Fiber Reinforced Polymer (FRP) reinforcement, in the form of longitudinal and
transverse reinforcement, are currently being developed for use in new buildings
and bridges (Nanni, 2001). The major driving force behind
this development is the superior performance of FRPs in corrosive environments
(El-Salakawy et al., 2005). FRP reinforcement
has high strength-to-weight ratio, favorable fatigue strength, electro-magnetic
transparency and low relaxation characteristics when compared with steel reinforcement,
offering a structurally sound alternative in most applications. However, FRP
reinforcement shows linear stress-strain characteristics up to failure, without
any ductility. The experimental results conducted by Deitz
et al. (2003) showed that even though some GFRP rebar experienced
the ultimate compressive strength approximately 50% of the ultimate tensile
strength, the main failure mode was due to low ductility of these bars. This
low ductility characteristic of FRP bars poses serious concerns about their
applicability to earthquake resistant structures, where seismic energy is expected
to be dissipated through inelasticity in members.
Several models were suggested to apply for steel reinforced concrete members
such as confined columns for increasing the ductility of those members (Pauly
and Priestly, 1991), these models can not directly applied to FRP reinforced
members due to non-ductility of this material. FRP bars have different ductility
because of different fiber material, new bars were developed to show higher
ductility for flexural and shear dominated concrete members (Benmokrane
et al., 2002; Tim and Chris, 2003; Toutanji
and Saafi, 2000). Experimental test results have shown that a concrete beam
reinforced with AFRP bars becomes more flexible in the postcracking range and
so a method has been suggested to provide a meaningful quantification of ductility
for FRP-reinforced beams (Rashid et al., 2005).
The interface behavior of rebars to concrete, bond, was an important issue that
were carried and considered in the last decade to show different influential
parameters (Maria Antonietta et al., 2007; Focacci
et al., 2000; Pecce et al., 2001).
Sharbatdar (2003) developed an analytical model for hysteretic
moment-displacement relationship for FRP reinforced columns under constant axial
loading. The primary moment-flexural displacement relationship defines the strength
boundary and initial stiffness. The main difference of the model, as opposed
to the hysteretic models for steel reinforced concrete members, is that the
unloading branches of hysteretic loops aim towards the origin of moment-displacement
relationship (zero point) due to the elastic behavior of FRP bars in tension.
Therefore, inelastic response of FRP reinforced structures can be analyzed by
using appropriate models. In order to prevent brittle failure, concrete crushing
can be obtained at ultimate prior to the tension failure of FRP provided that
the concrete is confined sufficiently and that the section is over-reinforced
as opposed to conventional design concept used for steel reinforced sections
(Sharbatdar, 2003; Deitz et al.,
Experimental research presented in this study has been underway at the
Structures Laboratory of the University of Ottawa to investigate seismic
performance of FRP reinforced concrete structural elements. Large scale
columns and beams have been tested under simulated seismic loading.
There are uncertainties about the FRP use in new construction; their elongation
before the material failure is very small which causes brittle failure. Therefore,
codes on FRP reinforced structures, such as Canadian Standards Association CSA
806-02 (2002), Canadian Network of Centres of Excellence on Intelligent Sensing
for Innovative Structures (2002) and American Concrete Institute ACI-440 (2001)
limited the replacement of steel reinforcement by FRP reinforcement only as
flexural reinforcement in beams and shear reinforcements (Nanni,
Ductility gains importance especially for seismic design of structures.
Therefore, investigations on the inelastic behavior of the FRP reinforced
sections under cyclic loading are needed in order to understand the overall
seismic behavior of the FRP reinforced structures. The inelastic analysis
of the structures generally requires complex and difficult calculations.
With the advent of computers, several programs have been developed for
nonlinear dynamic analysis of steel reinforced concrete structures. However,
there is still need for computer programs for the dynamic nonlinear analysis
of the FRP reinforced concrete structures.
The experimental program consists of two types of reinforced concrete
elements; (i) square columns and (ii) rectangular beams. They represent
portions of column and beam elements between rigidly attached adjoining
members and the points of contraflexure, as cantilever specimens. The
specimens were reinforced with carbon FRP bars and carbon FRP grids as
longitudinal and transverse reinforcement, respectively.
The columns had a 355 mm square cross section with either a 1900 mm height,
which resulted in a 2180 mm shear span when measured form the point of application
of lateral load, or 1000 mm height and 1280 mm shear span to increase the imposed
|| Details of column specimens
||Properties of test specimens
This implies that the two longer columns discussed in the study would behave
predominantly in the flexure mode and the shorter column would develop significant
shear stress reversals. Figure 1 shows the geometric details
of specimens. The columns were reinforced with 12-9.5 mm diameter carbon FRP
bars, resulting in 0.7% longitudinal reinforcement ratio. The longitudinal bars
were extended into the footing by 470 mm. Carbon fibre grids were used as column
ties. The grids had nine cells and were manufactured from 6x8 mm rectangular
FRP bars with overlapping fibres at intersecting joints. They had a square configuration
with an out-to-out dimension of 300 mm. One longitudinal bar was placed in each
perimeter corner. The grid spacing was either 88 or 175 mm. Table
1 provides a summary of column properties for the column specimens discussed
in this study. The beams had 305 mm width and 405 mm depth. They either had
a cantilever length of 1900 mm and a shear span of 1780 mm, or a length of 1000
mm and a shear span of 870. The beams were reinforced asymmetrically to simulate
the actual arrangement used in practice. Accordingly, top and bottom reinforcement
consisted of 6 and 4-9.5 mm diameter carbon FRP bars, respectively, resulting
in 0.39 and 0.26% tension reinforcement ratios in the strong and weak directions,
respectively. Carbon fibre grids were used as stirrups. The grids had two cells
and were manufactured using 6x8 mm rectangular FRP bars with overlapping fibers
at intersecting joints.
||Details of beam specimens
The perimeter dimensions of grids were 250x350 mm and the resulting total area
of transverse reinforcement effective against shear (in the direction of loading)
was 144 mm2. The grid spacing was either 90 or 180 mm. Figure
2 and Table 1 provide the details of beam properties.
Two different batches of Normal Portland Cement concrete were used to cast
the columns and beams. Separate batches of concrete were used to cast the footings,
which were used to secure the specimens to the laboratory strong floor. Concrete
strengths during the time of testing were 37 and 40 MPa for the columns and
beams, respectively. The longitudinal reinforcement for all specimens was manufactured
by Pultrall Inc. with a nominal diameter of 9.5 mm. They were made from high
strength carbon fibers and extremely durable vinyl ester resin. The bar surface
was sand-coated for improved bond. Sand coating increased the bar diameter to
approximately 12 mm. Coupon tests were conducted to establish the stress-strain
relationship in tension. The relationship indicates linear behavior with an
average tensile strength of 1450 MPa and a modulus of elasticity of 122,000
MPa. The stress-strain relationship of FRP bars in compression was difficult
to establish by tests because of the possibility of encountering stability failure.
Therefore, short samples, having lengths equal to 2-5 times the bar diameter
were tested under direct compression until failure. The average modulus of elasticity
in compression was found to be 23,000 MPa, which was approximately 20% of the
value in tension. The failure stress in compression varied between 240 and 310
MPa. These values correspond to 16-21% of tensile strength. The failure strain
in compression varied between 1 and 1.3%. The failure in direct compression
was caused either by delamination of fibers and crushing of resin or by splitting
of bars longitudinally. The NEFMAC grids used as transverse reinforcement were
manufactured by forming flat grids through a pin-winding process, similar to
filament winding. The product used in the current phase of experimental research
was reported to have a specific gravity of 1.4 t m-3 and a modulus
of elasticity of 100,000 MPa, by the manufacturer. The grids used as column
ties had a square configuration. These grids were manufactured using cross FRP
bars, each having a 6x8 mm rectangular cross-section, forming nine equal-size
square openings. The grids used as beam stirrups had two rectangular cells,
simulating a perimeter hoop with a crosstie. The stress-strain relationships,
established through coupon tests under direct tension indicate a maximum tensile
strength of 1230 MPa and an elastic modulus of 76,335 MPa. These grids are known
to have weak joints against maintaining the stability of compression bars against
buckling, there was no such failure observed during the column tested reported
in this study. The test setup consisted of a horizontal actuator to apply the
lateral load, supported by two steel A-frames and two vertical actuators for
the application of axial compression on the columns.
|| Test setup
||The deformation history for lateral loading
Figure 3 shows the test setup. The same setup was used for
both the columns and beams. Therefore, the beams were tested vertically. The
axial load was applied first and was maintained at a constant level through
the test. The horizontal load was applied in the deformation control mode. Lateral
deformation reversals were applied with three cycles at each of the incrementally
increasing drift level. The loading program followed is shown in Fig.
Force-deformation hysteretic characteristics of columns and beams, showing
their strengths and deformability, are expressed in term of hysteretic relationships.
Figure 5 and 6 show experimentally recorded
hysteretic moment-drift relationships for all the specimens. Figure
4 indicates that columns CFCL3 and CFCL4 both developed a 2% lateral drift
ratio with little strength degradation. Both columns showed flexure dominant
response. Column CFCL4, with closely spaced grids and approximately 62% of the
confinement reinforcement required by CSA S806-02, showed increased deformability,
with lateral drift ratios reaching up to 3 at 20% strength decay in the direction
of first load excursion and 25% decay in the opposite direction. The column
was able to sustain 4% lateral drift after 30 and 37% strength decay in the
two directions, respectively. The column failed during the second cycle at 5%
lateral drift when the compression bars showed local fibre buckling leading
to bar failure in compression. Maximum strains recorded in longitudinal bars
were 0.93 and 0.55% in tension and compression, respectively. The grids developed
0.5% strain in tension, a value higher than the conservative limit of 0.4% assumed
in CSA S806-021 for design. Column CFCL3, with a wider grid spacing of 175 mm
and about 31% of the confinement reinforcement required by CSA S806-021 showed
brittle behavior shortly after 2% lateral drift, experiencing about a 50% drop
in moment resistance at the end of 3% drift cycles.
||Experimentally recorded column moment-lateral drift hysteretic
The maximum strains recorded in longitudinal bars were 0.76 and 0.50% in tension
and compression, respectively. The confinement mechanism could not be fully
activated in this column because of the wide spacing of grids. The recorded
tensile strain in the grids was limited to 0.31%. Column CFCL8 was companion
to column CFCL3, except for its shorter length of 1000 mm. This resulted in
a shear span of 1280 mm and associated increase in shear force reversals. The
column was confined with 9-cell CFRP grids. This resulted in 192 mm2
of transverse reinforcement in each cross-sectional direction. This amount was
equal to 43% of the amount required by CSA S806-022 for concrete
confinement. The grid spacing was 175 mm, which was approximately twice the
maximum spacing permitted by CSA S806-02. The column was subjected to a constant
compressive force of 30% of its concentric capacity. The hysteretic relationship
shown in Fig. 5 indicates that the column showed stable hysteresis
loops up to 2% drift ratio, but developed severe strength decay during the second
cycle at 3% drift. The column failed during pulling at the second cycle of 3%
drift due to the instability of FRP bars in compression and crushing of concrete.
The transverse reinforcement provided was sufficient to prevent shear (diagonal
tension) failure, though the column showed increased diagonal cracking as compared
to the companion column with a longer height (CFCL3). Strain gauge readings
indicated that the FRP bars experienced a maximum of 0.84 and 0.63% strains
in tension and compression at the end of test, respectively. The FRP grids experienced
a maximum of 0.56% tensile strain, which is 80% higher, due to the increased
shear applied on the column, than that recorded in column CFCL3.
The beam hysteretic relationships are shown in Fig. 6.
Beams CFB4 and CFB5 were companion specimens with approximately 1.8 m
shear span, except for the grid spacing. CFB4 had 180 mm grid spacing,
corresponding to d/2 and CFB5 had 90 mm grid spacing, corresponding to
d/4. Both beams were designed to experience tensile rupturing of FRP bars
at or shortly after the onset of concrete crushing. The hysteretic relationships
indicate essentially elastic response, with gradual degradation of effective
elastic stiffness due to progressive cracking under reversed cyclic loading.
While both beams experienced 3% drift in the strong direction, the drift
capacity was limited to 3 and 2% for CFB4 and CFB5, respectively, in the
weak direction. The beams experienced softer response, as compared to
the columns discussed earlier, because of the absence of accompanying
axial compression. CFB4 experienced significant diagonal cracking on the
side faces, in addition to progressively increasing flexural cracking.
However, the diagonal cracking was controlled more effectively in CFB5
due to the reduced spacing of transverse grid reinforcement. This also
resulted in controlled softening of CFB5 in the strong direction, with
185 kN m of moment resistance at 3% drift, as compared to 156 kN m moment
resistance in CFB4 at the same levels of lateral drift. The strain readings
indicated 0.0125 and 0.0120 tensile strains in the longitudinal reinforcement
of CFB4 and CFB5, respectively, before rupturing in tension at about 4%
lateral drift. The maximum compressive strain readings in FRP longitudinal
bars were limited to 0.0038 and 0.0028 in CFB4 and CFB5, respectively.
The transverse grid reinforcement developed maximum tensile strains of
0.0044 to 0.0040. Beam CFB2 had reduced shear span and increased shear
stress reversals. However, the beam was reinforced for shear with closely
spaced grid reinforcement, resulting in 55% higher area of shear reinforcement
than that recommended by CSA S806-021. The reduction in grid spacing increased
the effectiveness of shear resistance while enhancing concrete confinement.
Hence, the beam was able to develop its flexural capacity without a premature
shear failure. The hysteresis loops showed near elastic behaviour with
some stiffness degradation due to concrete cracking, up to 3% drift ratio.
Degradation of strength began at the first cycle of 3% drift and continued
until 4% drift ratio. At this level of deformation, all the 4 positive
moment bars ruptured in tension causing the beam to loose its flexural
resistance in the weak direction. The beam was able to resist severely
reduced loading in the strong direction when pushed to 5% drift before
rupturing all the negative bars.
Flexural capacities of FRP reinforced concrete beams were computed using
the plane section analysis commonly employed for steel reinforced concrete members
in flexure. The analysis was conducted to establish moment-curvature relationships.
Experimentally recorded moment-curvature relationships were also obtained for
comparison. The experimental relationships were computed form strain gauge data
recorded at the beam critical section, as well as LVDT readings within the critical
beam region, giving average strains over a gauge length of 300 mm. The curvatures
obtained from LVDT readings provided slightly lower values than those obtained
by strain gauges, because they represented average curvatures over the gauge
length, rather than the maximums recorded at the critical section.
||Experimentally recorded beam moment-lateral drift hysteretic
Comparisons of moment-curvature relationships, shown in Fig.
7, indicate that the design approach employed for conventional steel reinforced
concrete beams can be used for the design of FRP reinforced concrete beams,
as recommended by CSA S806-02 (2002).
|| Moment-curvature relationships for FRP reinforced concrete
||Axial force-moment interaction diagrams for FRP reinforced
The applicability of the plane-section analysis to FRP reinforced column
design was also verified. Column axial force-moment interaction diagrams
were constructed twice, first using unconfined concrete and secondly the
confined concrete model the results plotted in Fig. 8
include experimentally obtained column capacities for two of the columns.
The figure clearly shows the applicability of conventional plane section
analysis to the computation of axial force-moment interaction diagrams
for FRP reinforced concrete elements.
Two design approaches may be used to design FRP reinforced concrete elements.
One is to allow sufficiently high safety margin in design against tensile
rupturing of bars and ensuring elastic behaviour during service loads
and accepting brittle failure under ultimate load conditions initiated
either by FRP rupturing in tension or concrete crushing in compression.
The other approach is to reduce the safety margin while maintaining elastic
behaviour under service loads and over-reinforcing members with FRP reinforcement
to prevent bar rupturing at ultimate and promoting ductile failure of
compression concrete at ultimate with appropriate concrete confinement.
Three different flexural modes of failure were observed in the specimens
tested; (i) crushing of concrete due to compression failure of concrete
(ii) stability failure of FRP bars in compression and (iii) rupturing
of tension reinforcement. The failure of columns under high axial compression
and reversed cyclic loading was initiated by the crushing of cover concrete,
followed by the buckling of FRP bars in compression and subsequent crushing
of core concrete. Tension rupturing of FRP reinforcement occurred in beams,
which were not over-reinforced. The type of tension failure was very similar
to that observed in coupon tests, where the failure took place within
approximately 150 mm segment of the bar, with fibers rupturing randomly
at different locations, displaying clear delamination of ruptured fibers
from the resin.
While both design philosophies; (i) allowing bar rupturing at ultimate
but providing increased safety margin in design and (ii) over-reinforcing
members while confining their compression concrete to promote concrete
crushing at ultimate prior to bar rupturing, may be defended for design
under gravity and wind loading. During a strong earthquake, however, because
of the uncertainties associated with seismic design loads, it may be prudent
not to allow bar rupturing under any condition. This implies that the
members should be confined with properly designed transverse reinforcement
in all critical regions. An important aspect of earthquake resistant design
is the dissipation of seismic induced energy through significant yielding.
This is not possible in FRP reinforced concrete, though some dissipation
of energy can be achieved through concrete confinement, as it is also
done in steel reinforced concrete columns subjected to high axial compression,
above the balanced load level. On the other hand, FRP reinforced concrete
exhibits a softer response within the effective elastic range, with extensive
cracking of concrete, because of low elastic modulus of FRP reinforcement.
This results in increased deformability, with elastic deformations approaching
to similar levels as those attained in steel reinforced concrete elements
beyond yielding. It also results in longer vibration periods in buildings,
with potentially reduced seismic force demands. Figure 9
shows a design response spectrum and the estimates of a typical low-rise
concrete building with steel and FRP reinforcement, qualitatively. As
demonstrated, elastic seismic force demands in FRP reinforced concrete
structures may be substantially lower, depending on the characteristics
of the design response spectrum and the dynamic characteristics of buildings.
||Effect of the lengthening of period on design force levels
The several conclusions can be drawn from the research presented in this
study. FRP reinforced concrete structural elements can be designed for
flexure, using the plane section analysis employed for steel reinforced
concrete elements, to sustain the required levels of seismic forces.
Tests of large-scale structural components under reversed cyclic loading
indicate that FRP reinforced concrete beams can attain a lateral drift
ratio of up to 3%, while essentially remaining elastic, though softened
due to cracking. FRP reinforced concrete columns under approximately 30%
of their concentric capacity can develop 2-3% lateral drift ratios, depending
on the level of concrete confinement. The deformability of these elements
is sufficiently high to suggest that they can be employed in seismically
Earthquake resistant design of FRP reinforced concrete structures may
be based on elastic member behaviour while taking advantage of relative
flexibility of FRP material and also FRP reinforced concrete and associated
elongations in the fundamental period of structures, attracting lower
More experimental tests including large scale external beam-column concrete
joints reinforced with FRP grids and rebars have been conducted at the
same lab to investigate the seismic behaviour of joints at this kind of
new concrete structures and also generate more data in this regard.
The research reported in this study was made possible by the financial
supports from the University of Ottawa. The authors would like to express
their gratitude to organization and staffs for their financial and technical