Composite Structures 200 (2018) 810–828

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Stress-strain behavior of concrete in circular concrete columns partially T

wrapped with FRP strips
⁎
Jun-Jie Zenga,b, Yong-Chang Guoa, , Wan-Yang Gaoc, Wei-Peng Chena, Li-Juan Lia
a
School of Civil and Transportation Engineering, Guangdong University of Technology, Guangzhou 510006, China
b
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong 999077, China
c
School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

A R T I C LE I N FO A B S T R A C T

Keywords: Fiber-reinforced polymer (FRP) wrapping has become an attractive strengthening technique for concrete col-
FRP umns. Within this strengthening technique, FRP jackets are wrapped around the concrete column with the fibers
Circular column in the jacket being oriented in the hoop direction. In practice, the FRP jackets can be either continuous or
FRP partially wrapped concrete discontinuous along the column height and thus the resulting column is referred to as FRP fully or partially
Stress-strain model
wrapped concrete columns. Existing research has demonstrated that the FRP partial wrapping strengthening
Confinement
technique by discrete FRP strips (rings) is a promising and economic alternative to the FRP full wrapping
FRP strip/ring
strengthening technique. Although a number of experimental investigations have been conducted on FRP par-
tially wrapped concrete columns, the stress-strain behavior of FRP-confined concrete in partially wrapped
concrete columns is not yet completely understood. This paper presents an experimental program to investigate
the axial compressive behavior of circular concrete columns partially wrapped with FRP strips. The test results
are presented and compared with five existing representative stress-strain models to examine the reliability and
accuracy of each model. It has been demonstrated that the Teng et al.’s (2007) model is superior to the other four
representative models and it provides reasonably accurate predictions of the ultimate axial stress of FRP partially
wrapped concrete while it usually underestimates the ultimate axial strain.

1. Introduction only requiring a considerable increase in axial deformation capacity. In

Over the past three decades, fiber-reinforced polymer (FRP) has
emerged as a favorite material for the strengthening of existing concrete
columns. The columns are usually wrapped with FRP jacket along the
entire column height, and the resulting column is referred to as an FRP
fully wrapped concrete column. Existing research has demonstrated
that the full FRP confinement can significantly enhance the compressive
strength and deformation capacities of circular concrete columns
[3,6,9–12,15–17,20,24,26–28,32,36,40–42,45,48]. Alternatively, the
concrete column can be wrapped with longitudinally discrete (i.e.,
spaced) FRP strips/rings (Fig. 1), which is referred to as a FRP partially
wrapped concrete column. Although most of the existing studies are
related to FRP fully wrapped concrete columns, FRP partially wrapped
concrete columns have also been demonstrated to process an adequate
increase in strength and a remarkable increase in axial deformation
capacity compared with their counterparts (i.e., un-confined concrete
columns) (e.g., [4,25,43,5,28,46,47]. As a result, FRP partially wrapped
concrete columns are comparatively favorable, especially for columns
addition, strengthening columns with discrete FRP strips is expected to
be able to avoid FRP buckling failure which will easily occurred in
concrete-filled FRP tubes as the axial stiffness of the FRP tube cannot be
neglected. Also, less FRP materials are needed for FRP partially
wrapped concrete columns and thus FRP partial wrapping strength-
ening can be applied easier and faster than FRP full wrapping
strengthening [28,46].
The interest to understand the behavior of FRP partially wrapped
concrete has led to a few experimental studies on the behavior of
concrete in FRP partially wrapped columns [4,25,5,28,37,29,46,47,18].
The confinement in the axially loaded circular columns fully wrapped
by FRP jackets is uniform. However, as for the confined concrete in FRP
partially wrapped circular columns, the confinement is non-uniform
(Fig. 2) within the concrete between two adjacent FRP strips [30,19].
The usage of discrete FRP strips results in the less efficient confinement
to the concrete between the two adjacent FRP strips. The confinement
mechanism is similar to the concrete confined by steel hoops or spirals,
in which the reduced confinement effect between two adjacent strips

⁎
Corresponding author at: Room 203, Structural Laboratory, School of Civil and Transportation Engineering, Guangdong University of Technology, Guangzhou Higher Education Mega
Center, Guangzhou, China.
E-mail address: guoyc@gdut.edu.cn (Y.-C. Guo).

https://doi.org/10.1016/j.compstruct.2018.05.001
Received 12 October 2017; Received in revised form 11 March 2018; Accepted 2 May 2018
Available online 04 May 2018
0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
J.-J. Zeng et al.
bf
sf' sf
D-sf' /2
D strain responses [44,46]. However, most of the existing models estab-
Fig. 1. Schematic of an FRP partially wrapped concrete column.
can be considered through the “arching action”. Based on the arching
action assumption, a parabola with an initial slope of 45° covering a
clear vertical spacing (s′f ) between two FRP strips is defined to separate
the effectively confined part from the ineffectively confined part of the
concrete section (Fig. 2a). As illustrated in Fig. 2b, the ineffective
confinement area is located between two adjacent FRP strips. Subse-
quently, the axial deformation of FRP partially wrapped concrete col-
umns is believed to concentrate at the ineffectively-confined concrete
between the FRP strips and some experimental evidence of the axial
deformation distribution has been observed experimentally by the au-
thors (e.g., [46,47]). The difference of confinement mechanisms be-
tween FRP fully and partially wrapped concrete implies different be-
haviors between them, which may lead to the difference in stress-strain
curves between the confined concrete in FRP partially wrapped con-
crete columns and that in FRP fully wrapped concrete columns. As a
result, the reliability and accuracy of existing theoretical models for
bf
sf' sf
FRP
D'
A D
(a) Arching action
Fig. 2. Arching action and effective confinement area in a circular column partially wrapped with FRP.
Composite Structures 200 (2018) 810–828
FRP fully wrapped concrete need to be carefully examined for their
applicability for FRP partially wrapped concrete.
The stress-strain behavior of FRP-confined concrete can be appro-
priately described by either a design-oriented stress-strain model which
has closed-form equations or an analysis-oriented stress-strain model
which is based on an incremental numerical procedure. Based on ex-
isting experimental observations, the stress-strain response of FRP-
confined concrete can be classified into two types [15,25,46], as shown
in Fig. 3. For the concrete with sufficient FRP confinement, the stress-
strain curve usually exhibits two segments (Type I): first ascending
segment and second linear ascending segment (Fig. 3a). During the first
segment, the concrete is initially in the elastic stage where the stress
increases linearly with strain and the load is predominantly sustained
by the concrete core. When near the turning point, the concrete begins
to crack and the dilation of concrete during the second segment is
constrained by the FRP jacket, leading to substantial increases in the
axial stress and strain of concrete. However, the stress-strain curve of
concrete with insufficient FRP confinement exhibits a strain softening
response (Fig. 3b and c) (Type II). In this type of stress-strain response,
the stress at ultimate strain is smaller than the peak stress. The strain
softening behavior of Type II can be further divided into Type II-I with
′ ⩾ fc0
fcu ′ (Fig. 3b) and Type II-I with fcu ′ < fc0
′ (Fig. 3c) [15]. It is seen
that the performances are different for the two different types of stress-
lished based on FRP fully wrapped concrete do not distinguish the
abovementioned two types, and thus some of them are only applicable
to Type-I stress-strain curve. Therefore, the reliability and accuracy of
the existing stress-strain models need to be examined for FRP partially
wrapped concrete.
In addition, the existing test results enable the evaluation and ver-
ification of available design guidelines (e.g., [8,7]) for FRP partially
wrapped concrete columns (e.g., [46,47]). As per these codes, a long-
itudinal confinement effectiveness factor is suggested for determining
the compressive strength and ultimate axial strain of FRP partially
wrapped concrete, and therefore, the FRP partially wrapped concrete
columns can be designed with an acceptable accuracy. In other words,
the different combinations of the FRP thickness and the FRP strip clear
spacing can obtain identical effective FRP confinements which are same
to those occurred in FRP fully wrapped concrete. However, an increase
in the FRP strip clear spacing may lead to the concrete crushing failure,
as found in a previous experimental study conducted by the authors
[46], which in turn influences the stress-strain behavior of FRP partially
wrapped concrete. Therefore, as stated in Zeng et al. [46], the accuracy
Section A-A
A
(b) Effective confinement area
811
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

(a) Type I: increasing type (b) Type II-I: decreasing type ( > )

(c) Type II-II: decreasing type ( < )

Fig. 3. Classifications of stress-strain curves of FRP fully wrapped concrete.

Fig. 4. Illustration of the FRP strip spacings considered in the experimental program.

of the longitudinal confinement effectiveness factor needs to be further
examined using more test data.
To this end, an experimental program was conducted to study the
axial compressive behavior of circular columns wrapped with CFRP
strips. In total 60 columns were prepared and tested. The main test
variable examined in this experimental program included the clear
spacing and width of FRP strips. The test results, in term of the effects of
the clear spacing and width of FRP strips on the axial compressive
behavior of FRP partially wrapped concrete, were presented and dis-
cussed. A widely accepted analysis-oriented stress-strain model pro-
posed by Teng et al. [33] was used to predict the stress-strain behavior
of FRP partially wrapped concrete; five representative models were
employed to predict the ultimate condition of FRP partially wrapped
concrete. The reliability and accuracy of these models were examined
by the comparisons between the test results and the model predictions.
2. Experimental program
2.1. Test specimens
Totally sixty FRP-confined cylindrical column specimens were
tested to investigate the effects of FRP strip width, clear spacing and
FRP thickness. All specimens had a diameter of 150 mm and a height of
300 mm. Note that the clear spacing and width of FRP strips are the
main parameters studied in the current study and they represent the
coefficient of FRP vertical efficiency. Note that the vertical confinement
effectiveness coefficient, which is a coefficient representing the coeffi-
cient of FRP vertical efficiency, is given by (1−s′f /2D)2 where s′f is the

812
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

Table 1
Key information of test columns and key test results.
Series Specimen Width of FRP strips Thickness of FRP strips Number of FRP strips s′f / D ′
fco ′
fcc εcc ′
fcu εcu ′ / fco
fcu ′ εcu/ εco εh,rup

/ SC-1 — — — / 23.27 / / / / / / /
SC-2 / 22.98 / / / / / / /
SC-3 / 23.86 / / / / / / /

I SP-1-1 / 0.167 / 0.00 23.40 62.13 0.0295 / / 2.66 10.54 −0.0141

SP-1-2 / 0.167 / 0.00 23.40 59.06 0.0293 / / 2.52 10.46 −0.0158
SP-2-1 / 0.334 / 0.00 23.40 88.55 0.0395 / / 3.78 14.11 −0.0156
SP-2-2 / 0.334 / 0.00 23.40 88.10 0.0389 / / 3.76 13.89 −0.0150
SP-3-1 / 0.501 / 0.00 23.40 108.98 0.0516 / / 4.66 18.43 −0.0154
SP-3-2 / 0.501 / 0.00 23.40 110.62 0.0507 / / 4.73 18.11 −0.0164

II S-1-3-25-1 25 0.167 3 0.75 23.40 25.70 0.0039 21.27 0.0116 0.91 1.39 −0.0107
S-1-3-25-2 25 0.167 3 0.75 23.40 25.38 0.0035 19.38 0.0140 0.83 1.25 −0.0122
S-1-3-30-1 30 0.167 3 0.70 23.40 25.48 0.0060 22.12 0.0135 0.95 2.14 −0.0098
S-1-3-30-2 30 0.167 3 0.70 23.40 25.26 0.0047 22.03 0.0097 0.94 1.68 −0.0110
S-1-3-35-1 35 0.167 3 0.65 23.40 25.21 0.0039 21.24 0.0160 0.91 1.39 −0.0138
S-1-3-35-2 35 0.167 3 0.65 23.40 26.60 0.0034 24.63 0.0142 1.05 1.21 −0.0115
S-2-3-25-1 25 0.334 3 0.75 23.40 27.43 0.0034 22.34 0.0134 0.95 1.21 −0.0109
S-2-3-25-2 25 0.334 3 0.75 23.40 26.89 0.0057 20.20 0.0124 0.86 2.04 −0.0115
S-2-3-30-1 30 0.334 3 0.70 23.40 29.54 0.0058 23.37 0.0165 1.00 2.07 −0.0126
S-2-3-30-2 30 0.334 3 0.70 23.40 30.14 0.0070 23.18 0.0178 0.99 2.50 −0.0145
S-2-3-35-1 35 0.334 3 0.65 23.40 29.23 0.0099 24.27 0.0224 1.04 3.54 −0.0126
S-2-3-35-2 35 0.334 3 0.65 23.40 29.08 0.0088 20.27 0.0248 0.87 3.14 −0.0117
S-3-3-25-1 25 0.501 3 0.75 23.40 29.72 0.0078 20.44 0.0279 0.87 2.79 −0.0108
S-3-3-25-2 25 0.501 3 0.75 23.40 28.19 0.0105 19.41 0.0214 0.83 3.75 −0.0109
S-3-3-30-1 30 0.501 3 0.70 23.40 27.93 0.0085 22.93 0.0200 0.98 3.04 −0.0110
S-3-3-30-2 30 0.501 3 0.70 23.40 29.28 0.0085 24.15 0.0282 1.03 3.04 −0.0106
S-3-3-35-1 35 0.501 3 0.65 23.40 33.37 0.0074 25.20 0.0154 1.08 2.64 −0.0093
S-3-3-35-2 35 0.501 3 0.65 23.40 32.23 0.0062 27.30 0.0207 1.17 2.21 −0.0084

III S-1-4-25-1 25 0.167 4 0.44 23.40 27.55 0.0092 / / 1.18 3.29 −0.0109
S-1-4-25-2 25 0.167 4 0.44 23.40 26.41 0.0092 / / 1.13 3.29 −0.0115
S-1-4-30-1 30 0.167 4 0.40 23.40 28.73 0.0132 / / 1.23 4.71 −0.0128
S-1-4-30-2 30 0.167 4 0.40 23.40 29.06 0.0099 / / 1.24 3.54 −0.0104
S-1-4-35-1 35 0.167 4 0.36 23.40 32.92 0.0099 32.60 0.0106 1.39 3.54 −0.0111
S-1-4-35-2 35 0.167 4 0.36 23.40 33.55 0.0117 33.36 0.0121 1.43 4.18 −0.0118
S-2-4-25-1 25 0.334 4 0.44 23.40 30.74 0.0098 30.64 0.0119 1.31 3.50 −0.0132
S-2-4-25-2 25 0.334 4 0.44 23.40 34.95 0.0106 34.54 0.0118 1.48 3.79 −0.0112
S-2-4-30-1 30 0.334 4 0.40 23.40 35.65 0.0142 / / 1.52 5.07 −0.0109
S-2-4-30-2 30 0.334 4 0.40 23.40 35.95 0.0154 / / 1.54 5.50 −0.0107
S-2-4-35-1 35 0.334 4 0.36 23.40 40.14 0.0207 / / 1.72 7.39 −0.0111
S-2-4-35-2 35 0.334 4 0.36 23.40 40.19 0.0161 / / 1.72 5.75 −0.0110
S-3-4-25-1 25 0.501 4 0.44 23.40 35.97 0.0193 35.52 0.0235 1.52 6.89 −0.0139
S-3-4-25-2 25 0.501 4 0.44 23.40 35.92 0.0248 35.25 0.0296 1.51 8.86 −0.0116
S-3-4-30-1 30 0.501 4 0.40 23.40 41.04 0.0328 / / 1.75 11.71 −0.0162
S-3-4-30-2 30 0.501 4 0.40 23.40 40.38 0.0280 / / 1.30 10.00 −0.0139
S-3-4-35-1 35 0.501 4 0.36 23.40 48.64 0.0397 / / 2.08 14.18 −0.0154
S-3-4-35-2 35 0.501 4 0.36 23.40 45.00 0.0448 / / 1.92 16.00 −0.0164

IV S-1-5-25-1 25 0.167 5 0.29 23.40 31.09 0.0187 / / 1.33 6.68 −0.0106

S-1-5-25-2 25 0.167 5 0.29 23.40 28.88 0.0177 / / 1.23 6.32 −0.0110
S-1-5-30-1 30 0.167 5 0.25 23.40 35.41 0.0228 / / 1.51 8.14 −0.0126
S-1-5-30-2 30 0.167 5 0.25 23.40 36.22 0.0224 / / 1.55 8.00 −0.0138
S-1-5-35-1 35 0.167 5 0.21 23.40 38.53 0.0232 / / 1.65 8.29 −0.0156
S-1-5-35-2 35 0.167 5 0.21 23.40 38.51 0.0232 / / 1.65 8.29 −0.0133
S-2-5-25-1 25 0.334 5 0.29 23.40 42.49 0.0365 / / 1.82 13.04 −0.0137
S-2-5-25-2 25 0.334 5 0.29 23.40 41.99 0.0304 / / 1.79 10.86 −0.0144
S-2-5-30-1 30 0.334 5 0.25 23.40 42.03 0.0238 / / 1.80 8.50 −0.0120
S-2-5-30-2 30 0.334 5 0.25 23.40 47.53 0.0331 / / 2.03 11.82 −0.0136
S-2-5-35-1 35 0.334 5 0.21 23.40 51.62 0.0287 / / 2.21 10.25 −0.0143
S-2-5-35-2 35 0.334 5 0.21 23.40 52.03 0.0323 / / 2.22 11.54 −0.0168
S-3-5-25-1 25 0.501 5 0.29 23.40 47.98 0.0507 / / 2.05 18.11 −0.0134
S-3-5-25-2 25 0.501 5 0.29 23.40 46.83 0.0541 / / 2.00 19.32 −0.0148
S-3-5-30-1 30 0.501 5 0.25 23.40 58.80 0.0381 / / 2.51 13.61 −0.0140
S-3-5-30-2 30 0.501 5 0.25 23.40 59.63 0.0365 / / 2.55 13.04 −0.0137
S-3-5-35-1 35 0.501 5 0.21 23.40 64.74 0.0422 / / 2.77 15.07 −0.0157
S-3-5-35-2 35 0.501 5 0.21 23.40 63.17 0.0369 / / 2.70 13.18 −0.0141

clear spacing of two adjacent FRP strips and D is the diameter of the
column [8]. The variation of coefficient of FRP vertical efficiency was
achieved by wrapping concrete columns with FRP strips (3, 4 and 5
strips) with different widths (i.e., 25, 30, and 35 mm) (Fig. 4). For ease
of reference, the specimens were divided into four series and the
number of FRP strips in each series is identical, as is shown in Table 1.
The design of strip number and strip width led to several different clear
spacing ratios in present study (Table 1). It is noted that the ratio be-
tween the clear spacing of two adjacent FRP strips and the column
diameter is referred to as clear spacing ratio (i.e., s′f / D ), as shown in
Fig. 1. Each column has two nominally identical specimens. The details
about the test columns (including the width of FRP strips, the FRP

813
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

jacket thickness, number of FRP strips, clear spacing ratio and un-
confined concrete strength) are given in Table 1.
Each column was given a name (Table 1), which starts with the
letter “S” to represent “specimen”, followed by a number indicating the
number of FRP layer/layers. This is followed by a number presenting
the total number of FRP strips and then followed by the width of the
FRP strips. The last number represents the number of the nominally
identical specimens. In addition, “SC” refers to unconfined concrete
cylindrical specimens and “SP” refers to fully wrapped specimens.

2.2. Preparation of specimens and material properties

Normal concrete made of river sand and natural gravels was used to
cast the column specimens. Standard cylinders were also cast with the
same batch concrete to determine the mechanical properties of the
unconfined concrete. After the concrete columns had been cured at
room temperature for 28 days, they were wrapped with carbon FRP
(CFRP) jacket (strips) using a wet layup process, with fibers oriented
(a) Instrumentations
only in the hoop direction. A 150-mm-long overlapping zone between
the starting end and the finishing end of each sheet was allowed to
ensure the fully-developed tensile strength of the FRP jacket. An addi-
tional 25-mm-width FRP strip was wrapped near each end of the
column to avoid unexpected failure there.
FRP tensile tests were conducted to determine the material prop-
erties of FRP jacket. In the tensile test, a unidirectional one-layer CFRP
coupon was employed. The test procedure followed what is specified by
the ASTM standard [1]. Five flat coupons were tested and results were
only based on coupons failed by rupture at the middle. The average
values of the tensile strength, tensile elastic modulus and the rupture
strain are 4477.6 MPa, 249.1 GPa and 1.66%, respectively, according to
ASTM D3039 [1].
The concrete mix proportioning is given in Table 2. The properties (b) Test setup
of unconfined concrete were obtained from compression tests on three
standard concrete cylinders in accordance with ASTM C469 [2] and the Fig. 5. Instrumentations and test setup.
unconfined concrete strength fco ′ and strain εco is determined by the
compressive strength and strain of concrete cylinders (i.e., fc′ and εc ), as simultaneously by a data logger.
given in Table 1.

2.3. Instrumentation and test setup 3. Test results and discussion

Two linear variable displacement transducers (LVDTs) were used to
measure the axial shortening of the column in the mid-height region.
The two LVDTs were installed at two opposite locations in each of the
columns (Fig. 5a) with a gauge length of 120 mm. The full-height axial
shortening was measured using a LVDT mounted in the testing ma-
chine.
All specimens were tested under concentric compression using a
loading machine with a load-carrying capacity of 5000 kN (Fig. 5b).
Each of the column ends were leveled using high-strength gypsum
mortar so that the axial load could be simultaneously applied on the
whole section. To ensure the centering of the specimen, each specimen
was loaded to around 30% of the loading capacity of the unconfined
column to check the test alignment. Calibration was carried out by
examining the readings from the four vertical strain gauges. The spe-
cimen was unloaded, realigned, and loaded again if the readings differ
each other by 10% at the time the applied load reached about 30% of its
unconfined column strength. Column compression test was realized
using displacement control with a rate of 0.4 mm/min was used. All test
data, including the loads, displacements, and strains were recorded
Table 2
Mix proportion of concrete.
Cement (kg) Sand (kg) Aggregate (kg) Water (kg)
1 2.02 3.59 0.71
3.1. Failure modes
Fig. 6 shows the failure models of selected FRP partially wrapped
concrete columns. Note that only failure modes of selected specimens
are presented in the Fig. 6 as examples because the failure modes are
similar for specimens in each group. Obviously, the damage of un-
confined specimens initiated with inclined cracking of concrete and
they failed by concrete crushing failure. It was not surprising to find
that the FRP fully wrapped columns failed by rupture of FRP jacket
(Fig. 6a). For these column specimens, some clicking sounds of epoxy
rupture were first heard, followed by a sudden explosive sound of FRP
rupture. FRP rupture occurred at or near mid-height and outside the
overlap zone, with crushing of concrete and sudden loss of load capa-
city simultaneously, which is in accordance with the experimental ob-
servations from some other researchers (e.g., [26,46]).
For columns partially wrapped with FRP strips, the failure depended
on their strip numbers and thickness. Specimens confined with three
FRP strips experienced either FRP rupture failure or concrete crushing
failure: specimens with one and two layers of FRP experienced FRP
rupture failure at the mid-height, while it was interesting to find that
specimens with three layers of FRP jacket experienced concrete
crushing failure (Fig. 6b). This was because the effective confinement in
the latter was larger than that in the former under a given deformation,
which led to substantial increase in axial stress of concrete. Subse-
quently, the concrete between two adjacent FRP strips failed by
crushing.

814
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

(a) Fully wrapped specimens

(b) Specimens confined with three FRP strips

(c) Specimens confined with four FRP strips

(d) Specimens confined with five FRP strips

Fig. 6. Failure modes of selected specimens.

The specimens confined with four or five FRP strips generally failed
by rupture of one of the FRP strips which were near the mid-height
(Fig. 6c-d). The failure initiated with the concrete between adjacent
FRP strips cracked when the average axial stress approximately reached
the unconfined concrete strength. Then the crack grew and FRP rupture
occurred, with crushing of concrete simultaneously. The concrete be-
tween two adjacent FRP strips did not experience crushing failure as the
clear spacing of FRP strips was small enough so the ineffective con-
finement area was not large. Likewise, the concrete crushing was more
evident for specimens with larger clear spacing of FRP strips, implying
concentration of localized failure of concrete between adjacent FRP
strips where the confinement was non-uniform (Fig. 2). Similarly, the
concrete outside the FRP strips cracked at the time the average axial
stresses approximately approached to the unconfined concrete strength
for specimens wrapped with five FRP strips. Subsequently, some strands
in the mid-height FRP strip ruptured with the loading increase. At last,
the mid-height FRP strip ruptured (Fig. 6c-d). Paradoxically, the
concrete cracking failure at mid-height of specimens wrapped with four
FRP strips was more obvious than that of specimens wrapped with three
and five FRP strips. This is probably because the concrete at mid-height
was not effectively confined for specimens wrapped with four FRP
strips and the FRP rupture failure did not take place at the column ends
because of the existence of additional friction from the loading platens
at the column ends [35].
3.2. Stress-Strain responses
The stress-strain responses of concrete are shown in Figs. 7 and
9–11. Note that in this article the stress-strain curves refer to axial
stress-axial strain curves and axial stress-hoop strain curves unless
otherwise specified. For stress-strain curves in present paper, the fol-
lowing sign convention is adopted: in concrete, compressive stresses
and strains are taken to be positive, while in FRP jackets, tensile stresses
and strains are taken to be negative. All the stress-strain curves in

815
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

70 100

90
60
80

Axial stress (MPa) 50 70

60

Axial stress (MPa)

40
50
30
40

20 30

SP-1-1 20 SP-2-1
10 SP-1-2 SP-2-2
10
Teng et al. (2007) Teng et al. (2007)
0 0
-20 -15 -10 -5 0 5 10 15 2025 30 35 -20 -10 0 10 20 30 40 50
-3 -3 -3 -3
Hoop strain 10 Axial strain 10 Hoop strain (10 ) Axial strain (10 )

(a) SP-1 (b) SP-2

120
110
100
90
80
70
Axial stress (MPa)

60
50
40
30
SP-3-1
20
SP-3-2
10 Teng et al. (2007)
0
-20 -10 0 10 20 30 40 50 60 70
-3 -3
Hoop strain (10 ) Axial strain (10 )

(c) SP-3
Fig. 7. Stress-strain curves for Series I specimens.

Figs. 7 and 9–11 are presented based on following rule: the axial stress- experimental study presented in the current paper. The axial stress-
hoop strain curve shown on the left while the axial stress-axial strain strain curve of FRP-confined concrete in partially wrapped circular
curve shown on the right. The axial stresses of concrete were calculated columns is appropriately divided into three segments (Fig. 8): (a) first
by dividing the axial load it takes by its area. The axial strains from linear ascending segment: during this segment the specimen is in the
LVDT readings are generally similar to those from strain gauge readings elastic stage where stress increases linearly with strain; (b) transition
before the cracking of concrete, but results from the two sources may segment: when microcracks propagate as loading increases, the hoop
differ substantially afterward due to localized post-cracking deforma- strain of the specimen increases steadily, the FRP jackets begin to work
tion of concrete. Therefore, the axial strains were obtained from read- gradually and the stress-strain curve changes to a smooth curve; (c)
ings of the two LVDTs at the mid-height section in this paper, which, is second ascending/descending segment: as the load grows further the
also suggested by most of previous researchers (e.g., [38]). The hoop hoop dilation of concrete is taken by FRP jackets, thus the slope of this
strains were averaged from the three hoop strain gauges outside the segment mainly depends on the FRP confinement stiffness. The differ-
FRP overlap zone at the mid-height section. In this article, the experi- ence of axial stress-axial strain curves of FRP fully wrapped concrete
mental stress-strain curves are terminated at the point when FRP rup- and FRP partially wrapped concrete is that latter exhibits three seg-
ture occurred, unless otherwise specified. ments while the former exhibits two segments. This difference was
Fig. 7 shows the stress-strain curves of FRP-confined concrete in probably because of the difference in concrete dilation properties and
fully wrapped columns (i.e., Series I). As can be seen in Fig. 7, the axial confinement mechanics between FRP partially and fully wrapped con-
stress-axial strain curves of FRP fully wrapped concrete show typical crete [39]: the FRP partially wrapped concrete exhibits a transition
two segments with the second ascending segment. The monotonic as- segment when the FRP confinement is not completely activated.
cending behavior can be attributed to the sufficient FRP confinement The axial stress-axial strain and axial stress-hoop strain curves of the
owing to the full wrapping. The strain hardening behavior (Type I) of FRP partially wrapped specimens are shown in Figs. 9–11. Fig. 9 shows
concrete in this series is in accordance with the typical experimental the stress-strain curves of concrete in columns confined with three FRP
observations [15], as illustrated in Fig. 3. strips (Series II). In each figure of Fig. 9, the number of FRP strips is
Fig. 8 illustrates the different segments observed on a typical stress- identical, allowing direct comparisons between specimens with dif-
strain curve of FRP partially wrapped concrete, based on the ferent FRP strip thicknesses and widths. It can be seen from Fig. 9 that

816
J.-J. Zeng et al.
(a) Type I: increasing type
(c) Type II-II: decreasing type (
Fig. 8. Typical stress-strain curves for FRP-confined concrete in partially wrapped concrete columns.
all the axial stress-axial strain curves of FRP-confined concrete in the
test columns exhibit typical three segments (Fig. 8): the first linear
ascending segment was followed by a transition segment, which con-
nected the first segment to a nearly linear second branch, which, was
different from that of FRP fully wrapped concrete. In this transition
segment, it was observed that the cracks in concrete between two ad-
jacent strips has developed to some extend but the dilation of concrete
in FRP was not large enough to activate a strain hardening behavior.
It is also seen from Fig. 9 that the specimens in Series II exhibited
strain softening behavior. This strain softening behavior can be attrib-
uted to that the FRP confinement in this series was not sufficient to
achieve strain hardening behavior. Note that some of these curves in
this series display Type II-I behavior and the others display Type II-II
behavior. It is interesting to find that the from Fig. 9 that the slope of
the third segment increase with FRP strip width while it is found that
the slope of the third segment is nearly independent to FRP strip
thickness. This implies that the slope of the third segment pre-
dominantly relates to the clear spacing of FRP strips and it contrasts
with a finding from Zeng et al. [46] that the slope of post-peak segment
increases with the confinement stiffness. The peak stress of concrete
( fcc′ ) generally increased with strip thickness and FRP layers, as can be
seen in Table 1.
Figs. 10 and 11 shows the stress-strain curves of concrete in columns
confined with four and five FRP strips (Series III and IV). Obviously, it is
seen from Figs. 10 and 11 that all the stress-strain curves, except that of
S-1-4-30-1, show monotonic ascending. The strain hardening behavior
was because the confinement efficiency was much better in columns
817
Composite Structures 200 (2018) 810–828
(b) Type II-I: decreasing type ( > )
< )
wrapped with four/five layers of FRP jackets than that columns
wrapped with three layers of FRP jackets. The clear spacing of adjacent
FRP strips reflects the effectiveness of confinement: an increase in the
clear spacing of adjacent FRP strips led to a decrease in the ultimate
axial stress, as is seen from Figs. 9–11. The reason was that confinement
efficiency was higher in a specimen processing a smaller clear spacing
provided that the other column parameters are identical. However, it is
found from Figs. 9–11 that the ultimate axial strain of concrete and
hoop rupture strain of FRP were approximately independent to the clear
spacing of adjacent FRP strips. This implies a behavioral difference
between FRP partially and fully wrapped concrete, as the ultimate axial
strain was found to be highly dependent on the FRP volume ratio for
FRP fully wrapped concrete. It should be noted that the stresses of the
two nominally identical specimens at the transition point of the first
and second segment were expected to be identical (i.e., approximately
equal to fco′ ) for such columns [15]. However, for the identical speci-
mens in each pair of the specimens, stresses at the transition point
should be approximately identical. However, for the stresses at the
transition point of each pair of nominally identical specimens were not
identical for two pairs of specimens (i.e., two S-2-4-25-1/ S-2-4-25-2
and S-2-4-35-1/ S-2-4-35-2). This was probably due to the in-
homogeneous and discreteness of concrete. The details of the ultimate
conditions will be discussed in detail in the following section.
3.3. Ultimate condition
The peak axial stresses ( fcc′ ), axial strains at peak stress (εcc ),
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

30 30

25 25

20

Axial stress (MPa)

20

Axial stress (MPa)

15 15

10 10

S-1-3-25-2 S-1-3-30-1
5 5
S-1-3-25-2 S-1-3-30-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 -15 -10 -5 0 5 10 15 20
-3 -3 -3 -3
Hoop strain 10 Axial strain 10 Hoop strain 10 Axial strain 10

(a) S-1-3-25 (b) S-1-3-30

30 30

25 25

20 20
Axial stress (MPa)

Axial stress (MPa)

15 15

10 10

S-1-3-35-1 S-2-3-25-1
5 S-1-3-35-2
5
S-2-3-25-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 -15 -10 -5 0 5 10 15 20
-3 -3 -3 -3
Hoop strain (10 ) Axial strain 10 Hoop strain 10 Axial strain 10

(c) S-1-3-35 (d) S-2-3-25

35
35

30
30

25
25
Axial stress (MPa)

Axial stress (MPa)

20 20

15 15

10 10
S-2-30-1 S-2-3-35-1
5 S-2-30-2 5 S-2-3-35-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 -15 -10 -5 0 5 10 15 20 25 30
Hoop strain (10-3) Axial strain (10-3) -3
Hoop strain (10 ) Axial strain (10-3)

(e) S-2-3-30 (f) S-2-3-35

Fig. 9. Stress-strain curves of FRP-confined concrete in columns wrapped with three FRP strips (Series II specimens).

ultimate axial stresses ( fcu ′ ), ultimate axial strains (εcu ) and FRP hoop ratio is 0.753, as is marked in Fig. 12. Fig. 12 also shows a variation of
strains at ultimate condition (εh,rup ) (at the mid-height section) for the FRP hoop rupture strain ratio, which was also been reported by some
test columns are listed in Table 1. The normalized ultimate axial other researchers (e.g., [15]). The ultimate condition of FRP fully
′ / fco′ ), normalized ultimate axial strains (εcu/ εco ) are also
stresses ( fcu wrapped concrete generally refers to the ultimate axial stress and the
given in Table 1. In Table 1, εh,rup refers to the average FRP hoop rup- corresponding strain of concrete recorded at the time the FRP jacket
ture strain from the three hoop strain gauges which were outside the ruptures. However, the FRP jacket on partially wrapped concrete does
overlap. For the column specimens failed without FRP rupture, the not necessarily rupture as the concrete crushing failure may take place
ultimate condition is defined as the level the load carrying capacity prior to FRP rupture.
reached to around 80% of the peak load and the FRP hoop strain at As shown in the Table 1, the normalized ultimate axial stresses
ultimate were also included in Table 1. It can be easily found from ′ / fco′ ) and normalized ultimate axial strains (εcu/ εco ) of FRP partially
( fcu
Table 1 that the FRP hoop rupture strains were independent to width of wrapped concrete varied in a considerable range with respect to the
FRP strip and volumetric ratio of FRP. The ratio between FRP hoop column parameters. It is found from Table 1 that the normalized ulti-
rupture strain and FRP coupon rupture strain (FRP hoop rupture strain mate axial stresses ranges from 0.83 to 2.77 and normalized ultimate
ratio) is given in shown in Fig. 12. The average FRP hoop rupture strain axial strains ranges from 1.21 to 19.32, which demonstrating excellent

818
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

35 35

30 30

25 25

Axial stress (MPa)

Axial stress (MPa) 20 20

15 15

10 10
S-3-3-25-1
S-3-3-30-1
5 S-3-3-25-2 5 S-3-3-30-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 25 30 -15 -10 -5 0 5 10 15 20 25 30
-3 -3 -3 -3
Hoop strain (10 ) Axial strain (10 ) Hoop strain 10 Axial strain 10

(g) S-3-3-25 (h) S-3-3-30

35

30

25
Axial stress (MPa)

20

15

10
S-3-35-1
5 S-3-35-2
Teng et al. (2007)
0
-15 -10 -5 0 5 10 15 20 25 30
-3 -3
Hoop strain (10 ) Axial strain (10 )

(i) S-3-3-35
Fig. 9. (continued)

strength and ductility of FRP partially wrapped concrete. The normal- shown in Fig. 14a. It is seen from Fig. 14a that an increase in the width
ized ultimate axial stresses of specimens experienced descending stress- of FPR strips leads to an increase in the normalized ultimate axial
strain curves (for specimens with a larger s′f ) are less than one. This is strain. Note that in Fig. 14, only data from specimens with ascending
believed to be due to the existence of a larger clear FRP strip spacing stress-strain behavior is included. The above conclusion is not applic-
easily leads to localized concrete crushing failure. able to specimens wrapped with three FRP strips, as has been explained
To examine the effects of width of FRP strips, the relationship be- earlier. Fig. 14b shows the relationship between the normalized ulti-
tween the normalized ultimate axial stress and the width of FRP strips is mate axial strain and the thickness of FRP strips. It is clear from Fig. 14b
shown in Fig. 13a, with the normalized axial stress being drawn in the that an increase in the thickness of FPR strips leads to an increase in the
vertical axis and the width of FRP strips drawn in the horizontal axis. normalized ultimate axial strain.
Note that in Fig. 13, only data from specimens with ascending stress-
strain behavior is included. For specimens associated with descending
3.4. Minimum amount of FRP for sufficient confinement in FRP partially
stress-strain behavior, it is obvious the number of FRP width or FRP
wrapped concrete
thickness do not related to the axial stress. As can be seen clearly in
Fig. 13a, the normalized axial stress at ultimate increases with an in-
Generally, FRP-confined concrete is believed to be insufficiently
crease in the width of FRP strips for specimens confined with 4 and 5
confined if confined concrete features a descending stress-strain curve
strips. For specimens confined with three FRP strips, the axial stress is
and the stress at the ultimate strain falls below the compressive strength
not given in Fig. 13 as these specimens experienced a descending stress-
of the unconfined concrete [15]. For FRP-confined concrete in which
strain behavior. Actually, the normalized axial stress seems to be in-
the FRP confinement is insufficient, the FRP is likely to experience
dependent to the width of FPR strips for specimens confined with three
premature failure with a small hoop rupture strain. Consequently, it is
FRP strips. This is because the specimens wrapped with three FRP strips
recommended that such weak confinement should not be allowed in
experienced concrete cracking failure.
practical design. To achieve sufficient FRP confinement, the minimum
It is not difficult to understand that FRP layer is related to FRP
amount of FRP needs to be properly defined. Mirmiran et al. [19]
volumetric ratio (confinement stiffness): increase of FRP layer leads to
suggested that the confinement ratio ( fl / fco′ ) should not be less than
an increase in FRP volumetric ratio, subsequently an increase in con-
0.15, while Spoelstra and Monti [31] believed that the stress of concrete
finement stiffness. Fig. 13b shows the relationship between the nor-
at the ultimate strain falls below fco′ if fl / fco′ < 0.07 . The latter was de-
malized ultimate axial stress and the thickness of FRP strips. As ex- monstrated to be accurate in the definition of weakly confined concrete
pected, the increase in FRP thickness generally leads to increase in both [15,34] while it was found that the actual confinement ratio (i.e.,
ultimate axial stress and ultimate axial strain. The relationship between fla / fco′ ) should be used [15].
the normalized ultimate axial strain and the width of FRP strips is Fig. 15 shows the relationship between the normalized ultimate

819
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

35
35

30
30

25
25

Axial stress (MPa)

Axial stress (MPa) 20
20

15
15

10
10
S-1-4-25-1 S-1-4-30-1
5 S-1-4-25-2 5 S-1-4-30-2
Teng et al. (2007)
Teng et al. (2007)
0
-15 -10 -5 0 5 10 15
0
-15 -10 -5 0 5 10 15
-3 -3 -3 -3
Hoop strain (10 ) Axial strain (10 )
Hoop strain 10 Axial strain 10

(a) S-1-4-25 (b) S-1-4-30

40 40

35 35

30 30

Axial stress (MPa)

25 25
Axial stress (MPa)

20 20

15 15

10 10
S-2-4-25-1
S-1-4-35-1
5 S-1-4-35-2 5 S-2-4-25-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15 20
-3 -3
Hoop strain (10-3) Axial strain (10-3) Hoop strain (10 ) Axial strain (10 )

(c) S-1-4-35 (d) S-2-4-25

40 50

35
40
30
Axial stress (MPa)

25
30
Axial stress (MPa)

20

20
15

10 S-2-4-30-1
10 S-2-4-35-1
S-2-4-30-2 S-2-4-35-2
5
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 -15 -10 -5 0 5 10 15 20 25
-3
Hoop strain (10 )
-3
Axial strain (10 ) -3 -3
Hoop strain (10 ) Axial strain (10 )

(e) S-2-4-30 (f) S-2-4-35

Fig. 10. Stress-strain curves of FRP-confined concrete in columns wrapped with four FRP strips (Series III specimens).

′ / fco′ ) and the actual confinement ratio (i.e., fla / fco′ ).

axial stress (i.e., fcu implies that the equation for minimum amount of FRP is conservative
Note that the actual confinement ratio was determined by Eq. (7) in while most of the specimens with an actual confinement ratio of smaller
which the ultimate FRP hoop strain was adopted. It is seen from Fig. 15 than 0.07 and a normalized ultimate axial stress larger than 1 experi-
that an approximately linear relationship between fcu ′ / fco′ and fla / fco′ is enced a descending behavior of the stress-strain curves. It is thus be-
found, demonstrating a high relevance between the normalized ulti- lieved that the conservativeness suggests little implication to the ac-
mate axial stress and the actual confinement ratio. The fla / fco′ is larger curacy of the criterion for fully wrapped concrete. Consequently, the
than 0.07 for specimens in Series I, III and IV, while the normalized criterion for FRP fully wrapped concrete is also valid for FRP partially
ultimate axial stresses of them are larger than one (i.e., fcu ′ / fco′ > 1). wrapped concrete.
However, for the 18 specimens in Series II, 6 specimens show fcu ′ / fco′ > 1 The range of the clear spacing ratio is given for each series, as
while only one of them has a fla / fco′ ratio of lager than 0.07. It is also shown in Fig. 16. It is seen that most of specimens with fcu ′ / fco′ > 1
seen from Fig. 15 that although the ultimate axial stresses are larger possess a clear spacing ratio of smaller than 0.44, while only 6 speci-
than one for most of the specimens confined with four FRP strips, the mens show fcu ′ / fco′ > 1 for specimens with a clear spacing ratio of larger
actual confinement ratios of 6 specimens are smaller than 0.07. These

820
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

40 45

40
35
35
30
30
25

Axial stress (MPa)

Axial stress (MPa) 25
20
20
15
15

10 10
S-3-4-25-1 S-3-4-30-1

S-3-4-25-2 S-3-4-30-2
5 5
Teng et al. (2007)
Teng et al. (2007)
0
0
-20 -10 0 10 20 30 40
-20 -15 -10 -5 0 5 10 15 20 25 30 35
-3 -3
Hoop strain (10-3) Axial strain (10-3)
Hoop strain (10 ) Axial strain (10 )

(g) S-3-4-25 (h) S-3-4-30

55
50
45
40
35
Axial stress (MPa)

30
25
20
15
10 S-3-4-35-1
S-3-4-35-2
5
Teng et al. (2007)
0
-20 -10 0 10 20 30 40 50
-3 -3
Hoop strain 10 Axial strain 10

(i) S-3-4-35
Fig. 10. (continued)

than 0.65. Fig. 16 shows the relationship between the normalized ul- The capability and accuracy of this model are determined by compar-
′ / fco′ ) and the clear spacing ratio. It is seen
timate axial stress (i.e., fcu isons between test results and predictions, as given in the following
from Fig. 16 that there may exist an optimum value of s′f / D (around section.
0.5), upon which the monotonic ascending stress-strain behavior is not
easily to obtain (i.e., fcu′ / fco′ > 1). Therefore, it is reasonable to re-
commend that a minimum ratio between the clear spacing of FRP strips 4.2. Teng et al. [33] model
and the diameter of the column (i.e., s′f / D ) for FRP partially wrapped
concrete is 0.5. The Teng et al. [33] model for FRP-confined concrete is composed
of following three elements: (1) an active-confinement base model; (2)
a hoop strain-axial strain relationship equation; (3) a relationship be-
4. Stress-Strain models tween the hoop strain in FPR jacket and the confining pressure supplied
by the jacket. The path-independence assumption was adopted in Teng
4.1. General et al. [33] model, suggesting that the axial stress and the axial strain of
FRP-confined concrete at a given hoop strain are equal to those of the
Extensive research has been conducted on FRP-confined concrete same concrete confined with a constant confining pressure (referred to
through axial compression tests on circular plain concrete cylinders as actively confined concrete hereafter) equal to that supplied by the
confined with an FRP jacket, leading to a number of stress-strain FRP.
models. These models include design-oriented models in closed-form The active-confinement model adopted in Teng et al. [33] model is
expressions (e.g., [34,23,22]) and analysis-oriented models (e.g., described by the following equations:
[33,21]) which predicts stress-strain curves using an incremental pro-
cedure [13]. Note that an analysis-oriented stress-strain is generally σc (εc / εcc∗ ) r
∗ =
more versatile and accurate than a design-oriented stress-strain model, fcc′ r −1 + (εc / εcc∗ )r (1)
which has been demonstrated by some researchers [13]. Teng et al.
[33] model has been shown to provide close predictions of FRP-con-
Ec
fined concrete (e.g., [14,17]), in terms of both the axial stress-strain r=
Ec−fcc′ ∗ / εcc∗ (2)
curve and the axial strain-hoop strain curve. Note that the Teng et al.
[33] model was developed and verified based on the test results of
concrete fully wrapped with an FRP jacket in the form of wet layup. fcc′ ∗ σl
Consequently, the model is not necessarily capable and accurate in = 1 + 3.5
fco′ fco′ (3)
generating stress-strain behavior of FRP partially wrapped concrete.

821
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

40
40

35 35

30 30

Axial stress (MPs)

25 25

Axial stress (MPa)

20 20

15 15

10 10
S-1-5-25-1 S-1-5-30-1
5 S-1-5-25-2 5 S-1-5-30-2
Teng et al. (2007) Teng et al. (2007)
0 0
-15 -10 -5 0 5 10 15 20 25 -15 -10 -5 0 5 10 15 20 25
-3 -3 -3 -3
Hoop strain (10 ) Axial strain (10 ) Hoop strain (10 ) Axial strain (10 )

(a) S-1-5-25 (b) S-1-5-30

50 50

40 40
Axial stress (MPa)

30

Axial stress (MPa)

30

20 20

10 S-1-5-35-1
10 S-2-5-25-1
S-1-5-35-2 S-2-5-25-2
Teng et al. (2007) Teng et al. (2007)
0 0
-20 -15 -10 -5 0 5 10 15 20 25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40
Hoop strain (10-3) Axial strain (10-3) -3
Hoop strain (10 )
-3
Axial strain (10 )

(c) S-1-5-35 (d) S-2-5-25

50 50

40 40
Axial stress (MPa)

Axial stress (MPa)

30 30

20 20

S-2-5-30-1 10 S-2-5-35-1
10
S-2-5-30-2 S-2-5-35-2
Teng et al. (2007) Teng et al. (2007)
0
0
-20 -15 -10 -5 0 5 10 15 20 25 30 35 40
-20 -15 -10 -5 0 5 10 15 20 25 30 35 40 -3 -3
Hoop strain 10-3 Axial strain 10-3 Hoop strain (10 ) Axial strain (10 )

(e) S-2-5-30 (f) S-2-5-35

Fig. 11. Stress-strain curves of FRP-confined concrete in columns wrapped with five FRP strips (Series IV specimens).

1.2
εcc∗ σ The following equation is used in Teng et al. [33] model to calculate
= 1 + 17.5 ⎛⎜ l ⎟⎞ the confining pressure provided by the FRP tube/jacket:
εco f ′ (4)
⎝ co ⎠
where σl is the confining pressure, and are respectively the peak fcc′ ∗ εcc′ ∗ Ef t f εh
σl =
axial stress and the corresponding axial strain of concrete under a R (6)
specific constant confining pressure σl .
The following axial strain-lateral strain relationship, originally where Ef and t f are the hoop elastic modulus and thickness of FRP
proposed by Teng et al. [33], was adopted: jacket; R is the radius of confined concrete core; εh is the hoop strain in
FRP jacket.
0.7
εc σ ⎧ ε ε ⎫ To the best of the authors’ knowledge, only the confined concrete
= 0.85 ⎜⎛1 + 8 l ⎟⎞ ⎡1 + 0.75 ⎛ h ⎞ ⎤ −exp ⎡−7 ⎛ h ⎞ ⎤
models provided by CNR-DT 200 R1 [7] and fib [8] are available for
⎜ ⎟ ⎜ ⎟

εco f ′co ⎨ ⎢ ε
⎝ ⎠⎦
co
⎥ ⎢
⎣ ⎝ ⎠⎦⎬
ε co
⎥
⎝ ⎠⎩⎣ ⎭ (5)
predicting the stress-strain response of FRP partially wrapped concrete.
where εh is the hoop strain. In both codes, a vertical efficiency coefficient which was originally

822
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

60 70

50 60

50
40
Axial stress (MPa)

Axial stress (MPa)

40
30
30

20
20
S-3-5-25-1 S-3-5-30-1
10 10 S-3-5-30-2
S-3-5-25-2
Teng et al. (2007) Teng et al. (2007)
0 0
-20 -10 0 10 20 30 40 50 60 -20 -10 0 10 20 30 40 50
-3 -3 -3 -3
Hoop strain (10 ) Axial strain (10 ) Hoop strain (10 ) Axial strain (10

(g) S-3-5-25 (h) S-3-5-30

70

60

50
Axial stress (MPa)

40

30

20
S-3-5-35-1
10 S-3-5-35-2
Teng et al. (2007)
0
-20 -10 0 10 20 30 40 50
-3 -3
Hoop stain (10 ) Axial strain (10 )

(i) S-3-5-35
Fig. 11. (continued)

where ρf is the FRP jacket volumetric ratio. The vertical confinement

effectiveness coefficient can be calculated by [8]:
s′f 2

k =
v
(1− ) 2D
≈ ⎛⎜1−
s′f ⎞2
⎟
1−ρsg ⎝ 2D ⎠ (8)

where ρsg is the reinforcement ratio of the longitudinal steel re-

inforcement with respect to the gross cross-sectional area. It should be
mentioned that the usage of vertical confinement effectiveness coeffi-
cient in the analysis-oriented model implies that it is assumed that the
actual equivalent confinement is uniform and the actual equivalent
confinement is exerted to the whole section.

4.3. Models for ultimate axial stress and ultimate axial strain

In addition to Teng et al. [33] model, the applicability of four re-

presentative models [42,22,28,8] is assessed herein using the test re-
sults of the present study. These four models for ultimate axial stress
and ultimate axial strain were selected as they are either from current
Fig. 12. FRP hoop rupture strain ratio.
design code or from recent literatures. The detailed equations for the
four models are not listed in current paper to limit the length of the
proposed by Sheikh and Uzumeri [30] is suggested for the FRP effective paper, whereas reader may refer to literatures [42,22,28,8] for details.
confinement pressure in the FRP partially wrapped columns, con- The actual confining stress for FRP partially wrapped concrete (Eq. (7))
sidering that the hoop confining pressure is only effective for concrete was adopted in all predictions. The test results presented in this study
in the effective confinement area where the confining pressure has fully were not used in the development of any of the above five stress-strain
developed due to the arching action (Fig. 2). Therefore, the effective models; as a result, the test results provide an independent assessment
confining stress in FRP partially wrapped concrete is: of their performance. The comparisons between the test results and the
predictions from these four models contains only the ultimate axial
ρf Ef εh stress and strain but not the stress-strain curves for concrete partially
σl,e = k v σl = k v
2 (7) wrapped with FRP strips as only Teng et al. [33] model is capable of

823
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

70 6

60 5
Ultimate axial stress (MPa)

Ultimate axial strain (%)

4
50

3
40
S-1-4-X S-1-4-X
S-2-4-X 2 S-2-4-X
S-3-4-X S-3-4-X
30
S-1-5-X S-1-5-X
1 S-2-5-X
S-2-5-X
S-3-5-X
S-3-5-X
20 25 30 35
25 30 35
Width of FRP strip (mm) Width of FRP strip (mm)
(a) Relationship between ultimate axial stress and width of FRP strips (a) Relationship between ultimate axial strain and width of FRP strips
70
6.0
65
5.5
Ultimate axial stress (MPa)

60
5.0

Ultimate axial strain (%)

55 4.5
50 4.0
45 3.5
40 3.0
S-X-4-25 S-X-4-25
35 2.5
S-X-4-30 S-X-4-30
30 S-X-4-35 2.0 S-X-4-35
S-X-5-25 S-X-5-25
25 1.5
S-X-5-30 S-X-5-30
S-X-5-35 1.0 S-X-5-35
20
1 2 3 1 2 3
Number of FRP layer
Number of FRP strip layer
(b) Relationship between ultimate axial stress and number of FRP layer
(b) Relationship between ultimate axial strain and number of FRP layer
Fig. 13. Relationship between ultimate axial stress and width of FRP strips/
number of FRP layer. Fig. 14. Relationship between ultimate axial strain and width of FRP strips/
number of FRP layer.

generating both axial stress-axial strain curves and axial stress-hoop

strain curves.

4.4. Comparisons and discussions

Test results from the present study are compared with the predic-
tions of Teng et al. [33] analysis-oriented stress-strain model. The
comparisons between test results and theoretical predictions are shown
in Figs. 7 and 9–11. The material properties presented earlier in this
paper were used and the predicted curves were terminated at a hoop
strain equal to the average FRP ultimate strain of the two nominally
identical specimens of a pair. Fig. 7 shows the predicted and test stress-
strain curves for Series I specimens. It can be seen from Fig. 7 that the
predicted axial stress-strain curves are slightly lower than the test re-
sults, but these small errors have little implication for the comparison.
Consequently, it is no doubt that the Teng et al. [33] model provides
close predictions for FRP-confined concrete in fully wrapped concrete
columns.
Figs. 9–11 shows the predicted stress-strain curves for specimens in
Series II, III and IV and the test results. The following three features can
be observed from Figs. 9–11: (1) the predicted axial strain is generally
much smaller than the corresponding test results; (2) the predicted axial
stress-axial strain curves generally show a two-segment behavior while Fig. 15. Relationship between normalized ultimate axial stress and actual
experimental axial stress-axial strain curves exhibit a three-segment confinement ratio.
behavior and the stress at the first and second segment transition point
is generally underestimated by the Teng et al. [33] model and (3) the
predicted axial stress is generally close to the experimental axial stress
when the hoop strain is relatively large, but the former becomes

824
J.-J. Zeng et al.
Fig. 16. Relationship between normalized ultimate axial stress and clear spa-
cing ratio.
considerably smaller than the latter when the hoop strain is small. The
first observation is believed to be due to the use of LVDTs covering a
gauge area containing both FRP strips and the concrete between two
adjacent strips. The axial strain in the concrete between two adjacent
strips was believed to be larger than that in the FRP strips for a partially
wrapped concrete column [46], although further research is needed to
clarify this concern. As a result, the ultimate axial strain in FRP partially
wrapped concrete is believed to be larger than that in FRP fully
wrapped concrete. Consequently, the axial strain-hoop strain relation-
ship in the analysis-oriented which is based on FRP fully wrapped
concrete needs to be refined for a more accuracy predicted ultimate
axial strain in FRP partially wrapped concrete while this issue needs to
be further verified by experimental and theoretical studies.
The second observation is believed to be due to the difference of
confinement mechanics for FRP partially wrapped concrete and FRP
fully wrapped concrete, as has explained earlier. The error of under-
estimation of axial stress at the first and second segment transition
point was resulted from the error in obtaining the unconfined concrete
strength, which was believed to be due to the intrinsic discreteness and
inhomogeneity of concrete. The third observation is because FRP par-
tially wrapped concrete generally exhibits a three-segment stress-strain
behavior while the predicted stress-strain curves show a typical two
segment behavior. This results from an error caused by the usage of the
vertical confinement effectiveness coefficient as the slope of the third
segment of stress-strain curve of FRP partially wrapped concrete may
depend on both the FRP material properties and the clear spacing of
FRP strips, instead of the FRP material properties only.
For Series II specimens (i.e., columns wrapped with 3 FRP strips),
the predicted stress-strain curves are less accurate than those of Series
III and IV specimens. One possible reason is that the confinement me-
chanics for FRP partially wrapped concrete is different from that for
FRP fully wrapped concrete, especially for FRP partially wrapped
concrete with insufficient confinement. Nevertheless, the above error is
not related to the vertical confinement effectiveness coefficient, im-
plying that the vertical confinement effectiveness coefficient provided
by current codes [8,7] is reasonably accurate.
Obviously, it is interesting to find that the predictions are generally
satisfactory for all batches of specimens in term of ultimate axial stress
while the Teng et al. [33] model is less accurate in predicting ultimate
axial strain for FRP partially wrapped concrete. Further research on
refinement of the analysis-oriented model for FRP partially wrapped
825
Composite Structures 200 (2018) 810–828
concrete is necessary.
The comparisons of ultimate axila stress and ultimate axial strain for
all the models are shown in Fig. 17. Note that the vertical confinement
effectiveness coefficient is adopted in calculating the effective confining
stress for all the selected models. The average absolute error (AAE), and
standard deviation (SD) of ratios between the theoretical and the ex-
perimental values (theoi / expi ) for each model are also given in the
figure. The AAE and SD are defined by the following equations:
n expi − theoi
∑i = 1 expi
AAE =
n (9)
2
SD =
n theo
∑i = 1 ⎡ exp i − exp
⎣ i
( )
theo
⎤
aver ⎦
n−1 (10)
where n is the number of data points; expi and theoi are the ith experi-
mental value and theoretical value, respectively. The AAE measures the
average error magnitude of model prediction; and the SD shows the
degree of variation or the magnitude of the associated scatter of pre-
diction.
It can be seen from Fig. 17 that for ultimate axial stress, Teng et al.
[33] model provides most accurate predictions as the AAE value of it is
the smallest; while the SD of Ozbakkaloglu and Lim [22] model is
smaller than that of Teng et al. [33] model, the AEE value of it is larger.
As for ultimate axial strain, the AAE values of Teng et al. [33] model
and Wei and Wu [42] model are much smaller than those of the other
three models, implying better performance of the two models than the
other three models in predicting ultimate axial strains. The above
comparisons suggest that, obviously, Teng et al. [33] model is superior
than the other four representative models.
5. Conclusions
The test results of 60 circular columns fully or partially wrapped
with FRP strips are presented in this paper. A widely accepted analysis-
oriented model proposed by Teng et al. [33] is used to predict the
stress-strain behavior of FRP partially wrapped concrete, and the re-
liability and accuracy of five existing models are then examined
through the comparisons between the test results and the model pre-
dictions. Based on the study presented in this paper, the following
conclusions can be drawn:
(1) Significant increases in the strength and axial deformation capacity
are obtained for FRP partially wrapped concrete. Similar to FRP
fully wrapped columns, FRP partially wrapped columns are mostly
failed due to the rupture of FRP strips. However, an increase in the
clear FRP strip spacing may lead to the crushing failure of the
concrete between two adjacent strips.
(2) A three-segment stress-strain curve can predict the behavior of FRP
partially wrapped concrete better compared with a two-segment
stress-strain curve which normally captures the behavior of FRP
fully wrapped concrete well. This three-segment stress-strain curve
is consisted of a first linear ascending segment, a transition segment
and a second ascending/descending segment.
(3) The stress-strain curves of FRP-confined concrete in fully wrapped
columns show monotonically ascending behavior owing to the
sufficient FRP confinement. For FRP partially wrapped concrete,
the slope of the third segment increases with an increase in the
width of FRP strips while it is nearly independent to the FRP strip
thickness. The axial strength increases with the width of FRP strips
for specimens confined with four or five FRP strips while it seems to
be independent to the width of FPR strips for specimens confined
with only three FRP strips. The ultimate axial strain of concrete
increases with the width of FRP strips and the FRP thickness. An
increase in the clear spacing of adjacent FRP strips leads to a
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

(a) fib (2001)

(b) Teng et al. (2007)

(c) Wei and Wu (2012)

Fig. 17. Performance of existing models in predicting the ultimate condition of test specimens.

decrease in the axial strength while the ultimate axial strain of
concrete and the hoop rupture strain of FRP are approximately
independent to the clear spacing of adjacent FRP strips.
(4) The equation for calculating the minimum amount of FRP is con-
servative. The s′f / d value is suggested to be larger than 0.5 to obtain
a monotonic ascending stress-strain behavior (i.e., fcu ′ / fco′ > 1).
(5) Teng et al. [33] model is reasonably accurate for predicting the
axial strength of FRP partially wrapped concrete, providing that the
confinement effectiveness coefficients [8,7] have properly been
considered. However, the predicted ultimate strain is significantly
underestimated and the stress at the turning point is also under-
estimated by Teng et al. [33] model. The comparisons between test
results and the predictions from five representative models indicate
that Teng et al. [33] model is superior to the other four models in
predicting ultimate axial stress and ultimate axial strain of concrete
partially confined with FRP strips.
Acknowledgements
The authors gratefully acknowledge the financial support provided
by the National Key R&D Program of China (No. 2017YFC0806001),
the National Natural Science Foundation of China (No. 11372076, No.
11672076 and No. 11472084), the Science and Technology Planning
Project of Guangzhou (No. 201704030057 and No. 201707010364),
the Education Department of Guangdong Province (No.
2017KZDXM030) and the Science and Technology Planning Project of
Guangdong Province (No. 2017A010103030, No. 2017B020238006
and No. 2014A020216053).

826
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

(d) Ozbakkaloglu and Lim (2013)

(e) Pham et al. (2015)

Fig. 17. (continued)

Appendix A. Supplementary data [13] Jiang T, Teng JG. Analysis-oriented stress-strain models for FRP-confined concrete.
Eng Struct 2007;29(11):2968–86.
[14] Liang M, Wu ZM, Ueda T, Zheng JJ, Akogbe R. Experimental and modeling on axial
Supplementary data associated with this article can be found, in the behavior of carbon fiber reinforced polymer confined concrete cylinders with dif-
online version, at http://dx.doi.org/10.1016/j.compstruct.2018.05. ferent sizes. J Reinf Plast Compos 2012;31(6):389–403.
001. [15] Lam L, Teng JG. Design-oriented stress-strain model for FRP-confined concrete.
Constr Build Mater 2003;17(6–7):471–89.
[16] G. Lin, T. Yu, J.G. Teng. Design-oriented stress-strain model for concrete under
References combined FRP-steel confinement. J Compos Constr, ASCE, 10.1061/(ASCE)CC.
1943-5614.0000651, p. 04015084.
[17] Lin G, Teng JG. Three-dimensional finite-element analysis of FRP-confined circular
[1] ASTM D3039. Standard test method for tensile properties of polymer matrix com-
concrete columns under eccentric loading. J Compos Constr, ASCE
posite materials (D3039M). West Conshohocken, USA; 2008.
2017;21(4):04017003.
[2] ASTM C469. Standard test method for static modulus of elasticity and poisson ratio
[18] Mai AD, Sheikh MN, Hadi MNS. Investigation on the behaviour of partial wrapping
of concrete in compression. West Conshohocken, USA; 2002.
in comparison with full wrapping of square RC columns under different loading
[3] Bai YL, Dai JG, Ozbakkaloglu T. Cyclic stress-strain model incorporating buckling
conditions. Constr Build Mater 2018;168:153–68.
effect for steel reinforcing bars embedded in FRP-confined concrete. Compos Struct
[19] Mander JB, Priestley MJ, Park R. Theoretical stress-strain model for confined
2017;182:185–98.
concrete. J Struct Eng, ASCE 1988;114(8):1804–26.
[4] Barros JAO, Ferreira DRSM. Assessing the efficiency of CFRP discrete confinement
[20] Matthys S, Toutanji H, Audenaert K, Taerwe L. Axial behavior of large-scale col-
systems for concrete cylinders. J Compos Constr, ASCE 2008;12(2):134–48.
umns confined with fiber-reinforced polymer composites. ACI Struct J
[5] Campione G, La Mendola L, Monaco A, Valenza A, Fiore V. Behavior in compression
2005;102(2):258–67.
of concrete cylinders externally wrapped with basalt fibers. Compos B Eng
[21] Nisticò N. R.C. square sections confined by FRP: a numerical procedure for pre-
2015;69:576–86.
dicting stress-strain relationships. Compos B Eng 2014;59:238–47.
[6] Chastre C, Silva MAG. Monotonic axial behaviour and modelling of RC circular
[22] Ozbakkaloglu T, Lim JC. Axial compressive behavior of FRP-confined concrete:
columns confined with CFRP. Eng Struct 2010;32(8):2268–77.
experimental test database and a new design-oriented model. Compos B Eng
[7] CNR-DT 200 R1. Guide for the design and construction of externally bonded frp
2013;55:607–34.
systems for strengthening existing structures, advisory committee on technical re-
[23] Ozbakkaloglu T, Lim JC, Vincent T. FRP-confined concrete in circular sections:
commendations for construction. National Research Council, Rome, Italy; 2013.
review and assessment of stress-strain models. Eng Struct 2013;49:1068–88.
[8] fib. Externally bonded frp reinforcement for RC structures, the international fed-
[24] Ouyang LJ, Gao WY, Zhen B, Lu ZD. Seismic retrofit of square reinforced concrete
eration for structural concrete. Lausanne, Switzerland; 2001.
columns using basalt and carbon fiber-reinforced polymer sheets: a comparative
[9] Eid R, Patrick P. Compressive behavior of FRP-confined reinforced concrete col-
study. Compos Struct 2017;162:294–307.
umns. Eng Struct 2017;132:518–30.
[25] Park TW, Na UJ, Chung L, Feng MQ. Compressive behavior of concrete cylinders
[10] Guo YC, Xie JH, Xie ZH, Zhong J. Experimental study on compressive behavior of
confined by narrow strips of CFRP with spacing. Compos B Eng
damaged normal- and high-strength concrete confined with CFRP laminates. Constr
2008;39(7–8):1093–103.
Build Mater 2016;107:411–25.
[26] Pessiki S, Harries KA, Kestner J, Sause R, Ricles JM. The axial behavior of concrete
[11] Hadi MNS. Behaviour of wrapped normal strength concrete columns under ec-
confined with fiber reinforced composite jackets. J Compos Constr, ASCE
centric loading. Compos Struct 2006;72(4):503–11.
2001;5(4):237–45.
[12] Ilki A, Peker O, Karamuk E, Demir C, Kumbasar N. FRP Retrofit of low and medium
[27] Pham TM, Doan LV, Hadi MN. Strengthening square reinforced concrete columns
strength circular and rectangular reinforced concrete columns. J Mater Civ Eng
by circularisation and FRP confinement. Constr Build Mater 2013;49:490–9.
ASCE 2008;20(2):169–88.
[28] Pham TM, Hadi MNS, Youssef J. Optimized FRP wrapping schemes for circular

827
J.-J. Zeng et al. Composite Structures 200 (2018) 810–828

concrete columns under axial compression. J Compos Constr, ASCE
2015;19(6):04015015.
[29] Saljoughian A, Mostofinejad D. Axial-flexural interaction in square RC columns
confined by intermittent CFRP wraps. Compos B Eng 2016;89:85–95.
[30] Sheikh BP, Uzumeri SM. Strength and ductility of tied concrete columns. J Struct
Div, ASCE 1980;106(5):1079–102.
[31] Spoelstra MR, Monti G. FRP-confined concrete model. J Compos Constr, ASCE
1999;3(3):143–50.
[32] Teng JG, Chen JF, Smith ST, Lam L. FRP strengthened RC structures. John Wiley &
Sons Ltd.; 2002.
[33] Teng JG, Huang YL, Lam L, Ye LP. Theoretical model for fiber-reinforced polymer-
confined concrete. J Compos Constr, ASCE 2007;11(2):201–10.
[34] Teng JG, Jiang T, Lam L, Luo YZ. Refinement of a design-oriented stress-strain
model for FRP-confined concrete. J Compos Constr, ASCE 2009;13(4):269–78.
[35] Teng JG, Lam L, Lin G, Lu JY, Xiao QG. Numerical Simulation of FRP-Jacketed RC
Columns Subjected to Cyclic and Seismic Loading. J Compos Constr, ASCE
2015;20(1):04015021.
[36] Teng JG, Wu JY, Casalboni S, Xiao QG, Zhao Y. Behavior and modeling of fiber-
reinforced polymer-confined concrete in elliptical columns. Adv Struct Eng
2016;19(9):1359–78.
[37] Triantafyllou GG, Rousakis TC, Karabinis AI. Axially loaded reinforced concrete
columns with a square section partially confined by light GFRP straps. J Compos
Constr, ASCE 2015;19(1):1–15. 04014035.
[38] Vincent T, Ozbakkaloglu T. Influence of fiber orientation and specimen end con-
dition on axial compressive behavior of FRP-confined concrete. Constr Build Mater
2013;47:814–26.
[39] Vincent T, Ozbakkaloglu T. Influence of shrinkage on compressive behavior of
concrete-filled FRP tubes: an experimental study on interface gap effect. Constr
Build Mater 2015;75:144–56.
[40] Wang LM, Wu YF. Effect of corner radius on the performance of CFRP-confined
square concrete columns: test. Eng Struct 2008;30(2):493–505.
[41] Wang ZY, Wang DY, Smith S, Lu DG. CFRP-confined square RC columns. I: ex-
perimental investigation. J Compos Constr, ASCE 2012;16(2):150–60.
[42] Wei YY, Wu YF. Unified stress-strain model of concrete for FRP-confined columns.
Constr Build Mater 2012;26(1):381–92.
[43] Wei H, Wu Z, Guo X, Yi F. Experimental study on partially deteriorated strength
concrete columns confined with CFRP. Eng Struct 2009;31(10):2495–505.
[44] Wu G, Wu ZS, Lu ZG. Design-oriented stress–strain model for concrete prisms
confined with FRP composites. Constr Build Mater 2007;21(5):1107–21.
[45] Xu JJ, Chen ZP, Xiao Y, Demartino C, Huang JH. Recycled aggregate concrete in
FRP-confined columns: a review of experimental results. Compos Struct
2017;174:277–91.
[46] Zeng JJ, Guo YC, Gao WY, Li JZ, Xie JH. Behavior of partially and fully FRP-con-
fined circularized square columns under axial compression. Constr Build Mater
2017;152:319–32.
[47] Zeng JJ, Guo YC, Li LJ, Chen WP. Behavior and three-dimensional finite element
modeling of circular concrete columns partially wrapped with FRP strips. Polymers
2018;10(3):253.
[48] Zhao JL, Teng JG, Yu T, Li LJ. Behavior of large-scale hybrid FRP-concrete-steel
double-skin tubular beams with shear connectors. J Compos Constr, ASCE
2016;20(5):04016015.

828