to prevent shear failure in the beams.
lower bond strength of FRP to concrete compared to that of
636 M. Nehdi, A. Said / Materials and Structures 38 (2005) 62 7-63 7
steel imposes difficulties in design, for instance in satisfying
rebar development length such as in the case of beam
reinforcement anchorage in exterior joints, for which using
FRP would require additional embedded length compared to
when steel rebars are used. Also the difficulty of
manufacturing bends in FRP makes it difficult to adopt this
material in reinforcing structurally complicated
configurations and needs to be addressed.
A major drawback of steel-free FRP-RC systems is their
low energy dissipation under earthquake loading, as
demonstrated by the performance of the tested FRP-
reinforced joint specimen (J4). The energy input from ground
motion is equal to the sum of potential, kinematic, damping
and hysteretic energy components [22]. The potential and
kinematic energy components vanish after the static
equilibrium of the structure is reached, while the damping
and hysteretic energy components are responsible for energy
dissipation. The hysteretic component becomes the major
contributor to energy dissipation when significant inelastic
deformations take place. Hence, an FRP-reinforced frame
may have to be designed with a high damping component so
that when added to its relatively limited hysteretic
Component, it can dissipate the energy input during an
earthquake. Design guidelines for framed RC buildings by
the Architecture Institute of Japan, as outlined by Kobayashi
et al. [23], entail ensuring seismic performance by
overcoming the ductility deficiency of FRP-RC frames. The
study recommended the use of the capacity spectrum
method. Performance demand and capacity spectra were
evaluated and a performance point, where the demand and
capacity spectra meet and members are still below their
flexural capacity, was defined as the safety limit. This
performance-based design approach was successfully applied
to the analysis of a 9-floor FRP-RC frame. The study also
pointed out the cruciality of damping in FRP-RC structures
and recommended the use of vibration control devices.
The use of hybrid steel-FRP RC systems could address
many of the drawbacks of steel-free RC systems. Steel
reinforcement can be used in lateral load resisting structural
members, which are not usually exposed to aggressive
media, while FRP reinforcement can be used in the
envelope of the structure to enhance durability.
Alternatively, a hybrid reinforcement configuration can
make use of the corrodible steel at a thick concrete cover,
while the more durable FRP stays at a minimum cover.
Thus, the structure can benefit from using such a hybrid
reinforcement system to provide both durability (using
FRP) and post-peak reserve strength (using steel).
The present study focussed only on comparing the
behaviour of FRP, hybrid steel-FRP, and steel-reinforced
beam-column joints. Full-scale tests on entire FRP and
hybrid-reinforced frames need to be performed to assess the
progress of failure globally. The results can be used to
calibrate numerical models that can be used to simulate the
behaviour of multi-storey FRP and hybrid-reinforced
frames with high degrees of redundancy, and accordingly
predict the progress of failure. Moreover, passive energy
dissipation devices can provide a source of energy
dissipation for FRP-reinforced frames, which needs further
focussed research. Overall, research efforts are still needed
to address many questions and uncertainties, and to develop
adequate design provisions dedicated to steel-free and
hybrid RC systems, before their widespread use in
demanding large-scale structural applications becomes
feasible and safe in seismic areas.
6. CONCLUSIONS
An effort was made to investigate the performance of
GFRP and hybrid steel-GFRP-reinforced beam-column
joints and to compare their behaviour to that of standard
steel-reinforced beam-column joints under reversed quasi-
static (cyclic) loading. Based on experimental observations
and analysis of test results, the following conclusions can
be drawn:
9 The GFRP-reinforced beam-column joint showed very low
plasticity features when tested under reversed cyclic loading.
This resulted in lower energy dissipation compared to that of
the steel and hybrid reinforced specimens.
9 The hybrid GFRP-steel-reinforced beam-colunm joint
showed lower stiffness than that of the conventional steel-
reinforced beam-column joint, but exhibited higher stiffness
than that of the GFRP-reinforced specimen.
9 The GFRP and hybrid-reinforced specimens showed
satisfactory drift capacity, assuming a minimum drift
requirement of 3% (0.03 rad) as recommended in the
literature for ductile RC flame buildings [24].
9 A hybrid RC system could be tailored to provide a range of
performance requirements such as durability, stiffness,
strength, ductility, etc. A designer may adapt the
reinforcement configuration of the hybrid system to
accommodate a balance between such design criteria.
9 This study was only focussed on the level of the
subassemblage. A more global concept should be adopted in
the design of moment-resisting frames. Thorough dynamic
analysis of GFRP and hybrid-RC structures should be
performed to better assess their capacity in meeting seismic
resistance requirements.
9 Design code provisions for the seismic design of RC
structures, which have been developed for ductile steel
reinforcement, need to be re-evaluated for FRP-RC
structures.
REFERENCES
[1] Sugita,M., 'NEFMAC grid type reinforcement', in 'Alternative
Materials for the Reinforcing and Prestressing of Concrete',
Clarke, J.L. (Ed.) (Blackie Academic and Professional, UK,
1993) 55-82.
[2] Brown,V.L. and Bartholomew, C.L., 'FRP reinforcing bars in
reinfoI~cedconcrete members', A CI Mat. J. 90 (1) (1993) 34-39.
[3] Nanni,A., 'Flexural behavior and design of RC members using
FRP reinforcement', J. Struct. Engrg. 119 (11) (1993) 3344-
3359.
[4] ACI Committee 440, 'Guide for the design and construction of
concrete reinforced with FRP bars - ACI 440.1R-01, ACI,
Detroit, 2001.
[5] CSA $806-02, 'Design and constructionof building components
using fibre-reinforced polymers', Canadian Standards
Association, Ontario, Canada, 2002.
 M. Nehdi, A. Said / Materials and Structures 38 (2005) 62 7-63 7
[6] ISIS Canada, 'Reinforcing concrete structures with fibre
reinforced polymers', Manual No. 3,2001.
[7] JSCE, 'Recommendations for the design and construction of
concrete structures using continuous fiber reinforcing materials',
Concrete Eng. Series 23, 1997.
[8] Canadian Highway Bridge Design Code (CHBDC), 'Fibre
reinforced structures', Sec. 16, 1998, 687-705.
[9] Nagasaka, T., Fukuyama, H. and Tanigaki, M., 'Shear
performance of concrete beams reinforced with FRP stirrups', in
'Int. Sym. on Fiber Reinforced Plastic Reinforcement for
Concrete Structures', ACI SP-138 (Vancouver, 1993) 789-811.
[10] Alsayed, S.H., A1-Salloum, Y.A. and Almusallam, T.H., 'Shear
design of GFRP bars', in '3 rd Int. Sym. on Non-Metallic
Reinforcement for Concrete Structures', FRPRCS-3, (Sapporo,
Japan, 1997) 285-292.
[11] Shehata, E.F.G., 'Fibre-reinforced polymer (FRP) for shear
reinforcement in concrete structures', Ph.D. Thesis, University
of Manitoba, Canada, 1999.
[12] Grira, M. and Saatcioglu, M., 'Reinforced concrete columns
confined with steel or FRP grids', in 'The 8th Canadian Conf. on
Earthquake Engineering', (Vancouver, 1999), 445-450.
[13] Fukuyama, H., Masuada, H., Sonobe, Y. and Tanigaki, M.,
'Structural performance of concrete flames reinforced with FRP
reinforcement', in '2ndInt. RILEM Sym. on Non-Metallic (FRP)
Reinforcement for Concrete Structures', FRPRCS-2, (Ghent,
Belgium, 1995), 275-286.
[14] Aiello, M.A. and Ombres, L., 'Structural performances of
concrete beams with hybrid (fiber-reinforced polymer-steel)
reinforcements', J. ofComp.for Const. 6 (2) (2002) 133-140.
[15] Leung, H.Y. and Balendran, R. V., 'Flexural behaviour of
concrete beams internally reinforced with GFRP rods and steel
rebars', Structural Survey 21 (4) (2003) 146-157.
[16] Tamzus, V. and Tepfers, R., 'Ductility of a non-metallic hdfbrid
fibre composite reinforcement for concrete', in 'the 2n Int.
637
RILEM Sym. on Non-Metallic (FRP) Reinforcement for Concrete
Structures', FRPRCS-2, (Ghent, Belgium, 1995), 18-25.
[17] Bakis, C.E., Nanni, A., Terosky, J.A. and Koehler, S., 'Self-
monitoring, pseudo-ductile, hybrid FRP reinforcement rod for
concrete application', Comp. Sci. andTech. 61 (6) (2001) 815-823.
[18] Harris, H.G., Somboonsong, W. and Ko, F.K., 'New ductile
hybrid FRP reinforcing bar for concrete structures', J. of Comp.
for Const. 2 (1) (1998) 28-37.
[19] Belarbi, A., Chandrashekhara, K. and Watkins, S.E.,
'Performance evaluation of fibre reinforced polymer
reinforcing bar featuring ductility and health monitoring
capacity', in 'The 4th Int. Sym. on FRP Reinforcement for
Concrete Structures', FRPRCS-4 (Baltimore, 1999) 1-12.
[20] CSA A23.3-94: 1994, 'Design of Concrete Structures',
Canadian Standards Association, Rexdale, Ontario, Canada.
[21] Joint ACI-ASCE Committee 352: 2002, 'Recommendations
for Design of Beam Column Connections in Monolithic
Reinforced Concrete Structures', AC1352R-02, ACI, Detroit.
[22] Priestly, M.J.N., Seible, F. and Calvi, G.M., 'Seismic Design
and Retrofit of Bridges', 1st Edn, (Wiley Interscience, New
York, 1996).
[23] Kobayashi, K., Fukuyama, H., Fujisaki, T., Fukai, S. and
Kanakubo, T., 'Design Practice of Framed Building Structures
Based on AIJ Design Guideline 2002', in '6 th Int. Sym. on
Fiber Reinforced Polymers in Concrete Structures', FRPRCS-
6, (Singapore, 2003), 1435-1444.
[24] Corley, W.G., 'Ductility of columns, walls, and beams-how
much is enough?', in 'The Thomas Paulay Sym., Recent
Developments in Lateral Force Transfer in Buildings', ACI SP-
157 (La Jolla, California, Sept. 20-22 1993) (1995) 331-350.
[25] NEFCOM Corporation, 'Mechanical Properties ofNEFMAC',
(Tokyo, 1996).
[26] Hughes Brothers Inc., 'GFRP ASLAN 100 Product
Specification', (Seward, Nebraska, 2001).