Engineering Structures 126 (2016) 365–378

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Axial and lateral stress-strain model for concrete-filled steel tubes

with FRP jackets
C.X. Dong a, A.K.H. Kwan a, J.C.M. Ho b,⇑
a
Department of Civil Engineering, The University of Hong Kong, Hong Kong, China
b
School of Civil Engineering, The University of Queensland, Brisbane, Australia

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

Article history: In a concrete-filled steel tube under axial compression, since the steel tube has a larger Poisson’s ratio than
Received 21 May 2015 the concrete, delamination at the steel-concrete interface occurs and the steel tube is ineffective in provid-
Revised 27 July 2016 ing confinement at the elastic stage. For resolving this problem and enhancing the concrete confinement,
Accepted 29 July 2016
the provision of an external FRP (fiber reinforced polymer) jacket to restrict the lateral expansion of the
Available online 9 August 2016
steel tube is an effective means. Herein, to investigate the effectiveness of FRP-jacketed steel tube confine-
ment, a theoretical model for evaluating the lateral strain, confining stress and axial stress in a concrete-
Keywords:
filled steel tube with FRP jacket at various stages of loading is developed. The theoretical model is first
Concrete confinement
Concrete-filled steel tube
applied to analyze specimens tested by other researches to verify its accuracy and then used to work
Confining stress out the FRP confining stiffness required to eliminate steel-concrete delamination and the FRP confining
FRP jacket stiffness required to achieve Level I ductility (no strain softening at the inelastic stage until failure).
Lateral strain of concrete Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
To improve the ductility of concrete members under axial com-
pression, especially those cast of high-strength concrete which
tends to be more brittle than normal-strength concrete, the tradi-
tional method is to install closely spaced transverse reinforcement
so as to provide better lateral confinement to the concrete. How-
ever, the effectiveness of the transverse reinforcement gradually
decreases as the concrete strength becomes higher and higher [1]
and thus for high-strength concrete members, the amount of trans-
verse reinforcement needed could be very large, thereby posing
difficulties to concrete placing due to steel rebar congestion.
In view of the above problems, the concrete-filled steel tube
(CFST) system has been advocated [2,3] to avoid the concreting dif-
ficulties due to steel rebar congestion and provide a more uniform
and continuous confinement to the concrete core. However,
despite these merits, delamination at the interface between the
steel tube and the concrete core could occur at the elastic stage
because the Poisson’s ratio of steel is larger than that of concrete
(the Poisson’s ratios of steel and concrete are around 0.3 and 0.2,
respectively). Such delamination would delay the development of
confining stress in the concrete core, decrease the effectiveness
of the steel tube confinement and even cause premature buckling
⇑ Corresponding author.
E-mail address: johnny.ho@uq.edu.au (J.C.M. Ho).
of the steel tube [4]. For this reason, FRP (fiber reinforced polymer)
jacketing of CFST has been proposed to restrict the lateral expan-
sion and restrain the outward buckling of the steel tube.
A lot of effort has been spent on testing FRP jacketed CFST spec-
imens to study the influence of shape of steel tube [5–8], thick-
ness/diameter ratio of steel tube [9–11], concrete strength [12],
FRP type [13,14] and FRP confining stiffness [15–17]. From the test
results, several general observations can be made: (1) the confining
stress developed in a CFST with FRP jacket is often negative or zero
(negative means tension) at the elastic stage, implying that delam-
ination could still occur even with FRP jacket provided; (2) the
additional confining stress provided by the FRP jacket could sub-
stantially enhance the axial load capacity; (3) the FRP jacket could
delay or even suppress the outward buckling of the steel tube; and
(4) FRP jacketing is an effective means of retrofitting the struc-
turally deficient members or upgrading the key elements in CFST
structures to modern seismic design standards. However, these
observations direct from the test results are mostly qualitative,
lacking quantitative analysis of the actual confining mechanism
in the structural system.
On the other hand, design formulas for evaluating the axial load
capacity of CFST with FRP jacket have been derived [11,13,14] by
modifying an existing design formula originally developed for CFST
without FRP jacket [18]. However, these formulas have certain lim-
itations: (1) There is still no general consensus on the definition of
axial load capacity, especially when there is no obvious yield point

http://dx.doi.org/10.1016/j.engstruct.2016.07.059
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
366 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378
Nomenclature
Ec Young’s modulus of concrete
Es Young’s modulus of steel tube
Efrp Young’s modulus of FRP
0
fc unconfined concrete strength (concrete cylinder
strength)
f cc peak axial stress on stress-strain curve of confined con-
crete
f sy yield strength of steel tube
ts thickness of steel tube
t frp thickness of FRP jacket
D outer diameter of steel tube
eco axial strain at peak axial stress of unconfined concrete
ecc axial strain at peak axial stress of confined concrete
eex elastic strain in x-direction
epx inelastic strain in x-direction
eTx total strain in x-direction (lateral strain in x-direction)
eey elastic strain in y-direction
ez axial strain in z-direction
in the axial load-strain curve. In practice, the use of different
definitions would lead to slightly different values of axial load
capacity. (2) The maximum confining stress (same as the confining
stress at rupture) offered by the FRP jacket is often used to evaluate
the confinement index for determining the axial load capacity. In
theory, the confining stiffness and rupture strain of the FRP should
be separately considered because they have different effects [19].
(3) The various design formulas do not agree with each other as
they were established from different sets of test results.
Regarding theoretical modelling, Choi and Xiao [20] developed
in 2010 an analytical model for predicting the full range axial
stress-strain behaviour of CFST with FRP jacket by employing
plasticity-based constitutive models for both the concrete and
the steel tube. In this model, a rather complex numerical procedure
is required to solve the total of twelve compatibility and equilib-
rium equations involved. In 2011, Hu et al. [10] developed a sim-
pler axial stress-strain model for FRP jacketed CFST by employing
a lateral strain model for the concrete and a plastic model for the
steel tube. Perfect bond between the concrete and steel tube is
assumed and the lateral strain of the confined concrete is taken
as an independent variable (input as prescribed values) in the anal-
ysis. In 2012, Che et al. [11] incorporated the confining effects of
the steel tube and FRP jacket by modifying the axial stress-strain
curve of the confined concrete. Since the properties of the steel
tube and FRP jacket are embedded in the constitutive model of
the confined concrete, the lateral strain of the confined concrete
and the confining stresses provided by the steel tube and FRP
jacket at different stages of loading cannot be evaluated directly.
In 2013, Teng et al. [21] also developed an axial stress-strain model
for FRP jacketed CFST, in which the lateral strain of concrete is
taken as an independent variable to be input as a prescribe value
rather than a dependent variable to be solved from the other stress
and strain components.
Herein, a theoretical axial and lateral stress-strain model for
FRP jacketed CFST, which incorporates a lateral-to-axial strain
model of confined concrete recently developed by the authors
[22], an axial stress-strain model of confined concrete developed
by Attard and Setunge [23], a plastic model of steel based on the
associated flow rule and von Mises yield criterion, and an elastic
stress-strain model of FRP, is proposed. The lateral-to-axial strain
model of confined concrete which is more generally applicable
has been applied to finite element analysis of cylindrical FRP con-
fined concrete under concentric loading [24] and eccentric loading
eez elastic strain in z-direction
ez0 axial strain at formation of splitting cracks
ez;s axial strain of steel tube
eh;s circumferential strain of steel tube
efrp longitudinal strain of FRP jacket
mc Poisson’s ratio of concrete
ms Poisson’s ratio of steel tube
rx normal stress (confining stress) in x-direction
ry normal stress (confining stress) in y-direction
rz normal stress (axial stress) in z-direction
rr confining stress in radial direction
rz;s axial stress of steel tube
rh;s circumferential stress of steel tube
rfrp longitudinal stress of FRP jacket
driz;s incremental axial stress of steel tube at step i
drih;s incremental circumferential stress of steel tube at step i
deiz;s incremental axial strain of steel tube at step i
deih;s incremental circumferential strain of steel tube at step i
[25], non-cylindrical FRP confined concrete [26], where the lateral
strain and confining stress could vary across the section. In this
model, the axial strain is applied incrementally, and then the lat-
eral strain and confining stress are evaluated by solving the equa-
tions governing the compatibility and equilibrium between the
confined concrete, the steel tube and the FRP jacket. The confining
stress so evaluated is then substituted into the axial stress-strain
model of confined concrete to determine the axial stress in the
concrete. The validity and accuracy of the theoretical model are
verified by comparing with published test results. Moreover, a
parametric study on the effects of the concrete, steel tube and
FRP jacket properties is carried out to evaluate the FRP confining
stiffness required to eliminate steel-concrete delamination and
the FRP confining stiffness required to eliminate strain softening
at the inelastic stage. Lastly, an appraisal of the effectiveness of
FRP jacketing is presented.
2. Proposed model for CFST with FRP jacket
In general, to analyze the axial and lateral stress-strain beha-
viour of confined concrete, a total of three constitutive models
are needed: (1) a lateral-to-axial strain model of concrete with var-
ious concrete strengths and under different confining stresses; (2)
an axial stress-strain model of concrete with various concrete
strengths and under different confining stresses; and (3) a confin-
ing stress-lateral strain model of the confinement taking into
account the stress-strain behaviour of the confining materials (in
this case, the steel tube and FRP jacket).
2.1. Lateral-to-axial strain model of confined concrete
In 2013, Ozbakkaloglu et al. [27] and Tao et al. [28] suggested
that further research on the lateral strain of confined concrete is
needed to enable more rigorous and generally applicable analysis
of concrete provided with FRP and/or steel tube confinement. In
a recent study [22], the authors separated the lateral strain of con-
fined concrete into an elastic component (lateral strain due to elas-
tic deformation) and an inelastic component (lateral strain due to
formation of splitting cracks), and by analyzing published test
results covering a wide range of concrete strength, have developed
a new lateral-to-axial strain model, as depicted below. As the
details have been published before, only the key features are pre-
sented herein.
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 367

Before formation of splitting cracks, the concrete remains elas- For the peak axial stress f cc and the corresponding axial strain
tic and isotropic, though it may exhibit slight nonlinearity. The ecc of the confined concrete, the authors found that the following
elastic lateral strains are given by the following equations (note formulas derived by Xiao et al. [29], which have been verified by
that the z-direction is the longitudinal direction of the concrete comparing with extensive test results, would give more accurate
column and the x- and y-directions are the lateral directions): results than the original formulas given by Attard and Setunge:

 rx
 ry
0 0:80
f cc =f c ¼ 1 þ 3:24 rr =f c
0
eex ¼ mc eez þ 1  m2c  mc þ m2c ð1aÞ ð7aÞ
Ec Ec

1:06
rx ry ecc =eco ¼ 1 þ 17:4 rr =f 0c ð7bÞ
e ¼ m e  ðmc þ m
e
y
e
c z cÞ
2
þ ð1  m cÞ
2
ð1bÞ
Ec Ec
where eex , eey and eez are the elastic strains in x-, y- and z-directions, rx 2.3. Stress-strain model of steel tube
and ry are the confining stresses in x- and y-directions, Ec is the
Young’s modulus, and mc is the Poisson’s ratio. For circular columns In theory, the steel tube is under axial, radial and circumferen-
under concentric load, the lateral strain and stress are the same in tial stresses, or in other words, under a triaxial stress state. How-
all radial directions. Taking eex ¼ eey and rx ¼ ry ¼ rr , Eqs. (1a) and ever, since the thickness of the steel tube is very small compared
to the diameter, the magnitude of the radial stress is negligible
(1b) can be simplified as:
compared to that of the circumferential stress. Hence, the steel

 rr
eex ¼ eey ¼ mc eez þ 1  mc  2m2c ð2Þ tube may be assumed to be under a biaxial stress state. Assuming
Ec that the steel is a linearly elastic-perfectly plastic material [30], the
As the axial strain increases, splitting cracks would eventually incremental stress-strain relations of the steel tube in the elastic
be formed. A formula for evaluating the axial strain at formation and plastic stages are given by Eqs. (8) and (9), respectively:
" #  " #
of splitting cracks has been derived by analyzing the published test driz;s Es 1 ms deiz;s
results, as given below: ¼ ð8Þ
drih;s 1  ms ms 1 deih;s
!
ez0  0 02

0  rr
2 3"
¼ 0:44 þ 0:0021f c  0:00001f c 1 þ 30 exp 0:013f c 0 " # #
eco fc driz;s
s2a
E s 4 1  sc ms  sascsb deiz;s
¼ 5 ð9Þ
ð3Þ drih;s 1  m s m  sa sb 1  sbc
s2 deih;s
s sc
in which ez0 is the axial strain at formation of splitting cracks, eco is
0
where
the axial strain at peak axial stress of the unconfined concrete, f c is
the cylinder strength of the unconfined concrete and rr is the con- sa ¼ sx þ ms sh ð10aÞ
fining stress. After formation of splitting cracks, the concrete
becomes inelastic and anisotropic. At this stage, the lateral strain sb ¼ ms sx þ sh ð10bÞ
eTx is taken as the summation of the elastic strain eex induced by Pois-
son’s ratio effect and the inelastic strain epx induced by splitting sc ¼ s2x þ s2h þ ms sx sh ð10cÞ
cracks:
1
eTx ¼ eex þ epx ð4Þ sx ¼ ð2ri1
z;s  rh;s Þ
i1
ð10dÞ
3
Having determined the value of ez0 , the inelastic strain epx may 1
be evaluated as: sh ¼ ðrz;s
i1
þ 2rh;s
i1
Þ ð10eÞ
3
n h

io
0 1:1
epx ¼ 19:1ðez  ez0 Þ1:5 0:1 þ 0:9 exp 5:3 rr =f c ð5Þ In the above equations, Es is the Young’s modulus, ms is the Pois-
son’s ratio, riz;s and rih;s are the stresses in the axial and circumfer-
The above formula was derived by correlating the inelastic ential directions, eiz;s and eih;s are the strains in the axial and
strain epx to the post-crack axial strain ðez  ez0 Þ, concrete cylinder
circumferential directions, driz;s and drih;s are the incremental stres-
strength f c and confining stress rr , and conducting regression anal-
0

ses in the axial and circumferential directions, deiz;s and deih;s are the
ysis of the test results of both passively and actively confined con-
crete specimens published in the literature. incremental strains in the axial and circumferential directions, and
i is the incremental step number. To check whether the steel tube
has yielded, the von Mises yield criterion is employed, as given in
2.2. Axial stress-strain model of confined concrete
the following equation:
For the axial stress-strain model, the one developed by Attard
 2
ri1 þ ðri1
h;s Þ  rz;s rh;s ¼ ðf sy Þ
2 2
z;s
i1 i1
ð11Þ
and Setunge [23], which has been shown to be applicable to a
broad range of concrete strength from 20 to 130 MPa, is adopted.
where f sy is the yield strength of the steel tube.
Its stress-strain curve is of the following form:

rz Aðez =ecc Þ þ Bðez =ecc Þ2 2.4. Stress-strain model of FRP jacket

¼ ð6Þ
f cc 1 þ ðA  2Þðez =ecc Þ þ ðB þ 1Þðez =ecc Þ 2
where rz and ez are the axial stress and axial strain, f cc and ecc are
the peak axial stress and the corresponding axial strain of the con-
fined concrete, and A and B are constants defining the shape of the
stress-strain curve. The constants A and B for the ascending and
descending portions have been given by Attard and Setunge [23]
and are therefore not repeated here for brevity.
Regarding the FRP jacket, the FRP is assumed to be linearly elas-
tic until rupture. Its axial stress-strain relation along the fiber
direction may be taken as:
rfrp ¼ Efrp efrp ð12Þ
where rfrp and efrp are the axial stress and strain along the fiber
direction, and Efrp is the Young’s modulus. Herein, only the case in
368 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378
which the fiber direction follows the circumference of the steel tube
with no inclination angle is considered.
2.5. Compatibility and equilibrium conditions
Previous research [31,32] has revealed that the bond strength at
steel-concrete interface ranges from 0.42 to 2.04 MPa, depending
on the concrete strength and steel surface roughness. As it is rela-
tively small and there is no consensus on its exact value, the bond
strength is neglected and simply taken as zero in this study.
Since the Poisson’s ratio of steel is generally larger than that of
concrete, at the initial stage of loading, the lateral expansion of the
steel tube could be larger than that of the concrete core even with
FRP jacket provided to restrict the lateral expansion of the steel
tube. As a result, delamination between the steel tube and the con-
crete core could occur. If there is delamination, the lateral strains of
the steel tube and concrete core would be independent and the
confining stress would be equal to zero. Under such circumstance,
the compatibility and equilibrium conditions between the steel
tube and the FRP jacket are given by:
eh;s ¼ efrp ð13Þ
2rh;s t s þ 2rfrp t frp ¼ 0 ð14Þ
in which ts is the thickness of steel tube and tfrp is the thickness of
FRP jacket.
If there is no delamination, the lateral strains of the steel tube
and concrete core would be forced to be equal and the confining
stress would become non-zero. Under such circumstance, the com-
patibility and equilibrium conditions between the concrete core,
the steel tube and the FRP jacket are given by:
eTx ¼ eh;s ¼ efrp ð15Þ
rr ðD  2ts Þ þ 2rh;s ts þ 2rfrp tfrp ¼ 0 ð16Þ
in which D is the outer diameter of steel tube.
3. Evaluation of lateral strain, confining stress and axial stress
The above lateral-to-axial strain model of confined concrete,
axial stress-strain model of confined concrete, biaxial stress-
strain model of steel tube and uniaxial stress-strain model of FRP
jacket are integrated together for the evaluation of the lateral
strain, confining stress and axial stress of the CFST with FRP jacket
based on the compatibility and equilibrium conditions. To obtain
the complete lateral-to-axial strain and axial stress-strain curves,
the axial strain is applied as prescribed axial strain to the concrete
and steel tube simultaneously and the lateral strain, confining
stress and axial stress are evaluated by the following procedures:
(1) Apply a prescribed axial strain ez to the concrete and steel
tube incrementally in small axial strain increments starting
from zero. From numerical experience, each small axial
strain increment may be taken as 0.0001.
(2) Evaluate the lateral strain of unconfined concrete and the
lateral strain of steel tube with FRP jacket and compare the
respective lateral strains of unconfined concrete and steel
tube to check the bond condition. Delamination occurs when
the magnitude of lateral strain of unconfined concrete is
smaller than that of steel tube. Delamination disappears
when the magnitude of lateral strain of unconfined concrete
is larger than that of steel tube.
(3) Evaluate the lateral strain eTx and confining stress rr using
the lateral-to-axial strain model of confined concrete and
stress-strain model of steel tube by trial and error. This
can be done either manually using a spreadsheet program
or automatically using a computerized iterative procedure
of alternately estimating the lateral strain from the esti-
mated confining stress and estimating the confining stress
from the estimated lateral strain until convergent results
are obtained. The confining stress is taken as zero when
delamination occurs.
(4) Substitute the confining stress rr so obtained by step (3) into
the axial stress-strain model of confined concrete to evaluate
the corresponding axial stress rz . Evaluate the axial and cir-
cumferential stresses of steel tube using the incremental
axial and circumferential strains so obtained by step (3).
(5) Multiply the axial stresses in confined concrete (or uncon-
fined concrete when delamination occurs) and steel tube
so obtained by step (4) by their respective cross-section
areas to obtain the axial load of CFST with FRP jacket.
(6) Increase the axial strain ez by a small increment and repeat
steps (1–5) until the complete lateral-to-axial strain and
axial load-strain curves are obtained.
4. Verification of proposed model
A total of 89 FRP jacketed CFST specimens tested by other
researchers [6,9–14,16,17] have been analyzed to verify the valid-
ity and accuracy of the new axial and lateral stress-strain model.
Their concrete properties, steel tube properties and FRP jacket
properties are listed in Tables 1–3, respectively. As can be seen
from the tables, the concrete strength ranges from 31.2 to
51.9 MPa, the steel tube thickness/diameter ratio ranges from
0.0050 to 0.0564, the steel yield strength ranges from 226 to
386 MPa, the FRP confining stiffness ranges from 0 to 2391 MPa,
and the FRP type covers carbon, glass and basalt fibers.
For verification, the predicted lateral-to-axial strain curves and
axial load-strain curves by the proposed model are compared to
the published test results in the following sections.
4.1. Lateral-to-axial strain curves
The lateral strain of the structural system varies during loading
and is the key factor governing the confining stresses provided by
the steel tube and FRP jacket. Hence, accurate prediction of the lat-
eral strain is crucial to the structural analysis. To assess the perfor-
mance of the proposed model, the lateral strains of 25 specimens
have been analyzed and the results are compared with the
measured lateral strains. Due to the large number of specimens
analyzed, it is not possible to present all the results. For brevity,
only the measured and predicted lateral-to-axial strain curves of
four representative specimens are presented and compared in
Figs. 1 and 2.
In Fig. 1(a), the specimen analyzed is ‘C1-0’ tested by Tao et al.
[6]. This specimen has a concrete cylinder strength of 46.0 MPa,
and is confined by a steel tube with thickness/diameter
ratio = 0.0192 and yield strength = 230 MPa, but is not provided
with any FRP jacket (FRP confining stiffness = 0 MPa). For this par-
ticular specimen, the predicted lateral strains by the proposed
model agree very well with the measured results. In Fig. 1(b), the
specimen analyzed is ‘F3-135’ tested by Hu et al. [10]. This speci-
men has a concrete cylinder strength of 42.1 MPa, and is confined
by a steel tube with thickness/diameter ratio = 0.0074 and yield
strength = 242 MPa, and a glass FRP jacket with confining stiff-
ness = 402 MPa. The comparison reveals that the predicted lateral
strains are slightly smaller than the measured results in magnitude.
In Fig. 2(a), the specimen analyzed is ‘F4-202’ tested by Hu et al.
[10]. This specimen has a concrete cylinder strength of 35.9 MPa,
and is confined by a steel tube with thickness/diameter
ratio = 0.0050 and yield strength = 231 MPa, and a glass FRP jacket
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 369

Table 1
Concrete properties of specimens analyzed.

Specimen label Reference from which obtained Concrete cylinder strength (MPa) Poisson’s ratio Young’s modulus (MPa) Axial strain at peak stress
of unconfined concrete
C1-0 [6] 46.0 0.18 35,800 0.0024
C1-1 [6] 46.0 0.18 35,800 0.0024
C1-2 [6] 46.0 0.18 35,800 0.0024
C2-0 [6] 46.0 0.18 35,800 0.0024
C2-1 [6] 46.0 0.18 35,800 0.0024
C2-2 [6] 46.0 0.18 35,800 0.0024
1–1.5 [9] 47.4 0.18 30,336 0.0025
1–2.5 [9] 47.4 0.18 30,336 0.0025
1–3.5 [9] 47.4 0.18 30,336 0.0025
1–4.5 [9] 47.4 0.18 30,336 0.0025
2–1.5 [9] 47.4 0.18 30,336 0.0025
2–2.5 [9] 47.4 0.18 30,336 0.0025
2–3.5 [9] 47.4 0.18 30,336 0.0025
2–4.5 [9] 47.4 0.18 30,336 0.0025
F0-102 [10] 42.2 0.18 27,900 0.0026
F1-102 [10] 42.2 0.18 27,900 0.0026
F2-102 [10] 42.2 0.18 27,900 0.0026
F3-102 [10] 42.2 0.18 27,900 0.0026
F0-135 [10] 42.1 0.18 28,800 0.0026
F2-135 [10] 42.1 0.18 28,800 0.0026
F3-135 [10] 42.1 0.18 28,800 0.0026
F4-135 [10] 42.1 0.18 28,800 0.0026
F0-202 [10] 35.9 0.18 26,700 0.0025
F2-202 [10] 35.9 0.18 26,700 0.0025
F3-202 [10] 35.9 0.18 26,700 0.0025
F4-202 [10] 35.9 0.18 26,700 0.0025
CSC A-0 [11] 46.5 0.22 34,500 0.0026
CSC B-0 [11] 46.5 0.22 34,500 0.0026
CSC C-0 [11] 46.5 0.22 34,500 0.0026
CSC D-0 [11] 46.5 0.22 34,500 0.0026
CSC E-0 [11] 46.5 0.22 34,500 0.0026
CSC A-1 [11] 46.5 0.22 34,500 0.0026
CSC B-1 [11] 46.5 0.22 34,500 0.0026
CSC C-1 [11] 46.5 0.22 34,500 0.0026
CSC D-1 [11] 46.5 0.22 34,500 0.0026
CSC E-1 [11] 46.5 0.22 34,500 0.0026
CSC A-2 [11] 46.5 0.22 34,500 0.0026
CSC B-2 [11] 46.5 0.22 34,500 0.0026
CSC C-2 [11] 46.5 0.22 34,500 0.0026
CSC D-2 [11] 46.5 0.22 34,500 0.0026
CSC E-2 [11] 46.5 0.22 34,500 0.0026
CFT [12] 46.6 0.18 30,139 0.0026
CCFT-2L-1 [12] 46.6 0.18 30,139 0.0026
CCFT-2L-2 [12] 46.6 0.18 30,139 0.0026
t4C40 [13] 38.4 0.18 28,008 0.0023
CF1t4C40 [13] 38.4 0.18 28,008 0.0023
CF2t4C40 [13] 38.4 0.18 28,008 0.0023
CF3t4C40 [13] 38.4 0.18 28,008 0.0023
GF1t4C40 [13] 38.4 0.18 28,008 0.0023
GF2t4C40 [13] 38.4 0.18 28,008 0.0023
GF3t4C40 [13] 38.4 0.18 28,008 0.0023
CF2t3C40 [13] 38.4 0.18 28,008 0.0023
CF2t5C40 [13] 38.4 0.18 28,008 0.0023
CF2t4C50 [13] 46.7 0.18 30,164 0.0025
CF2t4C60 [13] 51.9 0.18 31,418 0.0026
C-0-3.0 [14] 31.2 0.18 26,400 0.0020
C-B1-3.0 [14] 31.2 0.18 26,400 0.0020
C-B2-3.0 [14] 31.2 0.18 26,400 0.0020
C-C1-3.0 [14] 31.2 0.18 26,400 0.0020
C-C2-3.0 [14] 31.2 0.18 26,400 0.0020
C-0-4.5 [14] 34.7 0.18 27,900 0.0020
C-B1-4.5 [14] 34.7 0.18 27,900 0.0020
C-B2-4.5 [14] 34.7 0.18 27,900 0.0020
C-C1-4.5 [14] 34.7 0.18 27,900 0.0020
C-C2-4.5 [14] 34.7 0.18 27,900 0.0020
C-0-6.0 [14] 34.7 0.18 27,900 0.0020
C-B1-6.0 [14] 34.7 0.18 27,900 0.0020
C-B2-6.0 [14] 34.7 0.18 27,900 0.0020
C-C1-6.0 [14] 34.7 0.18 27,900 0.0020
C-C2-6.0 [14] 34.7 0.18 27,900 0.0020
C-0-7.5 [14] 34.7 0.18 27,900 0.0020
C-B1-7.5 [14] 34.7 0.18 27,900 0.0020
C-B2-7.5 [14] 34.7 0.18 27,900 0.0020

(continued on next page)

370 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378

Table 1 (continued)

Specimen label Reference from which obtained Concrete cylinder strength (MPa) Poisson’s ratio Young’s modulus (MPa) Axial strain at peak stress
of unconfined concrete
C-C1-7.5 [14] 34.7 0.18 27,900 0.0020
C-C2-7.5 [14] 34.7 0.18 27,900 0.0020
F0-60 [16] 48.3 0.18 30,557 0.0026
F1-60 [16] 48.3 0.18 30,557 0.0026
F2-60 [16] 48.3 0.18 30,557 0.0026
F3-60 [16] 48.3 0.18 30,557 0.0026
C3N [17] 37.5 0.18 27,761 0.0023
C3F-1 [17] 37.5 0.18 27,761 0.0023
C3F-3 [17] 37.5 0.18 27,761 0.0023
C4N [17] 37.5 0.18 27,761 0.0023
C4F-1 [17] 37.5 0.18 27,761 0.0023
C4F-2 [17] 37.5 0.18 27,761 0.0023
C4F-3 [17] 37.5 0.18 27,761 0.0023
C6N [17] 37.5 0.18 27,761 0.0023
C6F-1 [17] 37.5 0.18 27,761 0.0023
C6F-3 [17] 37.5 0.18 27,761 0.0023

with confining stiffness = 537 MPa. For this particular specimen,
the predicted lateral strains are slightly smaller than the measured
results in magnitude. In Fig. 2(b), the specimen analyzed is
‘GF2tC40’ tested by Lu et al. [13]. This specimen has a concrete
cylinder strength of 38.4 MPa, and is confined by a steel tube with
thickness/diameter ratio = 0.0313 and yield strength = 248 MPa,
and a glass FRP jacket with confining stiffness = 576 MPa. From
the figure, it is seen that at small axial strain, the predicted lateral
strains are larger than the measured results whereas at relatively
large axial strain, the predicted lateral strains are smaller than
the measured results.
Overall, both the measured and predicted lateral strain results
reveal that at the initial elastic stage, the magnitude of lateral
strain increases linearly with the axial strain, depending on the
thickness/diameter ratio of the steel tube and the confining stiff-
ness of the FRP jacket. If the lateral to axial strain ratio is larger
than the Poisson’s ratio of the concrete, steel-concrete delamina-
tion occurs and the confining stress is equal to zero, otherwise,
there is no steel-concrete delamination and the steel tube and
FRP jacket are providing confinement to the concrete. As the axial
strain increases further, sooner or later splitting cracks are formed
in the concrete and thereafter the magnitude of lateral strain
increases at a much faster rate, resulting in rapid increases in the
circumferential stresses in the steel tube and FRP jacket and thus
also rapid increase in the confining stress acting on the concrete
core.
To further assess the accuracy of the proposed model in the pre-
diction of the lateral strain, the measured lateral strains of the 25
specimens analyzed at every 0.005 axial strains are plotted against
the predicted lateral strains by the proposed model in Fig. 3. The
data points plotted reveal that on the whole, the measured and
predicted lateral strains agree quite well with each other, as indi-
cated by the closeness of most of the data points to the line of
equality and the relatively high R2 value of 0.835. There are a
few outliers but even with the outliers included, the average error
and average absolute error have been calculated as 0.3% and
20.3%, respectively.
4.2. Axial load-strain curves
At each prescribed axial strain, the axial stresses in the concrete
core and steel tube are evaluated and then the axial load is calcu-
lated by multiplying the axial stresses with the respective cross-
section areas. By increasing the axial strain in small increments
and evaluating the axial load at each increment step, the complete
axial load-strain curve is derived. To assess the performance of the
proposed model, the axial loads of 89 specimens have been ana-
lyzed and the results are compared with the measured axial loads.
As it is not possible to present all the results, only the measured
and predicted axial load-strain curves of four representative spec-
imens are presented and compared in Figs. 4 and 5. The specimens
presented in these four figures are the same as those presented in
Figs. 1 and 2.
In Fig. 4(a), the measured and predicted axial load-strain curves
agree quite well with each other and both exhibit obvious strain
softening in the post-peak stage. The predicted peak load is slightly
lower than the measured peak load but at larger axial strain, the
predicted axial load is slightly higher than the measured axial load.
In Fig. 4(b), the measured and predicted axial load-strain curves
are very close to each other at both the pre-peak and post-peak
stages. Particularly, the measured results indicate that FRP rupture
occurred at an axial strain of 0.018 whereas the proposed model
predicts that FRP rupture would occur at almost the same axial
strain. Hence, the proposed model is very accurate in predicting
FRP rupture.
In Fig. 5(a), the measured and predicted axial load-strain curves
are quite close to each other. The predicted axial load is slightly
lower and slightly higher than the measured result at the pre-
peak and post-peak stages, respectively. Moreover, the predicted
axial strain at which FRP rupture would occur agrees very closely
with the measured axial strain at FRP rupture. From Fig. 5(b), it
is evident that after yielding, the predicted axial load-strain curve
is significantly lower than the measured axial load-strain curve.
This specimen has a steel tube with a rather large thickness/diam-
eter ratio of 0.0313. The underestimation of axial load by the pro-
posed model is probably caused by the negligence of the strain
hardening effect of the steel tube, which is more pronounced when
the thickness of the steel tube is relatively large.
Overall, it is seen that at the initial elastic stage, the axial load
increases linearly with the axial strain. Moreover, the axial stiffness
(gradient of the axial load-strain curve) changes little with the
thickness/diameter ratio of the steel tube and the confining stiff-
ness of the FRP jacket because at this stage, the lateral strain and
confining stress are both relatively small and the concrete core
behaves just like unconfined concrete. As the axial strain increases
further, splitting cracks are formed, and the lateral strain and con-
fining stress both increase at faster rates, but because of gradual
damage of the concrete core, the axial load only increases at a
decreasing rate until the FRP ruptures. Once the FRP ruptures,
the axial load drops abruptly as the confining stress provided by
the FRP jacket suddenly vanishes. Nevertheless, the steel tube con-
tinues to provide confinement and the CFST is still able to sustain a
certain amount of axial load, depending on the thickness/diameter
ratio and yield strength of the steel tube.
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 371

Table 2
Steel tube properties of specimens analyzed.

Specimen label Outer diameter (mm) Thickness (mm) Poisson’s ratio Young’s modulus (MPa) Yield strength (MPa)
C1-0 156.0 3.00 0.30 206,000 230
C1-1 156.0 3.00 0.30 206,000 230
C1-2 156.0 3.00 0.30 206,000 230
C2-0 250.0 3.00 0.30 206,000 230
C2-1 250.0 3.00 0.30 206,000 230
C2-2 250.0 3.00 0.30 206,000 230
1–1.5 127.0 1.50 0.30 200,000 350
1–2.5 129.0 2.50 0.30 200,000 350
1–3.5 131.0 3.50 0.30 200,000 310
1–4.5 133.0 4.50 0.30 200,000 310
2–1.5 127.0 1.50 0.30 200,000 350
2–2.5 129.0 2.50 0.30 200,000 350
2–3.5 131.0 3.50 0.30 200,000 310
2–4.5 133.0 4.50 0.30 200,000 310
F0-102 204.0 2.00 0.30 203,000 226
F1-102 204.0 2.00 0.30 203,000 226
F2-102 204.0 2.00 0.30 203,000 226
F3-102 204.0 2.00 0.30 203,000 226
F0-135 203.0 1.50 0.30 204,000 242
F2-135 203.0 1.50 0.30 204,000 242
F3-135 203.0 1.50 0.30 204,000 242
F4-135 203.0 1.50 0.30 204,000 242
F0-202 202.0 1.00 0.30 203,000 231
F2-202 202.0 1.00 0.30 203,000 231
F3-202 202.0 1.00 0.30 203,000 231
F4-202 202.0 1.00 0.30 203,000 231
CSC A-0 127.0 1.50 0.28 203,000 330
CSC B-0 129.0 2.50 0.28 203,000 330
CSC C-0 131.0 3.50 0.28 203,000 330
CSC D-0 133.0 4.50 0.28 203,000 330
CSC E-0 136.0 6.00 0.28 203,000 330
CSC A-1 127.0 1.50 0.28 203,000 330
CSC B-1 129.0 2.50 0.28 203,000 330
CSC C-1 131.0 3.50 0.28 203,000 330
CSC D-1 133.0 4.50 0.28 203,000 330
CSC E-1 136.0 6.00 0.28 203,000 330
CSC A-2 127.0 1.50 0.28 203,000 330
CSC B-2 129.0 2.50 0.28 203,000 330
CSC C-2 131.0 3.50 0.28 203,000 330
CSC D-2 133.0 4.50 0.28 203,000 330
CSC E-2 136.0 6.00 0.28 203,000 330
CFT 152.0 2.95 0.30 200,000 356
CCFT-2L-1 152.0 2.95 0.30 200,000 356
CCFT-2L-2 152.0 2.95 0.30 200,000 356
t4C40 128.0 4.00 0.30 191,000 248
CF1t4C40 128.0 4.00 0.30 191,000 248
CF2t4C40 128.0 4.00 0.30 191,000 248
CF3t4C40 128.0 4.00 0.30 191,000 248
GF1t4C40 128.0 4.00 0.30 191,000 248
GF2t4C40 128.0 4.00 0.30 191,000 248
GF3t4C40 128.0 4.00 0.30 191,000 248
CF2t3C40 126.0 3.00 0.30 191,000 243
CF2t5C40 130.0 5.00 0.30 191,000 242
CF2t4C50 128.0 4.00 0.30 191,000 248
CF2t4C60 128.0 4.00 0.30 191,000 248
C-0-3.0 133.0 3.00 0.30 193,400 365
C-B1-3.0 133.0 3.00 0.30 193,400 365
C-B2-3.0 133.0 3.00 0.30 193,400 365
C-C1-3.0 133.0 3.00 0.30 193,400 365
C-C2-3.0 133.0 3.00 0.30 193,400 365
C-0-4.5 133.0 4.50 0.30 193,400 365
C-B1-4.5 133.0 4.50 0.30 193,400 365
C-B2-4.5 133.0 4.50 0.30 193,400 365
C-C1-4.5 133.0 4.50 0.30 193,400 365
C-C2-4.5 133.0 4.50 0.30 193,400 365
C-0-6.0 133.0 6.00 0.30 193,400 365
C-B1-6.0 133.0 6.00 0.30 193,400 365
C-B2-6.0 133.0 6.00 0.30 193,400 365
C-C1-6.0 133.0 6.00 0.30 193,400 365
C-C2-6.0 133.0 6.00 0.30 193,400 365
C-0-7.5 133.0 7.50 0.30 193,400 365
C-B1-7.5 133.0 7.50 0.30 193,400 365
C-B2-7.5 133.0 7.50 0.30 193,400 365
C-C1-7.5 133.0 7.50 0.30 193,400 365

(continued on next page)

372 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378

Table 2 (continued)

Specimen label Outer diameter (mm) Thickness (mm) Poisson’s ratio Young’s modulus (MPa) Yield strength (MPa)
C-C2-7.5 133.0 7.50 0.30 193,400 365
F0-60 165.0 2.75 0.30 200,000 386
F1-60 165.0 2.75 0.30 200,000 386
F2-60 165.0 2.75 0.30 200,000 386
F3-60 165.0 2.75 0.30 200,000 386
C3N 139.8 3.20 0.30 200,000 365
C3F-1 139.8 3.20 0.30 200,000 365
C3F-3 139.8 3.20 0.30 200,000 365
C4N 139.8 4.50 0.30 200,000 334
C4F-1 139.8 4.50 0.30 200,000 334
C4F-2 139.8 4.50 0.30 200,000 334
C4F-3 139.8 4.50 0.30 200,000 334
C6N 139.8 6.60 0.30 200,000 301
C6F-1 139.8 6.60 0.30 200,000 301
C6F-3 139.8 6.60 0.30 200,000 301

To further assess the accuracy of the proposed model in the pre-
diction of the axial load, the measured axial loads of the 89 speci-
mens analyzed at every 0.005 axial strains are plotted against the
predicted axial loads by the proposed model in Fig. 6. The data
points plotted reveal that on the whole, the measured and predicted
axial loads agree very well with each other, as indicated by the
closeness of most of the data points to the line of equality and the
relatively high R2 value of 0.877. There are a few outliers but even
with the outliers included, the average error and average absolute
error have been calculated as +2.4% and 7.7%, respectively. Hence,
although the error in prediction of lateral strain is relatively large,
the error in prediction of axial load is quite small. In conclusion,
the proposed model is fairly accurate in the generation of axial
load-strain curve for analyzing the full range axial behaviour and
evaluating the axial strength and ductility of CFST with FRP jacket.
5. Parametric study
5.1. Required FRP confining stiffness to eliminate delamination
Steel-concrete delamination due to larger Poisson’s ratio of steel
than concrete is one major problem associated with CFST column.
The delamination would destroy the bond at the steel tube-
concrete interface and delay the development of confining stress,
thus adversely affecting the effectiveness of the steel tube confine-
ment [4]. To resolve this problem, the provision of an external FRP
jacket to restrict the lateral expansion of the steel tube could be an
effective means. Moreover, the FRP jacket could provide additional
confinement to the concrete core.
A parametric study on the effect of FRP confining stiffness on
the onset of delamination has been carried out. It is found that a
certain minimum FRP confining stiffness is required to eliminate
the delamination and that the required FRP confining stiffness is
dependent solely on the Poisson’s ratios of the steel tube and con-
crete, and the thickness/diameter ratio of the steel tube. From the
parametric study, it has been worked out that at given Poisson’s
ratios (in this study, the Poisson’s ratios of the steel tube and con-
crete are taken as 0.30 and 0.18, respectively), the required FRP
confining stiffness increases with the steel tube thickness/diameter
ratio as a linear function, as shown in Fig. 7. This means that the
required FRP confining stiffness is generally higher for a steel tube
with a larger thickness/diameter ratio.
5.2. Required FRP confining stiffness to achieve Level I ductility
At potential plastic hinge zones (for example, near beam-
column joints or column-foundation connections), the ductility
demand is generally higher. Hence, the potential plastic hinge
zones should be designed to a higher level of ductility whereas
the other parts of the members could be designed to a lower level
of ductility. For a CFST column, the ductility at the potential plastic
hinges can be improved by increasing the thickness of the steel
tube. However, it is costly to increase the thickness of the entire
length of steel tube and impractical to locally increase the thick-
ness of the steel tube at the potential plastic hinge zones. Never-
theless, a good solution is to locally provide a FRP jacket to the
CFST, or if there is already a FRP jacket, locally increase the thick-
ness of the FRP jacket.
In previous studies [19,24], the authors have proposed to define
two different levels of ductility, namely: Level I ductility and Level
II ductility. At Level I ductility, the axial stress would continue to
increase with the axial strain even after reaching the yield stress
until failure (there is no strain softening). At Level II ductility, the
axial stress would start to decrease with the axial strain after
reaching the peak axial stress (there is strain softening) but would
increase again after reaching a certain minimum value not lower
than 85% of the peak axial stress. It is the authors’ view that col-
umns in seismic or impact resistant structures should be designed
to have Level I ductility at the potential plastic hinge zones.
A parametric study on the effects of various structural parame-
ters on ductility has been carried out to determine the required FRP
confining stiffness for achieving Level I ductility. Such required FRP
confining stiffness is plotted against the concrete cylinder strength
for different steel tube yield strengths (300 and 600 MPa) and dif-
ferent steel tube thickness/diameter ratios (from 0 to 0.06) in
Fig. 8. Overall, it is seen that the required FRP confining stiffness
increases as the concrete strength increases, decreases as the steel
tube yield strength increases, and decreases as the steel tube thick-
ness/diameter ratio increases. In general, the required ductility can
be met with by increasing the steel tube thickness and/or the FRP
confining stiffness.
5.3. Effects of FRP jacketing on confining stress and axial load
To study the effects of FRP jacketing, two representative FRP
jacketed CFST columns with concrete cylinder strength = 50 MPa,
steel tube yield strength = 600 MPa, steel tube thickness/diameter
ratio = 0.010, FRP confining stiffness = 1000 or 4000 MPa, and FRP
rupture strain = 0.010 have been analyzed. The variations of the
FRP provided confining stress (confining stress provided by the
FRP jacket) and the steel tube provided confining stress (confining
stress provided by the steel tube) are presented in Fig. 9, whereas
the corresponding variations of the concrete provided axial load
(axial load provided by the concrete core) and the steel tube pro-
vided axial load (axial load provided by the steel tube) are pre-
sented in Fig. 10.
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 373

Table 3
FRP jacket properties of specimens analyzed.

Specimen label Type Young’s modulus (MPa) Thickness (mm) Confining stiffness (MPa) Rupture strain
C1-0 N/A N/A N/A N/A N/A
C1-1 Carbon 255,000 0.17 556 0.0167
C1-2 Carbon 255,000 0.34 1112 0.0167
C2-0 N/A N/A N/A N/A N/A
C2-1 Carbon 255,000 0.17 347 0.0167
C2-2 Carbon 255,000 0.34 694 0.0167
1–1.5 Carbon 230,000 0.17 616 0.0055
1–2.5 Carbon 230,000 0.17 606 0.0055
1–3.5 Carbon 230,000 0.17 597 0.0055
1–4.5 Carbon 230,000 0.17 588 0.0055
2–1.5 Carbon 230,000 0.34 1231 0.0055
2–2.5 Carbon 230,000 0.34 1212 0.0055
2–3.5 Carbon 230,000 0.34 1194 0.0055
2–4.5 Carbon 230,000 0.34 1176 0.0055
F0-102 N/A N/A N/A N/A N/A
F1-102 Glass 80,100 0.17 134 0.0228
F2-102 Glass 80,100 0.34 267 0.0228
F3-102 Glass 80,100 0.51 401 0.0228
F0-135 N/A N/A N/A N/A N/A
F2-135 Glass 80,100 0.34 268 0.0228
F3-135 Glass 80,100 0.51 402 0.0228
F4-135 Glass 80,100 0.68 537 0.0228
F0-202 N/A N/A N/A N/A N/A
F2-202 Glass 80,100 0.34 268 0.0228
F3-202 Glass 80,100 0.51 402 0.0228
F4-202 Glass 80,100 0.68 537 0.0228
CSC A-0 N/A N/A N/A N/A N/A
CSC B-0 N/A N/A N/A N/A N/A
CSC C-0 N/A N/A N/A N/A N/A
CSC D-0 N/A N/A N/A N/A N/A
CSC E-0 N/A N/A N/A N/A N/A
CSC A-1 Carbon 228,000 0.167 600 0.0198
CSC B-1 Carbon 228,000 0.167 590 0.0198
CSC C-1 Carbon 228,000 0.167 581 0.0198
CSC D-1 Carbon 228,000 0.167 573 0.0198
CSC E-1 Carbon 228,000 0.167 560 0.0198
CSC A-2 Carbon 228,000 0.334 1199 0.0198
CSC B-2 Carbon 228,000 0.334 1181 0.0198
CSC C-2 Carbon 228,000 0.334 1163 0.0198
CSC D-2 Carbon 228,000 0.334 1145 0.0198
CSC E-2 Carbon 228,000 0.334 1120 0.0198
CFT N/A N/A N/A N/A N/A
CCFT-2L-1 Carbon 64,900 2.8 2391 0.0138
CCFT-2L-2 Carbon 64,900 2.8 2391 0.0138
t4C40 N/A N/A N/A N/A N/A
CF1t4C40 Carbon 250,000 0.111 434 0.0134
CF2t4C40 Carbon 250,000 0.222 867 0.0134
CF3t4C40 Carbon 250,000 0.333 1301 0.0134
GF1t4C40 Glass 109,000 0.169 288 0.0258
GF2t4C40 Glass 109,000 0.338 576 0.0258
GF3t4C40 Glass 109,000 0.507 864 0.0258
CF2t3C40 Carbon 250,000 0.222 867 0.0134
CF2t5C40 Carbon 250,000 0.222 867 0.0134
CF2t4C50 Carbon 250,000 0.222 867 0.0134
CF2t4C60 Carbon 250,000 0.222 867 0.0134
C-0-3.0 N/A N/A N/A N/A N/A
C-B1-3.0 Basalt 91,400 0.111 153 0.0260
C-B2-3.0 Basalt 91,400 0.222 305 0.0260
C-C1-3.0 Carbon 239,800 0.111 400 0.0170
C-C2-3.0 Carbon 239,800 0.222 801 0.0170
C-0-4.5 N/A N/A N/A N/A N/A
C-B1-4.5 Basalt 91,400 0.111 153 0.0260
C-B2-4.5 Basalt 91,400 0.222 305 0.0260
C-C1-4.5 Carbon 239,800 0.111 400 0.0170
C-C2-4.5 Carbon 239,800 0.222 801 0.0170
C-0-6.0 N/A N/A N/A N/A N/A
C-B1-6.0 Basalt 91,400 0.111 153 0.0260
C-B2-6.0 Basalt 91,400 0.222 305 0.0260
C-C1-6.0 Carbon 239,800 0.111 400 0.0170
C-C2-6.0 Carbon 239,800 0.222 801 0.0170
C-0-7.5 N/A N/A N/A N/A N/A
C-B1-7.5 Basalt 91,400 0.111 153 0.0260
C-B2-7.5 Basalt 91,400 0.222 305 0.0260
C-C1-7.5 Carbon 239,800 0.111 400 0.0170

(continued on next page)

374 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378

Table 3 (continued)

Specimen label Type Young’s modulus (MPa) Thickness (mm) Confining stiffness (MPa) Rupture strain
C-C2-7.5 Carbon 239,800 0.222 801 0.0170
F0-60 N/A N/A N/A N/A N/A
F1-60 Glass 80,100 0.17 165 0.0228
F2-60 Glass 80,100 0.34 330 0.0228
F3-60 Glass 80,100 0.51 495 0.0228
C3N N/A N/A N/A N/A N/A
C3F-1 Carbon 235,000 0.111 373 0.0149
C3F-3 Carbon 235,000 0.333 1120 0.0149
C4N N/A N/A N/A N/A N/A
C4F-1 Carbon 235,000 0.111 373 0.0149
C4F-2 Carbon 235,000 0.222 746 0.0149
C4F-3 Carbon 235,000 0.333 1120 0.0149
C6N N/A N/A N/A N/A N/A
C6F-1 Carbon 235,000 0.111 373 0.0149
C6F-3 Carbon 235,000 0.333 1120 0.0149

-0.020 -0.020
Measured Measured
Predicted Predicted
-0.015 Concrete cylinder strength = 46.0 MPa -0.015 Concrete cylinder strength = 35.9 MPa
Steel tube yield strength = 230 MPa Steel tube yield strength = 231 MPa
Steel tube thickness/diameter ratio = 0.0192 Steel tube thickness/diameter ratio = 0.0050
Lateral strain

Lateral strain
FRP confining stiffness = 0 MPa FRP confining stiffness = 537 MPa
-0.010 -0.010

-0.005 -0.005

C1-0 [6] F4-202 [10]

0.000 0.000
0.000 0.005 0.010 0.015 0.020 0.000 0.005 0.010 0.015 0.020
Axial strain Axial strain
(a) (a)
-0.020
-0.020
Measured
Measured
Predicted
Predicted
-0.015 Concrete cylinder strength = 42.1 MPa
Steel tube yield strength = 242 MPa -0.015 Concrete cylinder strength = 38.4 MPa
Steel tube thickness/diameter ratio = 0.0074 Steel tube yield strength = 248 MPa
Lateral strain

Steel tube thickness/diameter ratio = 0.0313

Lateral strain

FRP confining stiffness = 402 MPa

FRP confining stiffness = 576 MPa
-0.010
-0.010

-0.005
-0.005

F3-135 [10]
GF2tC40 [13]
0.000
0.000 0.005 0.010 0.015 0.020 0.000
0.000 0.005 0.010 0.015 0.020
Axial strain
(b) Axial strain
(b)
Fig. 1. Measured and predicted lateral strain-axial strain curves of (a) C1-0; and (b)
F3-135. Fig. 2. Measured and predicted lateral strain-axial strain curves of (a) F4-202; and
(b) GF2tC40.

From Fig. 9(a), it is seen that for a FRP jacketed CFST with FRP
confining stiffness = 1000 MPa, the FRP provided confining stress
increases with the axial strain right at the start of loading while
the total confining stress (FRP provided confining stress plus steel
tube provided confining stress) remains at zero due to steel-
concrete delamination. Later, after the formation of splitting
cracks, the concrete rapidly expands laterally and the concrete
eventually touches the steel tube to stop the delamination. Subse-
quently, the steel tube provided confining stress starts to increase
with the axial strain. However, at an axial strain of around 0.016,
the FRP ruptures and the FRP provided confining stress drops to
zero. Due to stress redistribution, the steel tube provided confining
stress abruptly increases and the total confining stress just drops
slightly. From Fig. 10(a), it is evident that at the initial stage, both
the concrete provided axial load and the steel tube provided axial
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 375

-0.030 2500
x
y= F4-202 [10]
-0.025
2000
R2 = 0.835
Measured lateral strain

-0.020

Axial load (kN)

1500
-0.015

1000 Concrete cylinder strength = 35.9 MPa

-0.010 Steel tube yield strength = 231 MPa
Steel tube thickness/diameter ratio = 0.0050
Tao et al. [6]
FRP confining stiffness = 537 MPa
-0.005 Hu et al. [10] 500
Xiao et al. [12] Measured
Lu et al. [13] Predicted
0.000 0
0.000 -0.005 -0.010 -0.015 -0.020 -0.025 -0.030
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Predicted lateral stain Axial strain
Fig. 3. Measured lateral strain versus predicted lateral strain at every 0.005 axial (a)
strain.
2000
GF2tC40 [13]
4000
C1-0 [6] 1500
3500
Axial load (kN)
3000

1000
Axial load (kN)

2500
Concrete cylinder strength = 38.4 MPa
2000 Steel tube yield strength = 248 MPa
Steel tube thickness/diameter ratio = 0.0313
Concrete cylinder strength = 46.0 MPa 500 FRP confining stiffness = 576 MPa
1500 Steel tube yield strength = 230 MPa
Steel tube thickness/diameter ratio = 0.0192 Measured
1000 FRP confining stiffness = 0 MPa Predicted
500 Measured 0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035
Predicted
0 Axial strain
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 (b)
Axial strain
Fig. 5. Measured and predicted axial load-strain curves of (a) F4-202; and (b)
(a) GF2tC40.

2500
F3-135 [10] 5000
x
2000 4500 y=
4000
R2 = 0.877
Measured axial load (kN)
Axial load (kN)

1500 3500

3000
Concrete cylinder strength = 42.1 MPa
1000 2500 Tao et al. [6]
Steel tube yield strength = 242 MPa
Gu et al. [9]
Steel tube thickness/diameter ratio = 0.0074
2000 Hu and Teng [10]
FRP confining stiffness = 402 MPa
Che et al. [11]
500 1500 Xiao et al. [12]
Measured Lu et al. [13]
Predicted 1000 Wei et al. [14]
Teng and Hu [16]
0 500
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 Park et al. [17]
0
Axial strain 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
(b) Predicted axial load (kN)
Fig. 4. Measured and predicted axial load-strain curves of (a) C1-0; and (b) F3-135.
Fig. 6. Measured axial load versus predicted axial load at every 0.005 axial strain.

load increase with the axial strain. After the steel tube has yielded,
From Fig. 9(b), it is seen that for a FRP jacketed CFST with FRP
the steel tube provided axial load starts to decrease while the con-
confining stiffness = 4000 MPa, both the FRP provided confining
crete provided axial load continues to increase. When the FRP rup-
stress and the total confining stress increase with the axial strain
tures, the total axial load abruptly drops and then gradually
right at the start of loading while the steel tube provided confining
decreases.
376 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378

20000 50
Required FRP confining stiffness (MPa)

Total confining stress

40 Steel tube provided confining stress
16000
FRP provided confining stress

Confining stress (MPa)

FRP confining stiffness = 1000 MPa
30 FRP rupture strain = 0.010
12000

20
8000

10
4000

0
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06
0.00 0.01 0.02 0.03 0.04
Steel tube thickness/diameter ratio
Axial strain
Fig. 7. Required FRP confining stiffness to eliminate steel-concrete delamination. (a)
50
1200 Total confining stress
Steel tube thickness/diameter ratio = 0.00 Steel tube provided confining stress
Required FRP confining stiffness (MPa)

40 FRP provided confining stress

Steel tube thickness/diameter ratio = 0.02
Confining stress (MPa)
Steel tube thickness/diameter ratio = 0.04 FRP confining stiffness = 4000 MPa
900 Steel tube thickness/diameter ratio = 0.06 30 FRP rupture strain = 0.010
Steel tube yield strength = 300 MPa

20
600

10

300
0

0 0.00 0.01 0.02 0.03 0.04

20 30 40 50 60 70 80 90 100 110 Axial strain
Concrete cylinder strength (MPa) (b)
(a)
Fig. 9. Confining stress-axial strain curves for FRP jacketed CFST with (a) FRP
1200 confining stiffness = 1000 MPa; and (b) FRP confining stiffness = 4000 MPa.
Steel tube thickness/diameter ratio = 0.00
Required FRP confining stiffness (MPa)

Steel tube thickness/diameter ratio = 0.02

Steel tube thickness/diameter ratio = 0.04 provided confining stress keeps on increasing with the axial strain
900 Steel tube thickness/diameter ratio = 0.06 until at an axial strain of around 0.034, the FRP ruptures and the
FRP provided confining stress drops to zero. Due to stress redistri-
Steel tube yield strength = 600 MPa
bution, the steel tube provided confining stress abruptly increases
600 but the total confining stress still drops quite dramatically. From
Fig. 10(b), it is evident that at the initial stage, both the concrete
provided axial load and the steel tube provided axial load increase
with the axial strain. After the steel tube has yielded, the steel tube
300 provided axial load ceases to increase while the concrete provided
axial load continues to increase. When the FRP ruptures, the total
axial load drops abruptly but still remains significantly higher than
the yield load of the structural system.
0
20 30 40 50 60 70 80 90 100 110 From the above, it can be seen that both the FRP confining stiff-
ness and rupture strain have great effects on the performance of
Concrete cylinder strength (MPa)
the FRP jacketed CFST system. With sufficient FRP confining stiff-
(b) ness provided, the steel-concrete delamination can be suppressed
Fig. 8. Required FRP confining stiffness to achieve Level I ductility at (a) steel tube
and thus the intrinsic problem of unavoidable delamination with
yield strength = 300 MPa; and (b) steel tube yield strength = 600 MPa. the pure CFST system can be resolved. Moreover, with sufficient
FRP confining stiffness and steel tube thickness provided, the resid-
ual axial load after FRP rupture can be maintained at higher than
stress becomes negative (the circumferential stress in the steel the yield load of the structural system, and thus the intrinsic prob-
tube is compressive and actually counteracting the FRP provided lem of brittle failure upon FRP rupture with the pure FRP confined
confining stress). No steel-concrete delamination occurs. The FRP concrete system can be resolved. Hence, the FRP jacketed CFST
 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378 377

2000 based on the von Mises yield criterion and associated flow rule.
Total axial load For the FRP jacket, the material is assumed to be linearly elastic
1800
Concrete provided axial load up to a certain rupture strain at which brittle failure would occur.
1600 Steel tube provided axial load The new model is used to analyze the lateral-to-axial strain
1400 curves and axial load-strain curves of a total of 89 FRP jacketed
FRP confining stiffness = 1000 MPa
CFST specimens tested by other researchers. The predicted
Axial load (kN)

1200 FRP rupture strain = 0.010

lateral-to-axial strain curves have been compared to a total of 25
1000 measured lateral-to-axial strain curves. It is found that the new
model can predict the lateral strains of the FRP jacketed CFST spec-
800
imens quite accurately with an average absolute error of 20.2%. On
600 the other hand, the predicted axial load-strain curves have been
400
compared to a total of 89 measured axial load-strain curves, and
the comparison reveals that the new model can predict the axial
200 loads of the FRP jacketed CFST specimens quite accurately with
0 an average absolute error of 7.7%. Hence, the new model has been
0.00 0.01 0.02 0.03 0.04 verified to have a reasonably good accuracy for the analysis of the
Axial strain full range behaviour of FRP jacketed CFST columns.
(a) Lastly, the new model is used to evaluate the minimum FRP
confining stiffness required to eliminate steel-concrete delamina-
2000
tion and the minimum FRP confining stiffness required to achieve
1800 Total axial load Level I ductility (no strain softening after yielding), and to study
Concrete provided axial load the effectiveness of the FRP jacket. It is found that the required
1600 Steel tube provided axial load
FRP confining stiffness to eliminate delamination is a linearly
1400 FRP confining stiffness = 4000 MPa increasing function of the steel tube thickness/diameter ratio. It
Axial load (kN)

1200 FRP rupture strain = 0.010 is also found that the required FRP confining stiffness to achieve
Level I ductility is higher at higher concrete strength, lower at
1000 higher steel tube yield strength and lower at larger steel tube
800 thickness/diameter ratio, and that in general, the required ductility
can be met with by increasing the steel tube thickness and/or the
600
FRP confining stiffness. More importantly, the study on the effec-
400 tiveness of FRP jacket, though rather preliminary, reveals that with
200 sufficient FRP confining stiffness provided, the intrinsic problem of
unavoidable delamination with the pure CFST system can be
0 avoided and with sufficient FRP confining stiffness and steel tube
0.00 0.01 0.02 0.03 0.04
thickness provided, the intrinsic problem of brittle failure upon
Axial strain FRP rupture with the pure FRP confined concrete system can be
(b) overcome. Hence, the FRP jacketed CFST system can be designed
to be superior to both the pure CFST system and the pure FRP con-
Fig. 10. Axial load-strain curves for FRP jacketed CFST with (a) FRP confining
stiffness = 1000 MPa; and (b) FRP confining stiffness = 4000 MPa.
fined concrete system. Further parametric studies on how the FRP
jacketed CFST should be designed are recommended.

system can be designed to be superior to both the pure CFST sys-
tem and the pure FRP confined concrete system. However, the
combined effects of the FRP confining stiffness and rupture strain
are fairly complicated. Further research is recommended to find
out the minimum FRP confining stiffness and minimum steel tube
thickness needed to ensure that after FRP rupture, the residual load
would remain higher than the yield load of the system so as to
avoid the brittle failure mode due to FRP rupture.
6. Conclusions
A new theoretical model for analyzing the axial and lateral
stress-strain behaviour of FRP jacketed CFST columns has been
developed by integrating a lateral-to-axial strain model of confined
concrete, an axial stress-strain model of confined concrete, a biax-
ial stress-strain model of steel and a uniaxial stress-strain model of
FRP. The lateral-to-axial strain model is capable of evaluating the
lateral strain of confined concrete at various stages of loading by
considering the elastic and inelastic lateral strains separately. The
axial stress-strain model of confined concrete is based on Attard
and Setunge’s model, which is applicable to a broad range of con-
crete strength and has been verified by extensive test results. For
the steel tube, the material is assumed to be linearly elastic-
perfectly plastic and the biaxial stress-strain model adopted is
References
[1] Ho JCM, Lam JKY, Kwan AKH. Effectiveness of adding confinement for ductility
improvement of high-strength concrete columns. Eng Struct 2010;32
(3):714–25.
[2] Schneider SP. Axially loaded concrete-filled steel tubes. J Struct Eng-ASCE
1998;124(10):1125–38.
[3] Sakino K, Nakahara H, Morino S, Nishiyama I. Behavior of centrally loaded
concrete-filled steel-tube short columns. J Struct Eng-ASCE 2004;130
(2):180–8.
[4] Giakoumelis G, Lam D. Axial capacity of circular concrete-filled tube columns. J
Constr Steel Res 2004;60(7):1049–68.
[5] Mao XY, Xiao Y. Seismic behavior of confined square CFT columns. Eng Struct
2006;28(10):1378–86.
[6] Tao Z, Han LH, Zhuang JP. Axial loading behavior of CFRP strengthened
concrete-filled steel tubular stub columns. Adv Struct Eng 2007;10(1):37–46.
[7] Liu L, Lu YY. Axial bearing capacity of short FRP confined concrete-filled steel
tubular columns. J Wuhan Univ Technol 2010;25(3):454–8.
[8] Park JW, Hong YK, Choi SM. Behaviors of concrete filled square steel tubes
confined by carbon fiber sheets (CFS) under compression and cyclic loads. Steel
Compos Struct 2010;10(2):187–205.
[9] Gu W, Guan CW, Zhao YH, Cao H. Experimental study on concentrically-
compressed circular concrete filled CFRP-steel composite tubular short
columns. J Shenyang Archit Civil Eng Univ 2004;20(2):118–20 [in Chinese].
[10] Hu YM, Yu T, Teng JG. FRP-confined circular concrete-filled thin steel tubes
under axial compression. J Compos Constr-ASCE 2011;15(5):850–60.
[11] Che Y, Wang QL, Shao YB. Compression performances of the concrete filled
circular CFRP-steel tube (C-CFRP-CFST). Adv Steel Constr 2012;8(4):331–58.
[12] Xiao Y, He WH, Choi KK. Confined concrete-filled tubular columns. J Struct
Eng-ASCE 2005;131(3):488–97.
[13] Lu YY, Li N, Li S. Behavior of FRP-confined concrete-filled steel tube columns.
Polymers 2014;6(5):1333–49.
378 C.X. Dong et al. / Engineering Structures 126 (2016) 365–378
[14] Wei Y, Wu G, Li GF. Performance of circular concrete-filled fiber-reinforced
polymer-steel composite tube columns under axial compression. J Reinf Plast
Compos 2014;33(20):1911–28.
[15] Fam AZ, Rizkalla SH. Concrete-filled FRP tubes for flexural and axial
compression members. In: Proceedings of third international conference on
advanced composite materials for bridges and structures, Ottawo, Ontario,
August 2000. p. 315–22.
[16] Teng JG, Hu YM. Theoretical model for FRP confined circular concrete-filled
steel tubes under axial compression. In: Proceedings of third international
conference on FRP composites in civil engineering, Miami, Florida, USA,
December 2006. p. 503–6.
[17] Park JW, Hong YK, Hong GS, Kim JH, Choi SM. Design formulas of concrete
filled circular steel tubes reinforced by carbon fiber reinforced plastic sheets.
Proc Eng 2011;14:2916–22.
[18] Han LH, Yao GH, Zhao XL. Tests and calculations for hollow structural steel
(HSS) stub columns filled with self-consolidating concrete (SCC). J Constr Steel
Res 2005;61(9):1241–69.
[19] Dong CX, Kwan AKH, Ho JCM. Effects of confining stiffness and rupture strain
on performance of FRP confined concrete. Eng Struct 2015;97:1–14.
[20] Choi KK, Xiao Y. Analytical model of circular CFRP confined concrete-filled
steel tubular columns under axial compression. J Compos or Constr-ASCE
2010;14(1):125–33.
[21] Teng JG, Hu YM, Yu T. Stress-strain model for concrete in FRP-confined steel
tubular columns. Eng Struct 2013;49:156–67.
[22] Dong CX, Kwan AKH, Ho JCM. A constitutive model for predicting the lateral
strain of confined concrete. Eng Struct 2015;91:155–66.
[23] Attard MM, Setunge S. Stress-strain relationship of confined and unconfined
concrete. ACI Mater J 1996;93(5):432–42.
[24] Kwan AKH, Dong CX, Ho JCM. Axial and lateral stress-strain model for FRP
confined concrete. Eng Struct 2015;99:285–95.
[25] Ouyang Y, Lo SH, Kwan AKH, Ho JCM. A new analysis method for polymer-
confined concrete columns In; Proceedings, institution of civil engineers,
structures and buildings; 2016, http://dx.doi.org/10.1680/jstbu.15.00140.
[26] Lo SH, Kwan AKH, Ouyang Y, Ho JCM. Finite element analysis of axially loaded
FRP-confined rectangular concrete columns. Eng Struct 2015;100:253–63.
[27] Ozbakkaloglu T, Lim JC, Vincent T. FRP-confined concrete in circular sections:
review and assessment of stress-strain models. Eng Struct 2013;49:1068–88.
[28] Tao Z, Wang ZB, Yu Q. Finite element modelling of concrete-filled steel stub
columns under axial compression. J Constr Steel Res 2013;89:121–31.
[29] Xiao QG, Teng JG, Yu T. Behavior and modeling of confined high-strength
concrete. J Compos Constr-ASCE 2010;14(3):249–59.
[30] Chen WF, Saleeb AF. Constitutive equations for engineering materials. New
York: Wiley; 1982.
[31] Hunaiti YM. Composite action of foamed and lightweight aggregate concrete. J
Mater Civil Eng-ASCE 1996;8(3):111–3.
[32] Mouli M, Khelafi H. Strength of short composite rectangular hollow section
columns filled with lightweight aggregate concrete. Eng Struct 2007;29
(8):1791–7.