ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 98-S27

Behavior of Axially Loaded Concrete-Filled Circular Fiber-

Reinforced Polymer Tubes
by Amir Z. Fam and Sami H. Rizkalla

This paper describes the behavior of concrete-filled glass fiber- confinement effect (Shahawy and Mirmiran 1998). The con-
reinforced polymer (GFRP) tubes under axial compression loading fining pressure provided by steel tubes is limited to a con-
conditions. The study includes completely filled and partially filled stant value once the tube yields, whereas FRP tubes provide
tubes with a central hole as well as a tube-in-tube system with con-
crete filling between the two tubes. The GFRP tubes were designed a continuously increasing confining pressure, which adds to
to provide strength in both the axial and transverse directions and both the ultimate confined strength and ductility (Samaan,
were axially loaded with the concrete core. The study showed that Mirmiran, and Shahawy 1998).
the strength and ductility of concrete are improved due to confine- For practical applications, concrete-filled FRP tubes are
ment using GFRP tubes. The highest confinement level was used to carry axial compression loads and may also be de-
achieved for completely filled tubes. Using a central hole reduces
the confinement effect; however, using an inner tube can enhance signed to resist bending moments. Most of the reported ex-
the confinement for this type of member. Test results indicate that perimental and analytical research in the area of concrete
loading of the GFRP tubes reduces the confinement effectiveness. confinement using FRP (Mirmiran et al. 1998) included FRP
The effects of laminate structure, hole size, interface condition tubes completely filled with concrete, where the fibers were
between the tube and the concrete core, stiffness of the tube, and mainly oriented in the hoop direction to provide maximum
failure modes are discussed.
stiffness and strength for confinement. Other studies consid-
ered the case of applying the axial load to the concrete core
Keywords: concrete; confined concrete; polymer; strain; stress.
only for an optimum use of the FRP tube in the hoop direc-
tion for confinement (Mirmiran and Shahawy 1997). In this
INTRODUCTION
case, slip could take place between the concrete and the outer
There is a great demand for columns and piles to be con-
tube, and consequently the member would not resist bend-
structed using more durable materials than traditional con-
ing. Confinement of concrete cylinders wrapped with FRP
struction materials. Additionally, these new products have to
withstand aggressive corrosive environments, such as the sheets and subjected to axial loading conditions was also
splash zone in the case of marine piles (Stapleman 1997). studied by other researchers (Nanni and Bradford 1995;
Similarly, bridge columns have to retain their structural in- Picher, Rochette, and Labossiere 1996).
tegrity in cold regions where salt is used for deicing roads.
One promising innovative structural system is concrete- OBJECTIVES
filled fiber-reinforced polymer (FRP) tubes, which provide The main objective of this paper is to describe the structural
many unique advantages (Seible 1996). The FRP tube acts as behavior of concrete-filled glass fiber-reinforced polymer
a stay-in-place structural formwork to contain the fresh con- (GFRP) tubular short columns (length-diameter ratio of 2:1)
crete, which may save the costs of formwork and labor used subjected to axial loads based on an experimental investiga-
by the cast-in-place or precast industries. At the same time, tion recently completed by the authors. A companion paper by
the FRP tube acts as noncorrosive reinforcement for the con-
the authors (Fam and Rizkalla 2001) describes the analytical
crete for flexure and shear. More importantly, the FRP tube
model proposed to predict the measured behavior. The spe-
provides confinement to the concrete in compression, which
significantly improves the strength and ductility. The con- cific objectives of the paper are:
tained concrete is protected from severe environmental ef- 1. Compare the behavior of a concrete-filled GFRP tube to
fects and deterioration resulting from moisture intrusion that of a concrete-filled steel tube;
(Mirmiran 1995). 2. Evaluate the behavior of concrete-filled GFRP tubes using
Concrete-filled steel tubes have been used for years as a central hole with and without an inner GFRP tube;
piles and columns, and extensive research has been estab- 3. Evaluate the effect of the condition of the GFRP tube/
lished (Furlong 1967; Knowles and Park 1969; Kilpatrick concrete interface. This includes a smooth (natural) and
and Rangan 1997). It should be noted, however, that in addi- roughened interface; and
tion to the corrosion problems of steel tubes, the confinement
4. Evaluate confinement as affected by the stiffness of the
effectiveness is reduced at low levels of loading if the tube is
tube.
also loaded in the axial direction. This is attributed to the fact
that Poisson’s ratio of concrete at low levels of loading, 0.15
to 0.2, is smaller than the 0.3 value of steel (Prion and Boe- ACI Structural Journal, V. 98, No. 3, May-June 2001.
MS No. 00-074 received April 4, 2000, and reviewed under Institute publication
hme 1994; Wei, Mau, and Mantrala 1995). On the other policies. Copyright © 2001, American Concrete Institute. All rights reserved, includ-
hand, Poisson’s ratio of FRP tubes can be controlled through ing the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion will be published in the March-April 2002 ACI Structural Jour-
selected design of the laminate structure to provide more nal if received by November 1, 2001.

280 ACI Structural Journal/May-June 2001

ACI member Amir Z. Fam is a postdoctoral research associate at North Carolina State
University. He received his MSc and PhD from the University of Manitoba, Canada in
1996 and 2000, respectively. He is a member of ACI Committee 440, Fiber Reinforced
Polymer Reinforcement. His research interests include the use of fiber-reinforced poly-
mers for reinforced and prestressed concrete structures, as well as stay-in-place form-
work for concrete members.

Sami H. Rizkalla, FACI, is Distinguished Professor of Civil Engineering and Construc-

tion and Director of the Constructed Facilities Laboratory (CFL) at North Carolina State
University. He is Chairman of ACI Committee 440, Fiber Reinforced Polymer Reinforce-
ment, and is a member of Joint ACI-ASCE Committee 550, Precast Concrete Structures.

RESEARCH SIGNIFICANCE
This paper extends the knowledge in the area of concrete-
filled FRP tubes, used as structural members, by addressing (a)
new parameters intended to simulate practical applications
and loading conditions. The paper is focused on the behavior
under axial compression loading conditions and addresses
the following aspects:
1. Axial loading applied to both the tube and the concrete
simultaneously. This loading condition represents the use of
the system for pile applications, which requires also resisting
bending moments;
2. FRP tubes with a laminate structure designed to provide
strength and stiffness in both the axial and transverse direc-
tions, rather than all fibers oriented only in the hoop direc-
tion; and
3. Concrete-filled FRP tubes with a central hole. The hole
could be maintained by a nonstructural tube such as a card-
board tube or FRP tube to enhance the strength and confine-
ment. The central hole provides large reduction in self-
weight, as well as material saving for large diameter piles.

EXPERIMENTAL PROGRAM
A total of 12 hybrid stubs were fabricated and subjected to
axial compression loads. Table 1 provides details of the stubs
including the cross section configurations, diameter, height,
and the tube number used to fabricate the stubs, as given in (b)
Table 2. The figure also provides material type of the tubes,
condition of the interface surface, and the unconfined con-
crete strengths. Stubs 1 and 7 were used to examine the effect Fig. 1—Details of glass fiber-reinforced polymer (GFRP)
of material type of the tube. Stubs 11 and 12 were used to ex- tension specimens in (a) hoop; and (b) axial directions.
amine the effect of laminate structure. Stubs 1, 3, and 5 were
used to examine the effect of cross section configuration. tion, the split-disk method was used where a narrow ring was
The effect of the GFRP/concrete interface condition was ex- cut from the hollow tube and tested, as shown in Fig. 1(a), us-
amined using Stubs (1 and 2), (3 and 4), and (5 and 6). The ef- ing two semicircular rigid plates of the same curvature as the
fect of hole size was studied using Stubs 2 and 4 as well as GFRP rings to apply the tension force. Strain gages were ap-
Stubs 8, 9, and 10. The effect of the stiffness of the tube in the plied in the hoop and axial direction within the expected fail-
hoop direction was examined using Stubs 2, 8, 11, and 12. ure zone. In the longitudinal direction, tension coupons were
tested, as shown in Fig. 1(b). Because the stress concentration
Materials at the gripping location can severely influence the strength of
Five different GFRP tubes and one steel tube were used to FRP coupons, the two ends of the GFRP strip were embed-
fabricate the stubs. The GFRP tubes were standard products. ded inside hollow steel tubes 305 mm long each, and epoxy
The continuous filament winding process (Murphy 1998) was resin was used to bond the coupon to the steel tubes. Grip-
used to fabricate the cylindrical GFRP tubes using E-glass fi- ping was applied to the steel tubes, and the tested part of the
bers and epoxy resin with fiber-weight fraction of 68%. Ta- GFRP coupon between the two steel tubes was 153 x 25 mm.
ble 2 provides details of the steel and GFRP tubes, including Axial compression tests were also performed on short GFRP
the diameter, thickness, number of layers, and stacking se- rings 25 mm in height to prevent buckling of the specimen.
quence including the direction of fibers. For example, Tube 2 The load was increased until the specimens were crushed.
has multiple layers oriented towards the axial and hoop direc- The mechanical properties of the steel tube were evaluated
tions at angles of 8 and 86 degrees, respectively, while Tube 3 from a bending test performed on the hollow steel tube. Ta-
has multiple layers oriented at 4 and 88 degrees. All angles are ble 2 summarizes the mechanical properties of the tubes.
measured with respect to the longitudinal axis of the tube. Two concrete mixtures were designed to induce pressure
Tension tests were performed to evaluate the GFRP proper- fit into the tubes after hardening of the concrete. Although
ties in both the axial and hoop directions. In the hoop direc- sealed concrete is expected to have much lower shrinkage

ACI Structural Journal/May-June 2001 281

 Table 1—Details and configurations of test stubs
Concrete
Diameter, Height, Tube from Interface strength,
Stub no. Configuration mm mm Table 2 Shell condition MPa

1 168 336 2 GFRP Roughened 58

2 168 336 2 GFRP Natural 58

3 168/95 336 2 GFRP Roughened 58

4 168/95 336 2 GFRP Natural 58

5 168/89 336 2, 6 GFRP Roughened 58

6 168/89 336 2, 6 GFRP Natural 58

7 169 336 1 Steel Roughened 58

8 219 438 3 GFRP Roughened 58

9 219/95 438 3 GFRP Roughened 58

10 219/133 438 3 GFRP Roughened 58

11 100 200 5 GFRP Roughened 37

12 100 200 4 GFRP Roughened 37

compared to exposed concrete, the coefficients of thermal ture has been used in real applications involving concrete-
expansion of FRP tubes and concrete cores are different; filled FRP tubes for piles in marine environments and was
therefore, a contact pressure is established to avoid possible used for Stubs 1 to 10 of this study. Mixture 2, used for Stubs
separation under different temperatures. This was achieved by 11 and 12, contains the following materials per cubic meter:
adding expansive agents, commercially known as CONEX or
566 kg of Type 10 cement; 5.7 kg of CPD; 768 kg of sand;
CPD, to the concrete mixtures. The first mixture, Mixture 1,
contains the following materials per unit cubic meter: 360 kg 715 kg of 3/8 in. stone; and 228 kg of water. Three cylinders
of Type 10 cement; 89 kg of CONEX; 803 kg of sand; 998 kg were tested from each mixture according to ASTM C 39
of 3/8 in. stone; 154 kg of water; 4.8 kg of high-range water- standards using a rate of loading of 0.1 mm/min. Figure 2
reducing admixture; and 5% air content. The designed slump shows the stress-strain curves in the axial direction of the
was 8 in. to provide good workability conditions. This mix- two types of concrete used in this study.

282 ACI Structural Journal/May-June 2001

Table 2—Details and mechanical properties of tubes used to fabricate stubs
Axial direction Hoop direction
Tube Diameter, Thickness, Material No. of Stacking sequence, Thickness of layers, fu(ten.), fu(comp.), fu(ten.),
no. mm mm type layers degrees* mm MPa MPa E, GPa υ MPa E, GPa
305 305 305
1 169 4.09 Steel — — [4.09] 203 0.3 203
(yield) (yield) (yield)
3.73 [0.08/0.36/0.28/0.25/
[+8/-86/-86/+8/-86/
2 168 (including GFRP 9
+8/-86/+8/-86]
0.38/0.25/0.36/0.25/ 282.9 224.1† 19.8 0.066 548 33.4
1.17 liner) 0.35]
3.70 [0.23/0.23/0.23/0.25/
[-88/-88/+4/-88/-88/ 0.25/0.28/0.25/0.24/
3 219 (including GFRP 9
+4/-88/+4/-88]
201.3 182.6† 19.8 0.055 548 33.4
1.49 liner) 0.25]
4 100 3.09 GFRP 1 [0] pultruded [3.09] 629.5 161.3 37.7‡ 0.242‡ 39‡ 8.7‡
[0.32/0.35/0.27/0.48/
[-87/+3/-87/+3/-87/
5 100 3.08 GFRP 9
+3/-87/+3/-87]
0.26/0.51/0.19/0.51/ 444 115 29 0.101‡ 398‡ 23.0‡
0.19]
2.28 [+15/-82/-82/+15/- [0.18/0.33/0.25/0.18/
6 89 (including GFRP 9
82/+15/-82/+15/-82]
0.25/0.18/0.25/0.18/ 209‡ 241 17.6‡ 0.155‡ 261‡ 27.6‡
0.22 liner) 0.25]
*Anglesmeasured with respect to longitudinal axis of tube.
†Manufacturer data.
‡
Classical lamination theory.

Fabrication of stubs
To provide a rough interface surface between the concrete
and the tubes, as in the case of Stubs 1, 3, and 5, the inner sur-
face of the GFRP Tubes 1 and 2 (Table 2) and the outer surface
of GFRP Tube 6 were roughened. This was achieved by ap-
plying a thin layer of epoxy to the surface, and later spraying
silica sand on top of the tacky epoxy. The tubes were inclined
at an angle of 16 degrees, as shown in Fig. 3(a), to facilitate
casting concrete. The central holes of Stubs 3, 4, 9, and 10
were achieved using cardboard tubes, while central holes for
Stubs 5 and 6 were achieved using GFRP Tube 6, which were
secured in position through wooden end plugs. In practice,
central holes may be constructed using mandrels that are
pulled out after concrete hardening, or by spinning the FRP
tube while it is partially filled with fresh concrete, similar to
the technique currently used in fabricating spun-cast concrete
Fig. 2—Typical stress-strain curves for two concrete mixtures.
poles. Strain gages were installed at midheight in the hoop di-
rection of the tubes to monitor the expansion of concrete dur-
ing curing in room temperature. Immediately after casting, porting steel plates to ensure uniform distribution of the
hoop strains were monitored for 7 weeks. Later, the stubs were applied pressure.
cut to lengths twice the diameter using a diamond saw to
maintain a smooth and square surface, as shown in Fig. 3(b). TEST RESULTS AND DISCUSSION
The following sections provide the test results and a dis-
cussion of the different aspects of the study, including the
Instrumentation and testing concrete expansion behavior during curing, behavior of con-
At four locations around the perimeter of each stub, at a fined concrete under axial loads, the effect of various param-
section located at midheight, strains were measured using eters on the behavior, and failure modes.
both strain gages and 100 mm displacement transducer PI-
gages. The PI-gages were attached in the axial direction on Concrete expansion during curing
two opposite sides. Strain gages were installed on the other Figure 5(a) shows the history of hoop strains developed after
two opposite sides in both the axial and lateral directions, as casting and during curing of the concrete for Stubs 1, 3, 5, and
shown in Fig. 4. This pattern of gages allows for axial strain 7. The behavior reflects the resultant of two mechanisms tak-
measurements at four sides around the perimeter for more ing place over time, including the internal pressure developed
accuracy, and also checks for any eccentricity, which is not due to expansion of concrete, as well as the drying shrinkage.
intended during loading. Stubs 5 and 6 included additional The behavior shows that the hoop strains reached the maxi-
strain gages installed in the lateral direction on the inner and mum values at approximately 20 days regardless of the cross
outer GFRP tubes 40 mm below the upper surface. section configuration or the material type of the tube (GFRP or
The axial load was applied to the entire cross section sur- steel). The strains were almost stabilized with little reduction
face, including the concrete core and the tube. Stubs were test- over time, which indicates that the shrinkage process could
ed, as shown in Fig. 4, using a closed-loop MTS 5000 kN have lasted longer than the expansion process. Knowing the
testing machine under stroke control with a 0.1 mm/min rate stiffness of the tubes in the hoop direction, (Et/R), one can es-
of loading. A thin layer of quickset plaster was placed be- timate the sustained radial pressure at the interface between
tween the end faces of the stub and both the loading and sup- concrete and the tube σ using the following equation

ACI Structural Journal/May-June 2001 283

 Fig. 3—Fabrication of test stubs.

tially filled tubes with small and large holes, Stubs 9 and 10,
respectively. This is attributed to the fact that the central
space provides some relief of the internal pressure, as well as
the fact that in partially filled tubes, there is less volume of
concrete to produce expansion as high as that in completely
filled tubes. The negative hoop strain developed during cur-
ing of Stub 10, as shown in Fig. 5(b), is very small (0.0001).
In this stub, the effect of concrete shrinkage, though small in
sealed concrete, has overcome the expansion effect of con-
crete after the first 10 days due to the large hole size. The
bond between concrete and tube was capable of transferring
the very small hoop compressive strain to the tube in this
case. The effect of hole size on the sustained pressure ratio
(σ/σo) is also shown in Fig. 6, and is based on Stubs 1 and 3
as well as Stubs 8, 9, and 10. σ and σo are the pressures built
in the partially and completely filled tubes, respectively, for
the same GFRP tube. Di and Do are the inner hole diameter
and the outer diameter, respectively. Curve fitting was im-
posed on the limited measured values, mainly to provide a
Fig. 4—Instrumentation and test setup. trend rather than specific equations. The figure clearly indi-
cates significant reduction of the pressure with increasing the
size of the inner hole. The trend is consistent for the two values
Et
σ = ----- ε (1) of the stiffness (Et/R) used in the investigation. It should be
R noted that the measured prestrain in the FRP tubes was only in
the range of 4% of the ultimate strain of the material, there-
where E is the elastic modulus of the tube in the hoop direc- fore, its effect was ignored in this study.
tion; t is the thickness of the tube; R is the radius of the outer
tube; and ε is the stabilized hoop strain measured on the tube, Confinement effect on stress-strain behavior of
as given in Fig. 5(a). The estimated pressures were 0.63, 0.49, concrete
0.42, and 5.13 MPa for Stubs 1, 3, 5, and 7, respectively. The To evaluate the beneficial effect of confinement on the to-
higher pressure in Stub 7 is attributed to the higher stiffness of tal load-carrying capacity of the stubs, the measured axial
the tube, mainly due to the higher elastic modulus of steel load-strain behavior of Stub 11 is compared with that ob-
(203 GPa compared with 33.4 GPa for GFRP), as well as the tained by superposition of both the measured axial load-
difference in wall thickness. The behavior also shows that strain behavior of the hollow GFRP tube, and of the plain
less pressure is built up in the case of central holes, which is concrete core individually, as shown in Fig. 7. At a strain of
also illustrated in Fig. 5(b), by comparing the history of con- approximately 0.004, the hollow GFRP tube suffered local
crete expansion of a completely filled tube, Stub 8, with par- damage at the loading end when tested in compression with-

284 ACI Structural Journal/May-June 2001

 (a) Fig. 6—Effect of hole size on pressure due to expansion of
concrete.

(b)

Fig. 5—Development of hoop strains over time due to

expansion of concrete: (a) Stubs 1, 3, 5, and 7; and (b) Stubs
8, 9, and 10. Fig. 7—Confinement effect on load-strain behavior of concrete.

out concrete filling. The figure clearly indicates that the ca-
pacity of the composite stub significantly exceeds the load
sharing capacity of the two individual materials. In fact, the
predicted behavior using superposition matches the mea-
sured response of the composite stub very well up to the vi-
cinity of the unconfined plain concrete strength, which
strongly indicates the beneficial effect of confinement be-
yond this stage, when concrete starts to expand excessively
at a stress level of approximately 87% of the unconfined
strength (Avram et al. 1981) and the confining mechanism is
activated through the GFRP tube.

Effect of material type of tube

The behavior of GFRP confined concrete, Stub 1, is com-
pared to that of steel confined concrete, Stub 7, in Fig. 8. The
plain concrete behavior is also shown for comparison. Al-
Fig. 8—Comparison between GFRP-confined concrete and
though the structural wall thickness of the GFRP tube is 37%
steel-confined concrete.
less than that of the steel tube, they both achieved the same
axial strength. The behavior of GFRP confined concrete is
characterized by a bilinear response with a transition zone yielded, followed by a plastic plateau with large deforma-
near the vicinity of unconfined concrete strength. The stiff- tions. The continuous increase of the load resistance of FRP
ness beyond this point (second slope) is governed by the confined concrete is attributed to the continuous increasing
stiffness of the GFRP tube. Once the tube reached its tensile of the confining pressure due to the linear nature of FRP. On
strength in the presence of the axial compressive stresses, it the other hand, once the steel tube yields, the confining pres-
fractured and the stub failed. On the other hand, steel con- sure is almost constant regardless of the degree of concrete
fined concrete behaved almost linearly until the steel tube expansion.

ACI Structural Journal/May-June 2001 285

 (a)
Fig. 9—Comparison between concrete confined using both
filament-wound and pultruded GFRP tubes.

It is also shown that the measured strength of the GFRP/

concrete hybrid system, Stub 1, is 41% higher than the com-
bined load capacities of the unconfined concrete and the
GFRP tube added together, whereas the increase in the ca-
pacity for the steel/concrete system, Stub 7, is only 18%. This
behavior is attributed to the fact that a steel tube attracts more
axial load in comparison to a GFRP tube under the same axial
strain level, mainly due to the higher elastic modulus of steel
as compared to GFRP in axial direction. It should also be
noted that the wall thickness of the steel tube is 1.59 times
that of the GFRP tube.

Effect of laminate structure

Stubs 11 and 12 were identical except that the GFRP tube
in Stub 11 is filament wound and the fibers are oriented in (b)
both longitudinal and hoop direction with a ratio of approxi-
Fig. 10—Effect of cross section configuration and interface
mately 2:1, whereas the GFRP tube of Stub 12 is pultruded
condition on load-strain behavior: (a) load-strain behavior
with all fibers oriented in the axial direction. Figure 9 com-
of Stubs 1 to 6; and (b) load-lateral strain behavior of Stub 5.
pares the load-axial and lateral strain behavior for both stubs.
The figure clearly illustrates the advantage of fibers in the
hoop direction for confinement, resulting in much higher activation of the inner tube in the confinement mechanism.
strength. The pultruded tube fails to confine concrete at the This is also evident by the lateral strain response on both the
vicinity of peak unconfined strength, when concrete starts to inner and outer GFRP tubes, as shown in Fig. 10(b).
expand excessively and the tube splits immediately due to the At early stages of loading, before concrete expands, both
lack of stiffness and strength in the hoop direction. This ob- tubes were under lateral tensile strains due to Poisson’s ra-
servation confirms similar findings reported by Kanathrana tio effect for tubes under axial compression. Once the con-
and Lu (1998). In some cases, however, pultruded tubes are crete starts to expand excessively, it presses against both
used as permanent formwork and also to provide additional tubes. At this point the outer tube is subjected to higher hoop
axial load resistance. tensile strains, whereas the hoop strain of the inner tube
starts to reverse toward the compression side. At this point,
Effect of cross section configurations and the stiffness is comparable to a completely filled tube. In
interface condition general, from Fig. 10(a), the interface condition had an insig-
Figure 10(a) shows the load-strain behavior of Stubs 1 to nificant effect on the behavior of Stubs 1 to 4. A more pro-
6, including completely and partially filled GFRP tubes, as nounced effect is recognized for Stubs 5 and 6 due to the
well as stubs with tube-in-tube configuration. The confine- larger surface area of GFRP in contact with concrete, yet the
ment effectiveness can be detected by the slope of the second maximum strength was not affected.
part of the bilinear response. It is evident that the completely
filled tube is more effective in confinement compared to the Effect of hole size
tubes with an inner hole. It is also evident, however, that pro- The stress-strain response of confined concrete is com-
viding an additional inner GFRP tube, as in Stubs 5 and 6, pared in Fig. 11(a) for stubs with various hole sizes. Stubs 8,
could enhance the confinement. The behavior of Stubs 5 and 9, and 10 are confined using the same GFRP tube with stiff-
6 shows a similar flat response to that of Stubs 3 and 4 at the ness (Et/R) of 680 MPa. Stubs 2 and 4 are confined using a
unconfined strength level. Soon, however, the stiffness in- similar GFRP tube with stiffness of 1040 MPa. The only
creases and the slope of the second part of the response ap- variable among each group is the hole size, which clearly af-
proaches a similar value to that of Stubs 1 and 2 due to fects the behavior and the confined strength. The confined

286 ACI Structural Journal/May-June 2001

 (a) Fig. 12—Effect of stiffness of tube on confinement effectiveness.

(b)
Fig. 13—Variation of lateral strain at different axial strains.
Fig. 11—Effect of void size on confinement level: (a) stress-
strain curves of Stubs 2, 4, 8, 9, 10, and plain concrete; and
(b) reduction in confined strength due to increase in central ratio or using GFRP tubes with higher elastic moduli in the
hole size. lateral direction.
In 1997, Mirmiran and Shahawy tested concrete-filled
strengths of the partially filled tubes are normalized with re- GFRP tubes with a range of stiffnesses (Et/R) very similar to
spect to the strength of completely filled tubes and were re- that used in this study for Stubs 1, 2, 8, 11, and 12. The
lated to the inner-outer diameter ratio (Di/Do), as shown in strength of the GFRP tubes in the hoop direction was also very
Fig. 11(b). The reduction in strength with increasing hole similar to that of Stubs 1, 2, and 8. The 152.5 x 305 mm stubs
included GFRP tube with 1.3, 2.1, and 3.0 mm wall thickness-
size is attributed to the fact that voided concrete cores result
es, filled with 32 MPa concrete, and tested in compression by
in reduction of the radial stresses under axial compression,
loading the concrete core only. The tubes consisted of a fila-
consequently reducing the contact pressure with the tube, ment-wound angle ply of polyester resin with E-glass fibers
and therefore reducing the confinement effect. at ±15 degrees with the hoop direction, resulting in elastic
moduli of 37.2 to 40.7 GPa and strengths of 524 to 641 MPa
Effect of tube stiffness on confinement in the hoop direction for the 1.3 to 3.0 mm thick tubes. The
The confinement effectiveness (fcc′ /fc′ ) is defined herein as results of their study are also plotted in Fig. 12 for compari-
the ratio between the maximum strength of the hybrid stub to son. It is clearly shown that the present study showed less
the strength of the unconfined concrete. This ratio is affected confinement effectiveness under the same stiffness level.
by both the strength and the stiffness (Et/R) of the tube in the This is mainly attributed to the fact that the GFRP tubes in
hoop direction, which control the level of confinement. the present study were axially loaded, and therefore, they ex-
panded outward due to Poisson’s ratio effect under their own
Stubs 1, 2, 8, 11, and 12 were selected to examine the effect
share of axial load, which results in less contact pressure
of stiffness of the tube on the confinement effectiveness, as
with concrete. Another factor is that the tubes were also bi-
shown in Fig. 12. The stubs consist of completely filled axially loaded under axial compression and lateral tension in
GFRP tubes with different stiffness values including 1040, the present study, which reduces their tensile strength in the
680, 1462, and 550 MPa for Stubs 1 and 2, 8, 11, and 12, re- hoop direction, whereas the GFRP tubes in the other study
spectively. The figure shows the increase in the strength of were fully utilized in the hoop direction under uniaxial ten-
confined concrete by increasing the stiffness of the tube. sile stresses, which allows the development of the full tensile
This could be achieved by using a higher thickness-diameter strength.

ACI Structural Journal/May-June 2001 287

 showed additional signs of crushing in the axial direction.
Once the tube fails, the concrete core immediately fails due to
the loss of containment, as was also reported by other re-
searchers (Fardis and Khallili 1981; Mirmiran et al. 1998).
Steel confined concrete, on the other hand, did not fail once
the steel jacket yielded but showed a ductile response where
the concrete core kept expanding, causing the stub to bulge
and sustain large axial strains.

SUMMARY AND CONCLUSIONS

This study described the behavior of axially loaded con-
crete-filled GFRP tubes with an emphasis on the beneficial
effects of confinement in practical applications and realistic
loading conditions. The experimental study investigated in
particular three aspects, including: 1) partially filled GFRP
tubes with inner holes; 2) tube-in-tube system; and 3) fully
composite sections, where the tube and the core are bonded
to resist the total applied load. Laminate structure of the
GFRP tubes was designed to provide resistance in the axial
and hoop directions.
The study showed that the confinement effect is highly de-
pendent on the boundary conditions, laminate structure of the
tube, and the size of the central hole. Experimental modeling
in previous studies could overestimate the confinement bene-
fits by assuming ideal cases such as tubes with maximum
Fig. 14—Failure mode of different stubs. strength and stiffness in the hoop direction or tubes that are
not axially loaded to be fully utilized in the hoop direction.
This comparison shows that loading the concrete core only The following conclusions were drawn from this study:
to preserve the strength and stiffness of the tube to be fully 1. Ignoring the effect of axial loading of the tube in a con-
utilized in the hoop direction could overestimate the benefi- crete-filled GFRP tube under compression and assuming the
cial effect of confinement. Most practical applications re- development of the full hoop strength, overestimate the con-
quire a full composite action as well as longitudinal strength finement effectiveness. The tube is biaxially loaded under
and stiffness for combined loading cases. the effect of axial compressive and hoop tensile stresses;
therefore, the material strength is governed by the biaxial
Axial-lateral strain behavior strength envelope. In addition, loading the tube causes lateral
The axial-lateral strain behavior of the tube is a good indi- expansion and results in less contact (confining) pressure
cator of the initiation of the confinement mechanism. Figure with concrete;
13 shows the axial-lateral strain behavior of completely 2. Completely filled GFRP tubes provide the most effec-
filled GFRP and steel tubes, Stubs 2 and 7, as well as stubs tive confinement. Although an inner hole offers material
with inner hole and tube-in-tube, Stubs 4 and 6. The behavior saving and reduced self-weight, it reduces the confinement
of completely filled tubes is generally characterized by a bilin- effect even though a high level of ductility is maintained;
ear response, with the first slope slightly higher than Poisson’s 3. If the hole is maintained by an inner GFRP tube, the
ratio of the tube, because at this stage the concrete undergoes confinement effectiveness is improved and could approach
small elastic lateral expansion corresponding to stable micro- that of a completely filled tube, depending on the stiffness of
cracking, which induces little tensile strains in the tube. Once the inner tube. The outer and inner tubes are subjected to
the concrete starts to expand excessively near the vicinity of hoop tensile and compressive stresses, respectively;
the unconfined strength, the tube is fully activated in confine-
ment and the rate of lateral straining increases, as shown by 4. GFRP tubes provide similar confined concrete strength
the second slope. The figure also shows that under the same to that provided by steel tubes while using a significantly
axial strain level, the lateral strain in Stub 4 is less than that smaller wall thickness of the tube; however, a lower level of
of Stub 2, which indicates less confinement due to the pres- ductility is achieved;
ence of an inner hole, as previously explained. The figure 5. Filament-wound GFRP tubes are superior to pultruded
also confirms that using an additional GFRP tube to maintain tubes in confinement. The pultruded tubes split immediately
the inner hole improves the confinement, as indicated by the once the concrete starts to expand excessively, and the stub
slope of the second branch for Stub 6. Steel tube shows a fails;
similar behavior with a higher lateral strain level due to the 6. Stress-strain response of GFRP confined concrete is bi-
higher Poisson’s ratio of steel. linear with the transition zone near the peak strength of the un-
confined concrete. The slope of the second branch is governed
Failure modes by the stiffness of the tube as well as the inner hole size;
Figure 14 shows the stubs after failure. The GFRP confined 7. Initiation of the confinement mechanism can be detected
concrete failed by fracture of the tube under a biaxial state of from the bilinear axial-lateral strain behavior of the tube. The
stress. The tubes were subjected to axial compressive stresses slope of the first part almost represents Poisson’s ratio of the
and lateral tensile stresses. In most cases, the apparent failure tube. A change of slope occurs when concrete expands, pro-
mode is rupture of the tube in the hoop direction, and few stubs ducing more lateral strains in the tube;

288 ACI Structural Journal/May-June 2001

 8. The interface condition between the tube and concrete Furlong, R. W., 1967, “Strength of Steel-Encased Concrete Beam Col-
has an insignificant effect on the behavior when both the umns,” Proceedings of the ASCE, V. 93, No. ST5, Oct., pp. 113-124.
Kanatharana, J., and Lu, L., 1998, “Strength and Ductility of Concrete
concrete core and the tube are axially loaded; and Columns Reinforced by FRP Tubes,” Proceedings of the First Interna-
9. The ultimate strength of the stub is governed by failure tional Conference on Composites in Infrastructure (ICCI’96), Tucson,
of the GFRP jacket. Unlike a steel tube, the GFRP tube frac- Ariz., Jan., pp. 370-384.
tures in a brittle manner. Kilpatrick, A. E., and Rangan, B. V., 1997, “Deformation-Control Analy-
sis of Composite Concrete Columns,” Research Report No. 3/97, School of
Civil Engineering, Curtin University of Technology, Perth, Australia, July.
ACKNOWLEDGMENTS Knowles, R. B., and Park, R., 1969, “Strength of Concrete Filled Steel
The authors wish to acknowledge financial support provided by the Net- Tubular Columns,” ASCE Proceedings, V. 95, No. ST12, Dec., pp. 2565-2587.
work of Centres of Excellence on Intelligent Sensing for Innovative Struc- Mirmiran, A., 1995, “Concrete Composite Construction for Durability
tures (ISIS Canada), the Natural Science and Engineering Research Council and Strength,” Symposium on Extending Life Span of Structures, Interna-
of Canada (NSERC), Lancaster Composite, and Ameron International for tional Association for Bridge and Structural Engineering (IABSE), San
fabricating and providing the FRP tubes. The authors are also grateful to Francisco, Calif., Aug., pp. 1155-1160.
Moray McVey and Grant Horeczy of the University of Manitoba for their as- Mirmiran, A., and Shahawy, M., 1997, “Behavior of Concrete Columns
sistance during the experimental program. Confined by Fiber Composites,” Journal of Structural Engineering, May,
pp. 583-590.
NOTATION Mirmiran, A.; Shahawy, M.; Michel, S.; El Echary, H.; Mastrapa, J. C.;
Di = diameter of inner hole of partially filled tube and Pico, O., 1998, “Effect of Column Parameters on FRP-Confined Con-
Do = diameter of outer tube crete,” Journal of Composites for Construction, ASCE, V. 2, No. 4, Nov.,
E = elastic modulus of tube in hoop direction pp. 175-185.
fc′ = compressive strength of unconfined concrete Murphy, J., 1998, Reinforced Plastic Handbook, Second Edition,
fcc′ = compressive strength of confined concrete Elsevier Science Ltd., Great Britain.
fu(comp.) = compressive strength of material of tube Nanni, A., and Bradford, N. M., 1995, “FRP Jacketed Concrete under
fu(ten.) = tensile strength of material of tube Uniaxial Compression,” Construction and Building Materials, V. 9, No. 2,
R = radius of outer tube pp. 115-124.
t = structural wall thickness of outer tube Picher, F.; Rochette, P.; and Labossiere, P., 1996, “Confinement of Con-
crete Cylinders with CFRP,” Proceedings of the 1st International Confer-
ε = strain in hoop direction in confining tube
ence on Composites in Infrastructure, H. Saadatmanesh and M. R. Ehsani,
σ = sustained radial pressure at interface between concrete core
eds., University of Arizona, Tucson, Ariz., pp. 829-841.
and tube due to expansion of concrete during curing
Prion, H. G. L., and Boehme, J., 1994, “Beam-Column Behavior of Steel
σo = sustained radial pressure due to expansion of concrete dur-
Tubes Filled with High-Strength Concrete,” Canadian Journal of Civil
ing curing in completely filled tubes
Engineering, V. 21, pp. 207-218.
υ = major Poisson’s ratio of tube, causing strain in hoop direc-
Samaan, M.; Mirmiran, A.; and Shahawy, M., 1998, “Model of Concrete
tion under axial loading
Confined by Fiber Composites,” Journal of Structural Engineering, Sept.,
pp. 1025-1032.
REFERENCES Seible, F., 1996, “Advanced Composite Materials for Bridges in the 21st
Avram, C.; Facaoaru, I.; Mirsu, O.; Filimon, I.; and Tertea, I., 1981, Century,” Proceedings of the Advanced Composite Materials in Bridges
Concrete Strength and Strains, Elsevier Scientific Publishing Co. and Structures, M. El-Badry, ed., CSCE, Montreal, Canada, pp. 17-30.
Daniel, I. M., and Ishai, O., 1994, Engineering Mechanics of Composite Shahawy, M., and Mirmiran, A., 1998, “Hybrid FRP-Concrete Beam-
Materials, Oxford University Press, New York. Columns,” Proceedings of ICCE/5 Fifth International Conference on Com-
Fam, A. Z., and Rizkalla, S. H., 2000, “Confinement Model for Axially posites Engineering, Las Vegas, Nev., July 5-11, pp. 619-620.
Loaded Concrete Confined by Circular FRP Tubes.” (submitted to ACI Stapleman, J., 1997, “Pile on the Abuse,” Composite Technology, Sept.-
Structural Journal for publication) Oct., pp. 56-58.
Fardis, M. N., and Khalili, H., 1981, “Concrete Encased in Fiberglass- Wei, S.; Mau, S. T.; and Mantrala, S. K., 1995, “Performance of New
Reinforced Plastic,” ACI JOURNAL, Proceedings V. 78, No. 6, Nov.-Dec., Sandwich Tube under Axial Loading: Experiment,” Journal of Structural
pp. 440-446. Engineering, V. 121, No. 12, Dec., pp. 1806-1821.

ACI Structural Journal/May-June 2001 289