ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 99-S8

Seismic Behavior of Concrete Columns Confined with

Steel and Fiber-Reinforced Polymers
by Shamim A. Sheikh and Grace Yau

Results from an experimental program are presented in which 12

356 mm diameter and 1473 mm long columns were tested under
constant axial load and reversed cyclic lateral load that simulated
forces from an earthquake. Each specimen consisted of a column
cast integrally with a 510 x 760 x 810 mm stub that represented a
beam-column joint area or a footing. The test specimens were
divided into three groups. The first group consisted of four columns
that were conventionally reinforced with longitudinal and spiral
steel reinforcement. The second group contained six reinforced
concrete columns that were strengthened with carbon fiber-rein-
forced polymers (CFRP) or glass fiber-reinforced polymers
(GFRP) before testing. The last group included two columns that
were damaged to a certain extent, repaired with fiber-reinforced
polymers (FRP) under axial load, and then tested to failure. The
main variables investigated were axial load level, spacing of spirals,
thickness, and type of FRP. From the results of the tests, it can be
concluded that carbon and GFRP can be used effectively to
strengthen deficient columns such that their behavior under simu-
lated earthquake loads matches or exceeds the performance of
columns designed according to the seismic provisions of the 1999
ACI Code. The use of FRP significantly enhances strength, ductility,
and energy absorption capacity of columns.

Keywords: column; concrete; ductility; polymer; strength. Fig. 1—Details of test specimen.

umns with that of conventionally reinforced columns. A stan-

INTRODUCTION AND RESEARCH SIGNIFICANCE
dard lateral load sequence that simulated earthquake forces
Repair, rehabilitation, and strengthening of existing struc-
was used for all the columns. The same loading sequence
tures has become a major part of construction activity in
North America. By some estimates, the money spent on ret- was used for testing of over 60 similar steel reinforced con-
rofitting of existing structures in recent years has exceeded crete columns with square and rectangular cross sections.3,4
that spent on new structures. There are more than 200,000 A direct comparison can thus be made between the perfor-
bridges in North America that represent approximately 40% mance of all the columns.
of the available inventory that are deemed deficient.1 Some
of these deficient bridges are damaged, while others need EXPERIMENTAL PROGRAM
strengthening because either the design codes have changed, A total of 12 specimens were tested. Each specimen con-
making these structures substandard, or larger loads are per- sisted of a 356 mm diameter and 1.47 m long column cast in-
mitted on the road. A similar scenario of other components tegrally with a 510 x 760 x 810 mm stub. All columns were
of infrastructure such as airports and parking garages. exists tested under lateral cyclic loading while simultaneously be-
where extensive retrofitting is required. Procedures that are ing subjected to constant axial load throughout the test. The
technically sound and economically feasible are needed to layout of the specimen is shown in Fig. 1. The column rep-
upgrade deficient structures. Traditional techniques that em- resented the part of a bridge column or a building column be-
ploy materials such as steel and cementitious composites tween the section of maximum moment and the point of
have been used successfully for many applications, but have contraflexure. The stub represented a discontinuity, such as
not proved very durable in many cases. For certain applica- a beam column joint or a footing. In all specimens, the ratio
tions, the traditional techniques are very cumbersome and of the core area measured to the centerline of spiral to the
expensive. In the research presented herein,2 relatively new
gross area of the column section was kept constant at 74%,
materials, fiber-reinforced polymer (FRP), have been used to
which is similar to that used in previous tests.3-5
retrofit circular columns. Continuous fibers of carbon or
glass were used in a circumferential direction to confine the
columns. The main purpose of this study was to evaluate the ACI Structural Journal, V. 99, No. 1, January-February 2002.
MS No. 01-028 received February 8, 2001, and reviewed under Institute publica-
effectiveness of FRP reinforcement in strengthening defi- tion policies. Copyright © 2002, American Concrete Institute. All rights reserved,
cient columns or repairing damaged columns. This was including the making of copies unless permission is obtained from the copyright pro-
prietors. Pertinent discussion will be published in the November-December 2002 ACI
achieved by comparing the behavior of FRP-retrofitted col- Structural Journal if received by July 1, 2002.

72 ACI Structural Journal/January-February 2002

Shamim A. Sheikh, FACI, is a professor of civil engineering at the University of Tor-
onto, Ontario, Canada. He is Chair of ACI-ASCE Committee 441, Reinforced Concrete
Columns, and is a member of ACI Committee 374, Performance-Based Seismic Design
of Concrete Buildings. His research interests include earthquake resistance and seismic
upgrade of concrete structures, confinement of concrete, use of FRP in concrete struc-
tures, and expansive cement and its applications. In 1999, he received the ACI Struc-
tural Research Award for a paper on the design of ductile concrete columns.

Grace Yau works as a structural engineer with Weiskopf and Pickworth, LLP, Con-
sulting Engineers in New York City. She received her Masters of Applied Science in
1998 from the University of Toronto.

Fig. 3—Tensile force-strain curves for FRP composites.

Fig. 2—Reinforcing cages of specimens. Fig. 4—Tensile stress-strain curves for reinforcing steel bars.

Table 1 gives the details of the test specimens. All the col- tained the same amount of spiral reinforcement as Specimens
umns contained six 25M (500 mm2) longitudinal steel bars, S-3NT and S-4NT; however, they were strengthened with
and the spirals were made of U.S. No. 3 (71 mm2) bars. The GFRP or CFRP before testing. Specimens ST-1NT to ST-6NT
reinforcement for the stub consisted of 10M (100 mm2) hor- fall in this group. The third group, Series R, included Specimens
izontal and vertical stirrups at 64 mm spacing. In addition, R-1NT and R-2NT that contained 50% less spiral reinforce-
10M bars with 135 degree hooks were placed at the top and ment compared with Specimens S-1NT and S-2NT. These
bottom of the stub at the same spacing (Fig. 2). The longitu- two columns were damaged to a certain extent under axial
dinal bars in the columns were completely extended into the and lateral loads, repaired under axial load with FRP, and
stub, whereas the spiral reinforcement was extended into the then tested to failure.
stub for 100 mm. The design of the specimens aimed at For Specimens ST-1NT and ST-2NT, the FRP composite
forcing the failure in the potential plastic hinge region of the was wrapped within the potential plastic hinge zones of the
column, that is, within a length of 800 mm from the face of columns, that is, for a length of approximately 800 mm start-
the stub. The length of 800 mm was chosen based on previ- ing from the stub face, and the failure occurred in the test
ous tests3,4 where it was observed that the length of the most zone. During the testing of Specimen ST-3NT, however,
damaged region of the column was approximately equal to crushing of concrete was observed outside the test region;
the section depth and located approximately 100 to 200 mm therefore, to ensure that the failure took place within the
away from the stub. Outside the test region, the spacing of plastic hinge zone, it was decided to wrap the whole column
spiral reinforcement was reduced to around 2/3 of the speci- for the rest of the specimens. Column ST-6NT was strength-
fied spacing in the test zone (Fig. 2). All specimens were cast ened with four 100 mm wide CFRP bands at a clear spacing
together in vertical positions. of 100 mm. The first band was applied at a distance of 50 mm
The test specimens are divided into three groups. The from the stub face. The glass fabric was 1.25 mm thick,
first group, Series S, consisted of columns S-1NT, S-2NT, whereas the carbon fabric was either 0.5 or 1.0 mm thick. The
S-3NT, and S-4NT. Only steel spirals were used as lateral re- type of fabric and the number of layers used were designed to
inforcement in these columns. Specimens S-1NT and S-2NT study a range of parameters for their effects on column behavior.
contained the amount of spiral reinforcement that satisfied
the 1999 ACI Code6 provisions for seismic resistance, Fiber-reinforced polymer (FRP)
whereas Specimens S-3NT and S-4NT contained much less A commercially available FRP system was used for retro-
spiral reinforcement (Table 1). These four columns were fitting. The epoxy consisted of two components, A and B,
tested to failure to establish the standard behavior against which were mixed for 5 min with a mixer at a speed of 400
which columns retrofitted with FRP could be compared. The to 600 rpm. The mixing ratio was 100 parts of A to 42 parts
second group, Series ST, consisted of six columns that con- of B by volume. The carbon or glass fabric was saturated

ACI Structural Journal/January-February 2002 73

Table 1—Details of test specimens
Lateral reinforcement in test zone
fc′ ,
Specimen Size Spacing, mm ρs Treatment Axial load ratio P/Po MPa Energy damage indicator E
Group I: Series S
S-1NT US No. 3 80 1.12 Control 0.54 40.1 69
S-2NT US No. 3 80 1.12 Control 0.27 40.1 778
S-3NT US No. 3 300 0.30 Control 0.54 39.2 5
S-4NT US No. 3 300 0.30 Control 0.27 39.2 9
Group II: Series ST
ST-1NT US No. 3 300 0.30 Strengthened with 1 layer of 1.25 mm GFRP 0.54 40.4 —
ST-2NT US No. 3 300 0.30 Strengthened with 1 layer of 1.25 mm GFRP 0.54 40.4 181
ST-3NT US No. 3 300 0.30 Strengthened with 1 layer of 1.00 mm CFRP 0.54 40.4 202
ST-4NT US No. 3 300 0.30 Strengthened with 1 layer of 0.50 mm CFRP 0.27 44.8 1028
ST-5NT US No. 3 300 0.30 Strengthened with 1 layer of 1.25 mm GFRP 0.27 40.8 1040
ST-6NT US No. 3 300 0.30 Strengthened with 1.00 mm CFRP bands 0.27 41.6 78
Group III: Series R
192*
R-1NT US No. 3 160 0.56 Damaged and repaired with 2 layers of 1.25 mm GFRP 0.54 42.8
310†
31*
R-2NT US No. 3 160 0.56 Damaged and repaired with 1 layer of 1.00 mm CFRP 0.54 43.9
86†
*Based on φ1 of repaired specimen.
†Based on φ1 of original specimen.

ticular specimen was obtained from the strength-age curve for

the age of that specimen and varied between 39 and 45 MPa.

Steel
Deformed bars were used in all the specimens. Grade 400,
25M bars were used to provide longitudinal steel contents of
3.0% in all the columns. U.S. No. 3, Grade 60 steel was used
for spiral reinforcement. Reinforcement in the stub was pro-
vided by a Grade 400, 10M bar. Figure 4 shows the stress-
strain curves for the three types of steel. Each curve shown
represents an average of at least three test results.

Patching materials
Two types of patching materials were used for column re-
pair. High-early-strength mortar was prepared by mixing
Fig. 5—Location of strain gages on longitudinal and fine sand with Type I portland cement in equal amounts by
spiral reinforcement. weight. The water-cement ratio was 0.15. The compressive
strength of the mortar reached 40 MPa in 2 days. The second
with the epoxy, and a layer of epoxy was also applied to the material was a commercially available shrinkage-compen-
surface of the column. The saturated fabric was then sated mortar called EMACO S77-CR. It can be mixed with
wrapped around the column with fiber orientation in the cir- water at a ratio of 14 to 18.5% by weight and yields a com-
cumferential direction, with an overlap length of 100 mm. pressive strength of 25 to 57 MPa in 7 days.
The thickness of epoxy was not strictly controlled, and ex-
cess amounts were squeezed out along with any air bubbles.
Three types of fabrics were used in this test series. The test Instrumentation
coupons were made from the fabric impregnated with epoxy Each specimen had a total of 18 strain gages installed on
and cured to harden. Figure 3 shows details of a typical test the longitudinal reinforcement. Moreover, the spiral rein-
specimen and the tensile stress-strain curves for the three forcement within the test region was instrumented with three
types of FRP. Each curve is the average of at least three tests. strain gages on each turn. Specimens S-1NT and S-2NT had
Since the thickness of the composite depends on the amount nine strain gages each attached to the spiral reinforcement,
of epoxy used, the tensile strength is represented in force per and all other specimens had six. Figure 5 shows the locations
unit width instead of stress. of the strain gages. The concrete core deformations were
measured using 18 linear variable displacement transducers
Concrete (LVDTs) with 10 on one side and 8 on the other side. The
Ready mixed concrete with a specified compressive gage lengths varied from 75 to 120 mm and covered a length
strength of 30 MPa was used. Development of concrete of about 515 mm. Transverse displacements of each speci-
strength with age was monitored by testing two or three cylin- men were also measured at six different locations along its
ders at one time. The strength of unconfined concrete in a par- length using LVDTs.

74 ACI Structural Journal/January-February 2002

Fig. 6—Test setup. Fig. 7—Specified displacement history.
Testing
The test setup is shown in Fig. 6. A hydraulic jack with a
capacity of 4450 kN was used to apply the axial load that was
measured by a load cell. The cyclic lateral load was applied by
an actuator with a 1000 kN load capacity and a ±150 mm
stroke capacity. A displacement control mode of loading was
used in all the tests to apply a predetermined displacement
history. The testing apparatus was specially designed to al-
low in-plane rotation of test specimens. Prior to testing, each
specimen was aligned both vertically and horizontally until Fig. 8—Damaged regions of Specimen R-2NT.
the centerline of the specimen matched the line of action of
axial load. All specimens were subjected to inelastic cyclic to the stub. During testing of Specimen ST-3NT, crushing
loading while simultaneously carrying a constant axial load of concrete outside the test region was observed in the ninth
throughout the test. The lateral load sequence (Fig. 7) con- cycle (∆ = 4∆1). The test was stopped immediately by bring-
sisted of one cycle to a displacement of 0.75∆1 followed by ing the specimen to zero displacement, and the axial load
two cycles each to ∆1, 2∆1, 3∆1 ... and so on, until the specimen was reduced to half of its original level. The column outside
was unable to maintain the applied axial load. Deflection ∆1 the test region was then strengthened with two layers of
was defined as the lateral deflection corresponding to the CFRP. After that, the test was continued by increasing the
maximum lateral load along a line that represented the ini- axial load to the original. In most cases, during the last
tial stiffness of the specimen. The lateral deflection ∆1 was loading cycle, rupture of FRP at the bottom of the columns
calculated using the theoretical sectional behavior of the col- occurred along with the buckling of longitudinal reinforcing
umn and integrating curvatures along the length of the spec- bars, which was an indication that failure had begun.
imen. This loading sequence is similar to the one used Specimen ST-1NT failed in an unpredictable manner.
previously by Sheikh and Khoury.3 The GFRP composite split along the extruded reinforcing
bars used for LVDT mounts at a distance of 390 to 560 mm
TEST OBSERVATIONS from the stub face. It is believed that during wrapping of the
The first signs of distress in all test specimens were the column, the GFRP was weakened by the extruded LVDT
cracks in the cover concrete at the top and the bottom (Fig. 6). bars, which in turn caused premature rupture of the compos-
For the S series specimens (Group I), it was at the first peak ite. To avoid this type of failure, one additional FRP strip
of the fourth cycle, that is, ∆ = 2∆1, that the cover at the top with a width of 75 mm was installed along the extruded
spalled followed by spalling of the cover at the bottom at the LVDT bars on all other specimens. In the case of Specimen
second peak. In all the S series specimens (Group I), the most ST-6NT, failure was initiated by debonding of the CFRP
extensive damage was concentrated approximately 295 to bands. During the eighth cycle (∆ = 4∆1), the first CFRP
350 mm from the stub face. Spalling of the cover, however, band adjacent to the stub debonded followed by the second
extended from close to the stub for a distance of about 585 to one in the next cycle, which brought about the termination of
740 mm. During the last cycle, buckling of the longitudinal the test. The most extensive damage for all of the columns
bars was observed after yielding of the spiral reinforcement, with FRP wraps concentrated at approximately 250 to 300
which indicated the commencement of failure. In Specimens mm from the stub face, which is also the location of the first
S-3NT and S-4NT, however, the spiral reinforcement did not fiber rupture. The failure mode for all specimens was dom-
yield. Fracture of the spiral occurred in Specimens S-1NT inated by flexural effects. No cracking was seen in the stub
and S-2NT and brought about the termination of the tests. in any specimen.
For the ST series specimens, separation of the FRP fabric Specimen R-1NT was subjected to three load cycles, that
from concrete along the circumference was observed within is, maximum displacement of ∆1 when cracks formed at both
the hinging zone, as indicated by a change in FRP color, during top and bottom. The specimen was further damaged with two
the fourth or fifth cycle when the concrete crushed. As the cycles of 1.4∆1. Vertical flexural cracks were observed in the
applied displacement increased, this separation in the FRP hinging zone at a distance of approximately 100 to 400 mm
wraps extended for a distance of 200 to 400 mm from close from the stub face. Some spalling of top cover occurred at a

ACI Structural Journal/January-February 2002 75

Fig. 9—Specimens after testing.

distance of 435 to 685 mm from the stub. Yielding of longitudi- reinforcement during the last loading cycle. Figure 9 shows
nal reinforcement was also observed. Inadvertently, Speci- the specimens at the end of the tests.
men R-2NT was damaged more extensively. It was loaded
up to the fifth cycle, that is, maximum displacement of 2∆1, RESULTS AND DISCUSSION
which resulted in the yielding of both longitudinal and spiral Figure 10 shows the idealization of a test specimen. Response
reinforcement. The top cover spalled off between 150 and 550 of each specimen can be obtained in the form of applied lat-
mm from the stub, while the bottom cover was lost for a dis- eral load-displacement at column-stub connection PL-δ, shear
tance of approximately 500 mm from close to the stub (Fig. 8). force-tip deflection V-∆, and moment-curvature M-φ curves
The damaged columns were repaired while they were sub- following the procedure used for previous specimens.3 The
jected to 2/3 of the originally applied axial load. The loose curvature was computed using the deformation readings
measured by upper and lower LVDTs located at the most
concrete was first removed in both columns. A high-early-
damaged region within the hinging zone. The moment
strength mortar was used for patching Column R-1NT, while
shown was also calculated at the same location. The moment
for Column R-2NT, the structural repair mortar EMACO M consists of two parts: the primary moment caused by the
S77-CR was used. The repair mortar was cured for 2 days lateral load, and the secondary moment caused by the axial
before the FRP was wrapped around the columns as detailed load. It should be noted that although the column section
in Table 1. Observations made during the testing of the two adjacent to the stub was subjected to the maximum mo-
repaired columns were similar to those of specimens in the ST ment, failure in all the columns initiated at a location that
series (Group II), except that in the case of Specimen R-2NT, was approximately 200 to 400 mm away from the stub. The
rupture of the fibers was caused by the fracture of the spiral additional confinement provided by the stub strengthened

76 ACI Structural Journal/January-February 2002

Fig. 10—Idealization of test specimen.

the critical section such that the failure took place at a lesser
moment away from the stub. The V-∆ and M-φ responses of
Specimen S-1NT are shown in Fig. 11. For the rest of the
specimens, only the moment-curvature curves are presented
herein (Fig. 12). Important events during testing such as
spalling of the concrete cover, yielding of the spiral, buck-
ling of the longitudinal bars, fracture of the spiral, and rup-
ture of the FRP are marked on the graphs.
There were reasonable similarities in form between the V-∆
and M-φ plots for all the specimens. Of primary concern
herein is the section behavior in the plastic hinge zone, as
represented by the M-φ relationship because in the postelastic
region further lateral displacement will take place as a result
of plastic rotation at the critical section of the column. A
number of variables can be examined by comparing differ-
ent specimens. Among the steel reinforced specimens
(Group I), effects of the level of axial load and the amount
of spiral reinforcement and spiral pitch can be examined. In Fig. 11—Behavior of Specimen S-1NT.
Group II specimens, the type of fiber used in FRP, amount of
FRP reinforcement, and the level of axial load are the main Responses of Specimens S-1NT and S-2NT can be com-
variables that can be studied. Specimens in Groups II and III pared to evaluate the effects of axial load level. Specimen
can be compared with those in Group I to evaluate the ben- S-1NT was tested under an axial load of 0.54Po while in S-2NT,
eficial effects of using FRP and the effect of pre-existing the axial load was 0.27Po . Both specimens were identical in
damage before the columns are repaired. all other aspects. It is evident that an increase in axial load
resulted in reduced ductility and deformability of the col-
Ductility parameters umn. The energy dissipation capacity of the section under
Ductility in elastoplastic structures can be defined easily. lower axial load is more than 10 times that of the section
In reinforced concrete members lacking such characteristics, under high axial load. Another pair of steel reinforced spec-
however, there is no universal definition for ductility. Figure 13 imens, S-3NT (P = 0.54Po ) and S-4NT (P = 0.27Po ) can also
describes various ductility parameters that have been used be studied for the effect of axial load. The amount of spiral
for steel reinforced concrete members.3,4 These include cur- reinforcement in both of these columns is only approximately
vature ductility factor µφ, cumulative ductility ratio Nφ and 30% of that required by the ACI code.6 Column behavior
energy damage indicator E. All of the terms are defined in even under lower axial load was quite brittle, but the column
Fig. 13 except Lf and h, which represent the length of the was able to undergo five cycles of lateral load excursions and
most damaged region measured from the test and the depth failed in the sixth cycle after undergoing a displacement of
of the column section, respectively. In members where no 3∆1. The Specimen S-3NT with P = 0.54Po failed in the fifth
strength degradation takes place and the section capacity keeps cycle with a maximum displacement of 2∆1. The only vari-
increasing with increased deformation until failure, toughness able different between FRP-retrofitted Specimens ST-1NT
and energy dissipation characteristics may define the section and ST-5NT is the axial load level. Since Specimen ST-1NT
performance better than other ductility parameters. Table 1 failed prematurely, a direct comparison of the two specimens
lists the total energy damage indicator for all the columns. cannot be made. The results, however, clearly indicate the
adverse effects of high axial load on the column’s ductility.
Axial load level
Axial load level in a column is generally indicated by two Amount and spacing of spiral reinforcement
indexes, P/fc′ Ag and P/Po , where Ag = gross cross-sectional The effect of the amount and spacing of spiral reinforcement
area of the column. Sheikh, Shah, and Khoury,4 based on an can be examined by comparing the behavior of Specimen
analysis of columns with fc′ ranging from approximately 30 S-1NT with that of S-3NT and the behavior of S-2NT with that
to 60 MPa, concluded that for different fc′ values, a compar- of S-4NT. An increase in the amount of transverse reinforcement
ison of the behavior of columns using the index P/fc′ Ag does provides higher confining pressure, and the reduced spiral
not remain valid. They recommended the use of index P/Po pitch improves the stability of the longitudinal bars, thus re-
to evaluate the relative performance of columns, particularly sulting in better ductile behavior of the columns. The energy
with regard to ductility. dissipation capacities of the columns with more spiral rein-

ACI Structural Journal/January-February 2002 77

Fig. 12—Moment versus curvature responses.
forcement are orders of magnitude larger than that of speci- first set was tested under an axial load of 0.54Po , while the
mens with smaller amounts of spiral steel. axial load for the second set was 0.27Po . The first set in-
cludes Specimens S-1NT, S-3NT, ST-1NT, ST-2NT, and
Retrofitting with FRP ST-3NT (Table 1; Fig. 11 and 12). Specimen S-3NT was
The effectiveness of strengthening deficient columns with similar to Specimens ST-1NT, ST-2NT, and ST-3NT in all
FRP is evaluated by considering two sets of specimens. The respects except for the lack of FRP. Both Specimens S-3NT

78 ACI Structural Journal/January-February 2002

 Fig. 13—Definitions of ductility parameters.
and larger spiral spacing in S-3NT compared with S-1NT is
more than compensated by the additional confinement pro-
vided by the FRP wraps. It should be noted that Specimens
ST-2NT and ST-3NT had no strength degradation; the sec-
tion moment capacity kept increasing until failure. Behavior
of the two FRP-strengthened specimens was even better than
that of Specimen S-1NT in which the spiral reinforcement
satisfied the seismic code provisions of the ACI code.6 A
comparison of Specimens ST-2NT and ST-3NT shows that
two layers of GFRP results in the improvement of column
behavior similar to that obtained using one layer of CFRP.
The second set of columns that were tested under P =
0.27Po includes Specimens S-2NT, S-4NT, ST-4NT, ST-
5NT, and ST-6NT (Table 1 and Fig. 12). Specimen S-4NT
was identical to Specimens ST-4NT, ST-5NT, and ST-6NT in
all respects except for the lack of FRP. Similar to the first set,
specimens strengthened with FRP displayed higher energy
dissipation capacity and strength than Specimen S-4NT. The
seismic resistance of retrofitted columns improved signifi-
cantly as a result of the confining action of the FRP composite
wraps. The overall responses of Specimens ST-4NT and ST-
5NT, retrofitted with one layer of GFRP and one layer of 0.5
mm thick CFRP, respectively, were similar to or better than
those of Specimen S-2NT in which the spiral reinforcement
was designed according to the seismic code provisions of the
Fig. 12 (cont.)—Moment versus curvature responses. ACI code.6 The FRP-retrofitted Specimens ST-4NT and ST-
5NT did not show a significant descending part in their re-
and ST-1NT behaved in a very brittle manner, and the energy sponses until the end of the test unlike Specimen S-2NT,
dissipation capacity in each specimen was poor. As men- which showed some strength loss with increased displacement
tioned previously, failure of Specimen ST-1NT was caused excursions. Specimens ST-4NT and ST-5NT displayed very
by premature rupture of the GFRP composite along the similar responses, which indicated that the column retrofitted
extruded LVDT bars. No improvement was therefore ob- with one layer of 0.50 mm CFRP performs as well as that with
served due to the strengthening by one layer of GFRP one layer of 1.25 mm GFRP. A thinner (0.5 mm) carbon fabric
wrap in Specimen ST-1NT. Comparisons of the behavior was selected for Specimen ST-4NT taking into consideration
of Specimen S-3NT with those of ST-2NT and ST-3NT show that it will provide half as much confining pressure as expect-
the remarkable beneficial effects of FRP wrapping on strength ed from a 1.0 mm carbon fabric in Specimen ST-3NT. It was
and ductility of columns. While Specimen S-3NT failed dur- later found, however, that the strength of 0.5 mm thick CFRP
ing the fifth load cycle (maximum displacement of 3∆1), in N/mm width was similar to that of 1.0 mm thick CFRP,
Specimens ST-2NT and ST-3NT, retrofitted with two layers which indicates a better quality fiber for thinner fabric (Fig. 3).
of GFRP and one layer of CFRP, respectively, were able to Under an axial load of 0.27Po , which is approximately equal
sustain 12 load cycles with a maximum displacement of 6∆1 to a balanced load, one layer of FRP increased the energy dis-
and 11 load cycles with a maximum displacement of 5∆1, re- sipation capacity of the section by a factor of more than 100.
spectively. The energy dissipation capacity of the critical The behavior of Specimen ST-6NT retrofitted with bands
sections of the columns increased by a factor of approxi- of CFRP was more ductile and stable than Specimen S-4NT,
mately 40 due to retrofitting with glass and CFRP. The ad- but not as good as S-2NT. As mentioned previously, failure
verse effect of the reduced amount of spiral reinforcement of Specimen ST-6NT was induced by debonding of the first

ACI Structural Journal/January-February 2002 79

two CFRP bands adjacent to the column-stub interface. As 2. Unlike the internal spiral reinforcement that only con-
the first CFRP band debonded, the column started to deteriorate fines the core concrete, the FRP wraps effectively confine
due to the loss of confinement. When debonding of the sec- the entire column section. The behavior of FRP retrofitted
ond band occurred, the column was unable to maintain the columns under simulated earthquake loads matched or ex-
axial load and failed rapidly. The lap splice used for the FRP ceeded the performance of slab-reinforced columns designed
was approximately 105 mm. according to the seismic provisions of the ACI Code;6
From a comparison of Specimens ST-3NT and ST-4NT, it 3. In steel reinforced columns, section and member ductil-
can be seen that the amount of confinement required to pro- ity decreased significantly with an increased spiral pitch and
duce comparably ductile behavior depends on the level of reduced amount of spiral reinforcement. The adverse effects
axial load. Specimen ST-4NT, tested under an axial load of of a reduced amount of spiral reinforcement and larger spacing
0.27Po , displayed considerably more ductile behavior than can be compensated for by the confinement provided by FRP;
Specimen ST-3NT in which the axial load was 0.54Po . Both 4. Column ductility deteriorates as the level of axial load in-
columns were confined to a similar degree with equivalent creases. The amount of FRP reinforcement needed to improve
lateral CFRP reinforcement. A similar conclusion can also column behavior depends on the level of axial load. It was ob-
be drawn by comparing Specimens ST-2NT and ST-5NT. It served that the amount of FRP reinforcement required under an
appears that a two-fold increase in the axial load requires axial load of 0.54Po is slightly more than twice that needed for
more than twice the amount of lateral reinforcement for a an axial load of 0.27Po for similar performance enhancement;
comparable improvement in a column’s ductile performance. 5. Columns retrofitted with FRP showed little strength
The original Specimens R-1NT and R-2NT were identical degradation with increased displacement excursions until
in all respects and were tested under an axial load of 0.54Po . failure; and
They were damaged to a certain extent, repaired with FRP 6. FRP composites are very effective for the rehabilitation
while subjected to axial loads, and then tested to failure. of damaged columns. The amount of FRP needed and the
Specimen R-1NT was repaired with two layers of GFRP, performance achieved is influenced by the extent of damage.
while Specimen R-2NT was wrapped with one layer of
CFRP. The repaired specimens were tested under the same ACKNOWLEDGMENTS
high axial load level until failure. The behavior of Specimen The research reported herein was supported by a Strategic Grant from
Natural Sciences and Engineering Research Council (NSERC) of Canada
R-1NT was more ductile than that of Specimen R-2NT, and and a grant from Intelligent Sensing for Innovative Structures (ISIS), an
its sectional response was also relatively stiffer. This appears NSERC Network of Centers of Excellence. Financial and technical support
to be partly due to the fact that Specimen R-2NT was more from Fyfe and Co., R. J. Watson, Inc., Roadsavers International, and Petro-
extensively damaged than R-1NT as mentioned previously. Canada is gratefully acknowledged. The experimental work was carried out
It should be noted, however, that the lateral load and section at the Structures Laboratories in the Department of Civil Engineering at the
University of Toronto. Assistance from the technical staff is greatly appreciated.
capacity of both repaired columns kept increasing with every
load cycle until failure. The responses of Specimens R-1NT
NOTATION
and R-2NT exceeded the performance of Specimen S-1NT Ac = area of concrete in column cross section
that was designed according to the seismic provisions of the Ag = gross cross-sectional area of column
ACI code.6 Repaired Specimens R-1NT and R-2NT were As = area of longitudinal steel in column
comparable in performance, respectively, to Specimens E = energy damage indicator
fc′= compressive strength of concrete as measured from standard (150 x
ST-2NT and ST-3NT that were strengthened without damage. 300 mm) cylinder
From the test results, it can be concluded that the amount of fy = yield strength of longitudinal steel
confinement required for repair depends on the extent of h = depth of column section
damage inflicted on the member. Lf = length of most damaged region of column
M = bending moment
Nφ = commutative ductility ratio
SUMMARY AND CONCLUSIONS P = axial load on column
Results from an experimental program are presented in PL = applied lateral load
which 12 column specimens were tested under constant axial Po = axial load capacity of column = 0.85fc' (Ag – Ac) + As fy
load and cyclic lateral load excursions that simulated earth- V = shear stress
∆ = tip deflection in column
quake forces. Each specimen consisted of a 356 mm diameter δ = deflection in column
and 1.47 m long column cast integrally with a 510 x 760 x φ = curvature of section
810 mm stub that represented a beam-column joint area or a µφ = curvature ductility factor
footing. Four columns were reinforced conventionally with
longitudinal and spiral steel. Of the remaining eight columns, REFERENCES
six were strengthened with carbon or GFRP before testing, and 1. Dunker, K. F., and Rabbat, B. G., “Why America’s Bridges are Crum-
two columns were tested to a certain damage level, repaired bling,” Scientific American, Mar. 1993, pp. 66-72.
2. Yau, G., and Sheikh, S. A., “Repair and Strengthening of Columns
with FRP under axial load, and retested to failure. FRP was with Fiber Reinforced Composites,” Research Report, Department of Civil
used only in the transverse direction of the column section to Engineering, University of Toronto, Toronto, Canada, Jan. 1998, 144 pp.
confine concrete. The following conclusions can be drawn 3. Sheikh, S. A., and Khoury, S. K., “Confined Concrete Columns with
from this study. Stubs,” ACI Structural Journal, V. 90, No. 4, July-Aug. 1993, pp. 414-431.
4. Sheikh, S. A.; Shah, D. V.; and Khoury, S. S., “Confinement of High-
1. Use of carbon and GFRP resulted in remarkable im- Strength Concrete Columns,” ACI Structural Journal, V. 91, No. 1, Jan.-
provement in the performance of columns, resulting in large Feb. 1994, pp. 100-110.
increases in ductility, energy dissipation capacity, and 5. Sheikh, S. A., and Uzumeri, S. M., “Strength and Ductility of Tied Con-
strength. For a column subjected to an axial load equal to crete Columns,” Journal of the Structural Division, ASCE, V. 6, No. ST5,
May 1980, pp. 1079-1102.
0.27Po , which is approximately equal to a balanced load, one 6. ACI Committee 318, “Building Code Requirements for Structural
layer of carbon or GFRP increased the energy dissipation Concrete (ACI 318-99) and Commentary (318R-99),” American Concrete
capacity by a factor of more than 100; Institute, Farmington Hills, Mich., 1999, 391 pp.

80 ACI Structural Journal/January-February 2002