Engineering Structures 101 (2015) 465–476

Contents lists available at ScienceDirect

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

Displacement and plastic hinge length of FRP-confined circular

reinforced concrete columns
Osama Youssf a, Mohamed A. ElGawady b,⇑, Julie E. Mills a
a
University of South Australia, Adelaide, Australia
b
Missouri University of Science and Technology, MO, USA

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

Article history: Confinement of both existing and newly constructed reinforced concrete (RC) columns by fibre reinforced
Received 22 January 2015 polymer (FRP) has been commonly used in recent decades. This is because of its ability to enhance the
Revised 23 June 2015 shear resistance and the ductility of the RC columns, which are the main parameters that govern the
Accepted 16 July 2015
behaviour of RC columns under lateral loading. This paper presents a finite element (FE) model that
was developed using the LS-DYNA program aimed at modelling the plastic hinge length (lp) for
FRP-confined RC columns. A FE parametric study was conducted to investigate the effect of
Keywords:
FRP-confinement on lp. Empirical models were proposed to predict lp and the ultimate drift ratio (du)
FRP-confined concrete
RC columns
for FRP-confined RC columns and the results were compared with similar previous models. The proposed
Plastic hinge length FE model was able to predict the plastic hinge region and lp value which can provide a simple way for
Ultimate drift ratio designers to investigate the behaviour of FRP-confined columns during the design process. The proposed
Finite element analysis du model reduced the average of errors (A) and standard deviation (SD) by 15.1%, 3.9%, respectively, com-
LS-DYNA pared to the best predictions by previous models.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction concrete columns subject to seismic loads is an important design

Reinforced concrete (RC) columns subject to seismic loading are
critical structural members and many of these columns have been
severely damaged or collapsed during moderate to large earth-
quakes due to inadequate strength, confinement, or ductility
[1–4]. By enhancing the ductility of a structure, the seismic
demand may decrease, leading to a more economical design, and
the displacement capacity may increase, leading to an improved
building performance [5,6].
In recent years, external confinement of concrete columns by
fibre reinforced polymer (FRP) has become increasingly popular
in the construction industry [7–12]. FRP-confinement increases
the column’s shear resistance and ductility because of its high ten-
sile stiffness and strength [13,14]. FRP-confinement prevents con-
crete cover from spalling and increases the inelastic deformability
of concrete in the potential plastic hinge region, which can increase
the lateral displacement capacity of the column under seismic
loads [15–19]. FRP shells provide stay-in-place formwork for new
structures, and a protective shell against corrosion, weathering
and chemical attacks [15,18,20]. Therefore, using FRP to confine
⇑ Corresponding author.
E-mail address: elgawadym@mst.edu (M.A. ElGawady).
or rehabilitation option to consider.
The parameters that can affect the behaviour of FRP-confined
columns are the column axial load ratio, the flexural reinforcement
ratio, cross section shape, cross section aspect ratio in the case of
rectangular sections, the column effective moment-to-shear span,
the mechanical characteristics of the used FRP fabric namely, the
FRP thickness, ultimate tensile strength, ultimate strain, and
E-modulus [5,21,22]. Increasing the thickness and/or strength of
the FRP increases its stiffness and then delays the FRP rupture
[15,23]. This delays the reinforcement buckling; hence, the column
ultimate displacement and peak strength increase. However, there
is threshold for confinement beyond which any increase in con-
finement does not increase the strength or ductility of a column
[22]. Furthermore, increasing the cross sectional aspect ratio for
columns having rectangular columns reduce the effectiveness of
the confinement. Hence, for a given confinement ratio, columns
having larger cross section aspect ratio develop smaller ultimate
drift compared to columns having smaller cross section aspect
ratio [22].
The plastic hinge region in a member is defined as the physical
region over which the member experiences inelastic deformations
and severe damage [10]. The performance of a plastic hinge is crit-
ical to the deformation capacities of flexural members and hence
requires extensive detailing to prevent failure of structural

http://dx.doi.org/10.1016/j.engstruct.2015.07.026
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
466 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
members from extreme events such as earthquakes. Identification
of the plastic hinge length (lp) is a key step in estimating the ulti-
mate drift ratio ðdu Þ of concrete columns. The ultimate drift ratio
of a column is defined here as the ratio of maximum lateral dis-
placement experienced by the column to its shear span. Priestley
et al. [24] and Elsanadedy and Haroun [25] showed that the plastic
hinge length of an FRP-confined RC column is smaller than that of a
traditional RC column. However, other experimental tests have
also shown that the lp of FRP-confined columns are larger than
those of traditional RC columns [15]. Other researchers reported
that the lp of FRP-confined columns is equal to that of a traditional
RC columns [26,27]. Gu et al. 2010 [16] have observed an increase
and then decrease in the lp depending on the columns confinement
ratio. They have related the increase in the lp to the increase of the
cross-section moment capacity caused by the effect of the
FRP-confinement. This results in significant increase in the plastic-
ity zone and hence the lp increases. However, they related the
decrease in the lp with the confinement ratio increase to the addi-
tional frictional bond between concrete and longitudinal bars
caused by the confinement pressure that able to decrease the rein-
forcement strains along the column height and hence the lp
decreases [16]. Hence, there is no consensus among researchers
on the quantification of lp for FRP-confined columns. Therefore,
quantification of the plastic hinge zone is important, not only for
the design of new structures but also for the rehabilitation of old
structures [28].
Several analytical models (e.g. [21,29–31]) have been devel-
oped for estimating lp for unconfined columns. These models
resulted in a wide range of lp values ranging from 0.4 to 2.4 of
the column’s diameter. However, limited studies were conducted
to determine lp for FRP-confined columns [31]. Priestly and Park
[32] have proposed an analytical model for lp estimation based
on experimental observations as shown in Eq. (1). Paulay and
Priestly [29] have improved Eq. (1) based on curvature integra-
tion of typical member to account for different grades of longitu-
dinal reinforcement as shown in Eq. (2). Lu et al. [33] have
modified Eq. (1) to express the length of the plastic hinge based
on a regression analysis using relevant experimental results as
shown in Eq. (3).
lp ¼ 0:08H þ 6db ð1Þ
lp ¼ 0:08H þ 0:022f sy db ð2Þ
lp ¼ 0:077H þ 8:16db ð3Þ
where lp is the plastic hinge length, H is the column shear span, db is
the longitudinal rebar diameter, and f sy is the longitudinal rebar
yield strength.
The first term in these equations (Eqs. (1)–(3)) accounts for col-
umn bending effect, while the second term accounts for steel ten-
sile strain continuing into the footing due to the finite bond stress
(tensile strain penetration effect), with consideration of different
reinforcement grades incorporated in Eq. (2).
Bae and Bayrak [21] have conducted analytical parametric stud-
ies on the influence on lp of various parameters using four full-scale
square reinforced columns. The effect of axial load ratio, reinforce-
ment ratio, and shear span to column depth ratio were investigated
through moment–curvature analysis and the axial strain profile of
the longitudinal compression bars. However, the strain penetration
effect on the lp was kept constant as it did not change with columns
parameters, except for the flexural rebar diameter. Their proposed
model is shown in Eq. (4).
     
P As
lp ¼ 0:3 þ3  0:1 H þ 0:25D ð4Þ
P0 Ag
where lp is the plastic hinge length, P is the column axial load, P0 is
the column axial load capacity, As is the total area of longitudinal
reinforcement, Ag is the gross area of column cross section, H is
the column shear span, and D is the column diameter.
Mortezaei and Ronagh [5] have proposed modifications to that
proposed by Bae and Bayrak [21] (Eq. (4)) based on finite element
(FE) parametric studies on FRP strengthened reinforced columns
subjected to far-fault and near-fault ground motions. The parame-
ters that they investigated were the same parameters investigated
by Bae and Bayrak [21]. Eqs. (5) and (6) show their proposed
models.
     
P As
lp ¼ 0:4 þ3  0:1 H
P0 Ag
þ 0:65D ðFor near-fault ground motionÞ ð5Þ
     
P As
lp ¼ 0:4 þ3  0:1 H
P0 Ag
þ 0:55D ðFor far-fault ground motionÞ ð6Þ
where lp is the plastic hinge length, P is the column axial load, P0 is
the column axial load capacity, As is the total area of longitudinal
reinforcement, Ag is the gross area of column cross section, H is
the column shear span, and D is the column diameter.
Gu et al. [16] have conducted an experimental investigation on
FRP-confined reinforced concrete columns subjected to seismic
load using different types and thicknesses of FRP. Based on the
results of this experimental investigation, Gu et al. [31], have mod-
elled lp of FRP-confined columns by taking into consideration the
effect of confinement ratio ðkl Þ as shown in Eq. (7).
lp ¼ ð0:59  2:30kl þ 2:28k2l ÞH þ 0:022f sy db ð7Þ
where lp is the plastic hinge length, kl is the confinement ratio, H is
the column shear span, f sy is the longitudinal rebar yield strength,
and db is the longitudinal rebar diameter.
Measurements of lp in previous experimental studies (e.g.,
[21,34,35]) were based on visual observations of the
column-damage regions. Ozbakkaloglu and Saatcioglu [15,18] have
introduced a physical technique to measure lp for FRP-confined col-
umns using the recorded FRP lateral strains during the cyclic load-
ing. This technique is based on the intimate relationship that exists
between the lateral expansion of the FRP tube and the level of
damage sustained by concrete inside the tube [10]. Higher
FRP-tube hoop strains correspond to the most highly damaged
regions of the columns, since the concrete undergoes rapid expan-
sion inside the FRP tube within these regions. The technique
assumes that, at the ultimate column displacement, the plastic
hinge region terminates at a height above the column footing
where the recorded hoop strain values are below 1/3 of the maxi-
mum recorded strain. This technique was verified through mea-
surements of column rotations and strains on longitudinal
reinforcement [10,36].
By estimating lp for a given column, du can be estimated due to
the good correlation between them. Paulay and Priestly [29] have
proposed a model predicting du as shown in Eq. (8) by conducting
curvature analysis of a cantilever column. This model defined the
ultimate drift ratio as the summation of two components. The first
component accounts for the lateral displacement at the yielding of
the steel reinforcement and the second component accounts for
the plastic displacement occurring after column yielding. This
model has not taken into consideration the effect of
FRP-confinement on the ultimate drift ratio of a reinforced column.
;y H ð;u  ;y Þlp ðH  0:5lp Þ
du ¼ þ ð8Þ
3 H
 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
where du is the ultimate drift ratio, H is the column shear span, ;y
and ;u are the yield and ultimate curvatures, respectively, and lp
is the plastic hinge length calculated by Eq. (2).
Gu et al. [31] have modified Eq. (8) to take into consideration
the effect of FRP-confinement on the plastic displacement of a
FRP-confined column, assuming that the FRP-confinement does
not significantly affect the column yield state. Eq. (9) shows their
proposed model.
2:45E sy H
du ¼
3D
" # 
0:008 þ 0:09kl lp lp
þ  2:45E sy 1  0:5
0:19 þ ðPP0 Þð0:72  0:67kl Þ H H D
where du is the ultimate drift ratio, E sy is the yield strain of the lon-
gitudinal reinforcement, H is the column shear span, D is the col-
umn diameter, kl is the confinement ratio, P is the column axial
load, P0 is the column axial load capacity, and lp is the plastic hinge
length calculated using Eq. (7).
In the current research, FE models developed using LS-DYNA
were used to investigate the lp in different FRP-confined RC col-
umns. Six FRP-confined RC columns were used to calibrate the pro-
posed FE model. A parametric study was conducted using the
developed FE model to investigate the effect of confinement on
the lp. The previous plastic hinge models discussed above were
used in this study to find the model that best fit the results of
the parametric study. Then, more accurate empirical models were
proposed to determine the values of lp and the du of FRP-confined
columns. The proposed du model predictions were compared to
the predictions of two previous models. This FE study applies only
to bridge columns/piers for which the seismic design philosophy is
the ‘‘weak’’ column that allows plastic deformations under a signif-
icant earthquake event; and to cantilever columns that deform in
the form of a single curvature under lateral load. This study can
help structural designers to investigate such column structures
during the design process. It can also help to minimize the extent
of retrofitting work using external FRP jacketing by knowing the
extent of the predicted concrete damage zone, so that the optimum
jacket length can be constructed, as the flexural jacketing only
needs to cover the plastic hinge region.
2. Finite element modeling
2.1. Model geometry
Each simulated column in this study consisted of a footing, a
FRP-confined column, and a loading stub. The concrete footing, col-
umn, and loading stub, as shown in Fig. 1, were modelled using
8-node constant stress solid hexahedron elements, which have
six degrees of freedom at each node. Single point volume integra-
tion was carried out by Gaussian quadrature. Hourglass control
with an hourglass coefficient of 0.1, as recommended by
LS-DYNA support [37], was provided in order to avoid the zero
energy modes. FRP jackets were modelled using 4-node shell ele-
ments with six degrees of freedom at each node. This element
includes membrane, bending and shear deformation capabilities.
The section attribute for this element was thickness alone. The
Belytschko–Tsay [38] element formulation was used for the shell
elements, which is the default theory for shell elements in
LS-DYNA [39] due to its computational efficiency. It is based on a
combined co-rotational and velocity strain formulation. The
co-rotational portion of the formulation avoids the complexities
of non-linear mechanics by embedding a co-ordinate system in
the element [40]. The reinforcement bars (longitudinal and
467
Axial load
Loading stub
Lateral load
FRP-confined column
 
H
ð9Þ Foong
Fig. 1. Geometry of the FE model.
transverse) were modelled using 2-node beam elements, called
Hughes–Liu, with six degrees of freedom at each node. The column
was laterally loaded in displacement control uniformly applied to
all nodes at the mid-height of the column loading stub (see
Fig. 1) using the BOUNDARY_PRESCRIBED_MOTION card. Another
node set was generated for the bottom and side nodes of the col-
umn footing. This node set was fully constrained in the global X,
Y, and Z axes displacements and rotations. The node-set constraint
actions were defined using the BOUNDARY_SPC_SET card. The axial
load on the column was applied on the top surface of the column
loading stub, using the LOAD_SEGMENT_SET card.
2.2. Material modelling
The K–C model, ⁄MAT_CONCRETE_DAMAGE_RELIII, was used
for simulating the concrete material for the columns [41]. It was
used in conjunction with an equation of state, EOS_
TABULATED_COMPACTION, which gives the current pressure,
P = C (E v ), as a function of current and previous volumetric strain.
In this function, C is a constant depends on the yield or failure state
of the material, and E v is the volumetric strain defined as the ratio
of the volume change to the original volume. This function was
entered as a series of P, E v pairs in the keyword input file [42].
The K–C model has the capability of auto-generating the required
model parameters, as well as, the equation of state parameters
based solely on the unconfined compressive strength of the con-
crete. Once the pressure is known, the stress tensor can be calcu-
lated as being a point of a moveable surface that can be a yield
surface or a failure surface. By knowing the unconfined concrete
strength of the column under modelling, the material model
parameters were automatically generated using a standard
100  200 mm cylinder and then these parameters were imple-
mented in the full scale model. However, parameter b1 was cali-
brated based on the full scale model and a value of 2.5 was used.
This parameter is a damage scaling parameter that governs the
strain softening in compression. It changes the rate at which the
damage occurs, and the stress–strain curve expands or compresses
along the strain axis depending on the value of b1 [42]. More
details and features of this material model can be found in [43,44].
The FRP Jacket was modelled using an orthotropic material model
called MAT_ORTHOTROPIC_ELASTIC_002 material model [39].
468 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
This material model allows the user to control the FRP shell prop-
erties in three dimensions and control the FRP-shell fibre direction.
In addition, it adopts laminated shell theory for the purpose of cor-
recting the assumption of a uniform constant shear strain through-
out the thickness of the composite shell; thus, avoiding very stiff
results [45]. The MAT_ADD_EROSION card was used to control
the failure stress of the FRP shell. The schematic stress–strain curve
of the used FRP material model is shown in Fig. 2(a), where, ft, E f ,
and Ef are the ultimate tensile strength, the ultimate tensile strain,
and the modulus of elasticity, respectively of the FRP.
The contact between the FRP jacket and the concrete column
was assumed as a perfect bond by sharing the same nodes at the
contact surface, as recommended by previous FE studies (e.g.
[14,40,46]). At the contact surface, the shell elements were mod-
elled as the slave segment set and the solid elements’ outer sur-
faces were modelled as the master segment set. The static and
dynamic coefficient of friction used was equal to 0.6, in conjunc-
tion with the soft constraint formulation (SOFT = 1.0) [37]. Recent
research [47] showed that using this contact approach was able
to capture not only the global behaviour of bridge columns but also
the peak strength in concrete at rupture. The height of the FRP shell
was 2 mm shorter than the column height at each end to avoid
local damage to the FRP shell due to bearing against the surround-
ing footing/loading stub during the lateral loading. The CONTACT_
NODE_TO_SURFACE card was used to model the contact between
the FRP jacket edges and both the loading stub and footing.
Since the behaviour of the loading stub and the footing are basi-
cally linear up to failure, they were represented using a linear elas-
tic material model (MAT_ELASTIC_001), having compressive
strength equal to that of the actual loading stub and column foot-
ing. The reinforcement bars were modelled using an elastoplastic
material model (MAT_PLASTIC_KINEMATIC_003). This material
model includes tensile stress–strain properties in which elasticity
modulus, strength, elongation, and Poisson’s ratio are the inputs.
The reinforcement was modelled using a discrete representation
with perfect bond between any reinforcement bars and the con-
crete. The schematic stress–strain curve of the used rebar material
model is shown in Fig. 2(b), where, fsy, fsu, E sy , E su , and Es are the
yield strength, the ultimate strength, the yield strain, the ultimate
strain, and the modulus of elasticity of the rebar.
The DATABASE_CROSS_SECTION_PLAN card was used to record
the required lateral force during the column loading until failure.
The cross section in this card was located horizontally just
above the footing surface. The DEFINE_CURVE card was used to
define the lateral displacement pattern and the axial load applied
to the column. Two loading steps were used for the analysis of
the models. During the first step, the axial load was applied as a
distributed stress on the top surface of the loading stub in the
Fig. 2. Schematic stress–strain curves for: (a) FRP material model and (b) rebar material model.
negative Z direction, see Fig. 1. The second loading step consisted
of a monotonic push in the X-direction simulated by a linearly
increasing lateral displacement until the failure of the model
occurred and the analysis was not able to proceed any further.
The stress–strain curve that can be obtained from monotonic load-
ing of RC column is almost identical to the envelope (backbone)
curve of the stress–strain behaviour of RC columns under cyclic
loading as known from the literature [48–52]. However, slight dif-
ferences could be observed in the ultimate strength and strain
when comparing similar columns under cyclic or monotonic load-
ing. In addition, more cover spalling and reinforcement buckling
can be experienced by columns under cyclic loading [48–52]. The
FE analysis for each simulated column was terminated when the
recorded lateral force decreased by 20% of the ultimate recorded
load which was then taken as the column failure point.
3. Experimental database
A total of 37 circular FRP-confined RC columns were collected
from the literature as an experimental database in this study, see
0
Table 1. In this table, f c is the unconfined concrete compressive
strength, q is the longitudinal reinforcement ratio calculated as
the area of longitudinal steel reinforcement divided by the column
cross section area, tf is the saturated FRP thickness, ft is the FRP ulti-
mate tensile strength in the circumferential direction, E f is the FRP
rupture strain in the circumferential direction, and Ef is the FRP
modulus of elasticity in the circumferential direction. The axial
load capacity ðP0 Þ of each column was calculated using Eq. (10)
according to ACI 318-11 [53]
0
P0 ¼ 0:85ðAg  As Þf c þ As f sy ð10Þ
where P0 is the column axial load capacity, Ag is the gross area of
column cross section, As is the total area of longitudinal reinforce-
0
ment, f c is the unconfined concrete compressive strength, and f sy
is the longitudinal rebar yield strength.
The data in Table 1 were collected from Desprez et al. [13],
Ozbakkaloglu and Saatcioglu [15], Idris and Ozbakkaloglu [36], Li
and Sung [54], Gu et al. [16], Sheikh and Yan [55], Youssf et al.
[23], Xiao et al. [56], and Wang et al. [1]. In the collected database,
D ranged from 150 to 760 mm, H ranged from 630 to 2045 mm,
0
H/D ranged from 1.5 to 7.4, f c ranged from 18.6 to 95.0 MPa, P/P0
ranged from 3% to 68%, tf ranged from 0.11 to 10.16 mm, ft ranged
from 552 to 4300 MPa, and Ef ranged from 38 to 260 GPa. The data-
base included slender and short columns having only flexural only
or flexural–shear model of failure as reported by each individual
author in this database. The FRP confinements in this study were
formed using carbon, glass, aramid, and dyneema fibres oriented
 O. Youssf et al. / Engineering Structures 101 (2015) 465–476 469

Table 1
Experimental database of FRP-confined RC columns.
0
No. ID D H H/D fc q db fsy E sy P P/P0 FRP tf ft Ef Ef Source Mode of
(mm) (mm) (MPa) (%) (mm) (MPa) (%) (kN) (%) type (mm) (MPa) (%) (GPa) failure

1 P1C 300 2045 6.8 35.8 2.53 19.5 415 0.23 224 8 C 1.02 849 1.20 71 [13] Flexural
2 P2C 300 2045 6.8 34.9 2.53 19.5 415 0.23 867 31 C 1.02 849 1.20 71 [13] Flexural
3 P3C 300 2045 6.8 34.4 2.53 19.5 415 0.23 249 9 C 1.02 849 1.20 71 [13] Flexural
4 P4C 300 2045 6.8 34.3 2.53 19.5 415 0.23 805 29 C 1.02 849 1.20 71 [13] Flexural
5 RC2 270 2000 7.4 75.2 2.81 16.0 500 0.24 1480 34 C 0.33 3800 1.67 227 [15] Flexural
6 RC3 270 2000 7.4 49.7 2.81 16.0 500 0.24 1480 47 C 0.33 3800 1.67 227 [15] Flexural
7 RC1 270 2000 7.4 90.1 2.81 16.0 500 0.24 1580 31 C 0.66 3800 1.67 227 [15] Flexural
8 RC4 270 1200 4.4 75.3 2.81 16.0 500 0.24 1480 34 C 0.33 3800 1.67 227 [15] Flexural
9 CFFT 150 1000 6.7 95.0 3.56 10.0 600 0.30 725 41 A 0.60 2900 2.50 116 [36] Flexural
10 FCS-1 760 1750 2.3 18.6 1.87 19.0 426 0.21 422 4 C 0.41 4170 1.80 232 [54] Flexural
11 FCS-2 760 1750 2.3 18.6 1.87 19.0 426 0.21 422 4 C 0.28 4170 1.80 232 [54] Flexural
12 CH1 360 1100 3.1 34.9 6.09 25.0 382 0.19 1200 24 C+D 0.34 2346 2.67 109 [16] Flexural
13 CH3 360 1100 3.1 34.9 6.09 25.0 382 0.19 1200 24 C 0.25 3945 1.52 260 [16] Flexural
14 J1 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 D 0.26 1832 3.05 60 [16] Flexural
15 J2 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 C 0.11 4232 1.84 230 [16] Flexural
16 J3 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 C 0.11 4232 1.84 230 [16] Flexural
17 CL1 360 800 2.2 34.9 6.09 25.0 382 0.19 1200 24 D 1.03 1832 3.05 60 [16] Flexural
18 CL2 360 800 2.2 34.9 6.09 25.0 382 0.19 1200 24 C 0.42 3945 1.52 260 [16] Flexural
19 ST-2NT 356 1470 4.1 40.4 2.96 25.0 500 0.25 2570 54 G 1.25 828 2.00 41 [55] Flexural
20 ST-3NT 356 1470 4.1 40.4 2.96 25.0 500 0.25 2630 55 C 1.00 805 1.40 58 [55] Flexural
21 ST-4NT 356 1470 4.1 44.8 2.96 25.0 500 0.25 1380 27 C 0.50 1675 1.40 120 [55] Flexural
22 CCF2 240 1500 6.3 62.5 1.55 12.0 550 0.28 110 4 C 0.26 4300 1.80 234 [23] Flexural
23 CRCF2 240 1500 6.3 46.4 1.55 12.0 550 0.28 110 5 C 0.26 4300 1.80 234 [23] Flexural
24 CS-ISJ-RT 610 915 1.5 35.9 1.94 19.0 303 0.15 655 6 G 10.16 552 1.50 38 [56] Flexural–shear
25 CS-CSJ-RT 610 915 1.5 35.9 1.94 19.0 303 0.15 655 6 G 7.62 552 1.50 38 [56] Flexural–shear
26 J4 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 D 0.516 1832 3.05 60 [16] Flexural
27 J5 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 C 0.222 4232 1.84 230 [16] Flexural
28 J6 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 C+D 0.369 2552 2.68 111 [16] Flexural
29 J7 300 850 2.8 28.0 4.87 19.0 400 0.20 100 3 C+D 0.3135 2255 2.84 90 [16] Flexural
30 CH2 360 1100 3.1 34.9 6.09 25.0 382 0.19 1200 24 D 0.65 1832 3.05 60 [16] Flexural
31 CL3 360 800 2.2 34.9 6.09 25.0 382 0.19 1200 24 C 0.58 3945 1.52 260 [16] Flexural
32 C1n1 180 630 3.5 54.8 2.67 12.0 354 0.18 628 45 C 0.80 3430 1.80 230 [1] Flexural
33 C1n2 180 630 3.5 54.8 2.67 12.0 354 0.18 767 55 C 0.80 3430 1.80 230 [1] Flexural
34 C1n3 180 630 3.5 54.8 2.67 12.0 354 0.18 906 65 C 0.80 3430 1.80 230 [1] Flexural
35 C2n1 180 630 3.5 71.2 2.67 12.0 354 0.18 815 47 C 0.80 3430 1.80 230 [1] Flexural
36 C2n2 180 630 3.5 71.2 2.67 12.0 354 0.18 997 57 C 0.80 3430 1.80 230 [1] Flexural
37 C2n3 180 630 3.5 71.2 2.67 12.0 354 0.18 1178 68 C 0.80 3430 1.80 230 [1] Flexural
C Carbon fiber A Aramid fiber
G Glass fiber D Dyneema fiber

in each column’s circumferential direction as specified by each
individual author in this database. The FRP confinement included
both filament wound tubes and FRP wraps. Researchers [57,58]
reported that there was no significant difference in behaviour
between concrete encased in these two different types of FRP
tubes.
all results of the FE model. The error % in this table was calculated
using Eq. (11). As shown in the table, du predictions were more
accurate than the predictions of Fu. However, the maximum error
in both Fu and du is only 10.9%. The FE model predicted du with
absolute average of errors (A) and standard deviation (SD) of
4.4% and 1.7%, respectively and predicted Fu with A and SD of
6.9% and 3.4%, respectively.

jEXP: result  FE resultj

4. Validation of the proposed FE model
The first six specimens of the database described above were
used to calibrate the proposed FE model. The calibration was car-
ried out by comparing the predicted backbone curves with the cor-
responding experimental ones. The ultimate lateral force (Fu) and
ultimate displacement (du) were also compared. Table 2 shows
Error ð%Þ ¼ x100 ð11Þ
EXP: result
Fig. 3 shows the FE model backbone curve predictions compared to
the corresponding experimental ones. As shown in the figure, the FE
models were able to capture the elastic and plastic behaviour of the
column specimens quite well. The errors in the backbone predic-
tions were calculated as the errors in the predicted forces at

Table 2
Results of the proposed FE model.

No. ID D (mm) H (mm) F u EXP: (kN) F u FE (kN) Error (%) du EXP: (mm) du FE (mm) Error (%) Source Mode of failure

1 P1C 300 2045 61 64 6.2 286 270 5.61 [13] Flexural failure
2 P2C 300 2045 89 87 2.8 243 227 6.42 [13] By FRP tensile
3 P3C 300 2045 63 69 8.8 286 299 4.62 [13] Rupture in the
4 P4C 300 2045 88 86 2.3 150 148 0.90 [13] Plastic hinge
5 RC2 270 2000 77 86 10.9 200 192 3.95 [15] Region in all
6 RC3 270 2000 81 89 10.4 181 173 4.76 [15] Columns
470 O. Youssf et al. / Engineering Structures 101 (2015) 465–476

Fig. 3. Experimental versus FE backbone curve predictions.

different drifts and then plotted in Fig. 4. As shown in the figure,
about 90% of the compared curves data resulted in errors less than
10% and for the remaining 10% the predicted errors ranged from
10% to 14%. In addition to the FE backbone predictions, the FE model
was able to show the location of the high strain concentration
region (plastic hinge region), as well as the rupture of the FRP shell
as shown in Fig. 5. This can help to investigate and model the plastic
hinge length. The failure in all the simulated columns was due to
the FRP-shell tensile rupture in the circumferential direction which
is the same direction as the fibre orientation. Noting that, all spec-
imens used in this study were confined using unidirectional FRP.
5. Modelling of the plastic hinge length
The plastic hinge length of a cantilever column is the summa-
tion of two components namely; the column bending component
(lpb) and the strain penetration component (lps) as explained in
Eq. (2) [29]. Since the strain penetration component is independent
of the column properties and depends only on the reinforcement
properties [16], it was kept as ½lps ¼ 0:022f sy db  as recommended
by [16,29], and the bending component was measured using the
previously described FE model. The FE model was able to show
the general behaviours of the columns and their backbone curves
 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
Fig. 4. Errors in the predicted force versus drift for backbone curves.
quite well (Fig. 3). Based on that, the lpb was measured using the
technique previously proposed by [15,18] and reported for each
specimen. This technique is based on the intimate relationship that
exists between the lateral expansion of the FRP tube and the level
FRP-shell
(a)
FRP-shell
(c)
Fig. 5. Performance of plastic hinge region in the FE model: (a) FRP tensile strain concentration (b) FRP tensile rupture in the circumferential direction, (c) FRP compressive
strain concentration, and (d) concrete compressive strain concentration.
471
of damage sustained by concrete inside the tube. Higher FRP-tube
hoop strains correspond to the most highly damaged regions of the
columns, since the concrete undergoes rapid expansion inside the
FRP tube within these regions. This technique assumes that, at the
ultimate column displacement, the plastic hinge region terminates
at a height above the column footing where the recorded hoop
strain values are below 1/3 of the maximum recorded strain. This
technique was verified through measurements of column rotations
and strains on longitudinal reinforcement carried out by [10,36].
In this study, the maximum tensile hoop strain experienced at
each level of the column height above the column footing was
recorded at each column’s ultimate displacement. Then the
hoop-strain profile along the column height was used to determine
lpb as the height above the column footing where the recorded
hoop strain values were below 1/3 of the maximum recorded
strain. Fig. 6 shows an example of measuring the lpb for specimen
RC3 in Table 1.
A parametric study was conducted on the six simulated col-
umns in Table 2 to better understand the effect of the
FRP-confinement on lpb using the above described FE model by
FRP-shell
FRP-rupture
(b)
concrete
(d)
472 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
Fig. 6. Example of measuring the lpb for specimen RC3 in Table 1.
running each model eight times with different values of kl . The
h i
2f t
confinement ratio kl ¼ Dft 0 f of each column was changed by
c
changing the confinement thickness, tf. ACI 440-2R [59] recom-
mended that the confinement ratio should be larger than 0.08.
Moreover, in practice, achieving a passive confinement ratio as
large as 0.70 is quite difficult considering the large size of real con-
crete columns and the cost of confinement materials [60]. In previ-
ous experimental studies, the confinement ratio usually ranged
between 0.10 and 0.50. Therefore, the chosen confinement ratio
limits in this parametric study ranged between 0.08 and 0.70 to
cover the recommended ranges.
Fig. 7 shows the results of the parametric study. Generally,
increasing the confinement ratio increases the plastic hinge length.
Up to a confinement ratio of 0.4 the rate of increase in lpb is rela-
tively higher than the rate of increase beyond a confinement ratio
of 0.4. The increase in lpb with the increase in kl could be attributed
to the increase in the concrete compressive strength caused by the
effect of confinement, resulting in a column cross section moment
capacity increase and hence increase in the section curvature. This
conclusion is not in agreement with that of Gu et al. [31] in which
lpb increased with increasing kl until a value of 0.1 and then it
decreased as kl increased from 0.1 to 0.5. The decreased signifi-
cance of the FRP-confinement at high kl in this parametric study
can be attributed to the column failure mode. At relatively low
confinement ratios the column’s failure was mainly due to tensile
rupture of the FRP. However, at relatively high confinement ratios
the column’s failure was mainly due to rupture of the longitudinal
reinforcement rather than rupture of the FRP, thus reducing the
significance of the FRP-confinement.
The plastic hinge lengths ½lp ¼ lpb þ lps  measured for the original
six calibrated specimens and the parametric study resulted in a
total of fifty four lp values. The previous plastic hinge models dis-
cussed in the introduction section were used in this study to find
Fig. 7. Effect of confinement ratio on the plastic hinge length.
the model that best fit the results of these fifty four lp values. By
investigating the A and the SD in the predictions of these fifty four
lp values using those previous models, the Paulay and Priestley [29]
model (Eq. (2)) gave the lowest A and SD compared to the other
previous models. This model defines the plastic hinge length as
the summation of two components. The first component is the
effect of the moment gradient along the column height presented
as a function of column height, and the second component is the
effect of additional rotation resulting from the strain penetration
of longitudinal reinforcement into column footing.
In this study, the Paulay and Priestley [29] model was modified
to take into consideration the effect of the FRP-confinement and
hence generate better lp predictions for FRP-confined columns,
with lower A and SD. The final form of the modified lp model is
shown in Eq. (12). The A and SD of the proposed lp model were
29.1% and 17.9%, respectively, compared to the A and SD of 57.6%
and 9.5%, respectively of the Paulay and Priestley [29] model pre-
dictions, through the fifty four lp values. Thus, the proposed lp
model reduced the A by 69% compared to the earlier model [29].
The proposed lp model will be used in developing the prediction
of the ultimate drift ratio of FRP-confined columns in next section.
lp ¼ 0:8kl H þ 0:022f sy db ð12Þ
where lp is the plastic hinge of FRP-confined column, kl is the con-
finement ratio, H is the column shear span, f sy is the expected rebar
yield strength, and db is the longitudinal rebar diameter.
6. Modelling of the ultimate drift ratio
In order to model the column ultimate drift ratio, the Paulay
and Priestly [29] model (Eq. (8)) and Gu et al. [31] model (Eq.
(9)) discussed earlier, were used to predict the results of the col-
lected database as shown in Table 3. The error % in this table was
calculated using Eq. (13).
jExperimental drift ratio  Model drift ratioj
Error ð%Þ ¼
Experimental drift ratio
 100 ð13Þ
To keep the consistency of the respective models it should be
noted that lp in Eq. (8) was calculated using Eq. (2) and lp in Eq.
(9) was calculated using Eq. (7). Since the FRP-confinement does
not significantly affect the column yield state [61], the yield curva-
ture in Eq. (8) was determined using the unconfined column for-
h i
2:45e
mula as, ;y ¼ D sy [24]. However, the ultimate curvature in Eq.
(8) should be determined using formula that takes into considera-
tion the FRP-confinement effect on the column ultimate state
[26,31]. In order to do that, the Binici [26] ;u model and the Gu
et al. [31] ;u model were used and the results were compared.
Gu et al.’s ;u model was able to decrease the A and the SD in the
du predictions by 52.4% and 2.5%, respectively, compared to
Binici’s ;u model. Therefore, Gu et al.’s ;u model, shown in Eq.
(14), was used in this study.
0:008 þ 0:09kl
;u ¼ ð14Þ
ð0:19 þ ð0:72  0:67kl Þ  nÞD
where ;u is the ultimate curvature, kl is the column confinement
ratio, n is the column axial load ratio, and D is the column diameter.
The Paulay and Priestly [29] model (Eq. (8)) was able to
decrease the A and the SD in the predict du by 17.4% and 44.2%,
respectively, compared to the A and the SD in the predict du by
the Gu et al. [31] model through the experimental database.
Thus, it was decided in this study to modify the Paulay and
Priestly model to achieve better predictions of du . The final modi-
fied model is shown in Eq. (15).
 O. Youssf et al. / Engineering Structures 101 (2015) 465–476 473

Table 3
Predicted du values using different models.

No. ID EXP. du Gu et al. [31] Paulay and Priestly [29] Proposed model (Eq. (15)) Source
du Error (%) ;u (Eq. (14)) du Error (%) ;u (Eq. (15c)) du Error (%)

1 P1C 0.140 0.194 38.5 3.1E04 0.105 24.6 3.0E04 0.124 11.1 [13]
2 P2C 0.119 0.122 3.3 2.0E04 0.070 41.2 1.8E04 0.079 33.7 [13]
3 P3C 0.140 0.189 35.5 3.1E04 0.105 24.8 3.0E04 0.125 10.8 [13]
4 P4C 0.073 0.127 73.0 2.1E04 0.072 1.0 1.9E04 0.082 12.3 [13]
5 RC2 0.100 0.117 17.3 1.7E04 0.061 38.5 1.7E04 0.066 33.7 [15]
6 RC3 0.091 0.110 21.3 2.0E04 0.068 24.9 1.7E04 0.077 15.5 [15]
7 RC1 0.121 0.139 14.8 2.6E04 0.089 26.1 2.3E04 0.105 13.3 [15]
8 RC4 0.040 0.076 89.3 1.7E04 0.045 13.0 1.7E04 0.048 19.7 [15]
9 CFFT 0.103 0.124 20.1 4.8E04 0.097 5.8 3.8E04 0.107 3.2 [36]
10 FCS-1 0.068 0.076 12.6 1.9E04 0.056 17.9 1.6E04 0.072 6.5 [54]
11 FCS-2 0.058 0.075 29.5 1.4E04 0.042 27.1 1.3E04 0.050 13.8 [54]
12 CH1 0.050 0.066 32.4 1.6E04 0.042 16.1 1.6E04 0.044 11.1 [16]
13 CH3 0.080 0.072 10.1 1.8E04 0.048 39.8 1.7E04 0.051 36.4 [16]
14 J1 0.085 0.096 13.5 2.9E04 0.059 30.5 3.1E04 0.065 23.4 [16]
15 J2 0.086 0.096 11.9 2.8E04 0.059 31.5 3.1E04 0.065 24.5 [16]
16 J3 0.086 0.096 11.9 2.8E04 0.059 31.5 3.1E04 0.065 24.5 [16]
17 CL1 0.068 0.074 9.3 3.1E04 0.071 3.9 2.3E04 0.070 3.5 [16]
18 CL2 0.060 0.071 17.8 2.7E04 0.062 4.0 2.2E04 0.062 4.0 [16]
19 ST-2NT 0.046 0.060 31.2 1.1E04 0.041 11.4 1.1E04 0.042 9.0 [55]
20 ST-3NT 0.046 0.053 15.1 9.3E05 0.034 25.4 9.5E05 0.036 22.3 [55]
21 ST-4NT 0.080 0.078 2.4 1.3E04 0.048 39.5 1.4E04 0.051 35.6 [55]
22 CCF2 0.097 0.197 102.7 4.1E04 0.106 9.6 4.1E04 0.123 27.0 [23]
23 CRCF2 0.098 0.199 103.2 4.9E04 0.125 27.2 4.4E04 0.153 56.1 [23]
24 CS-ISJ-RT 0.125 0.053 57.4 4.2E04 0.075 40.4 2.9E04 0.106 15.0 [56]
25 CS-CSJ-RT 0.173 0.049 71.7 3.2E04 0.058 66.8 2.4E04 0.077 55.7 [56]
26 J4 0.126 0.119 5.5 4.5E04 0.093 26.6 4.0E04 0.104 17.1 [16]
27 J5 0.112 0.119 6.2 4.5E04 0.092 17.8 4.0E04 0.104 7.2 [16]
28 J6 0.125 0.119 4.8 4.5E04 0.092 26.2 4.0E04 0.104 16.7 [16]
29 J7 0.109 0.111 1.5 3.7E04 0.076 30.7 3.5E04 0.084 23.2 [16]
30 CH2 0.130 0.077 41.1 2.1E04 0.055 57.9 1.8E04 0.058 55.5 [16]
31 CL3 0.060 0.080 33.7 3.7E04 0.085 42.4 2.7E04 0.085 41.3 [16]
32 C1n1 0.114 0.091 20.2 9.3E04 0.121 5.5 5.2E04 0.137 19.3 [1]
33 C1n2 0.116 0.083 28.4 8.5E04 0.110 5.5 4.6E04 0.121 3.8 [1]
34 C1n3 0.126 0.077 39.2 7.8E04 0.101 19.9 4.2E04 0.108 14.1 [1]
35 C2n1 0.136 0.068 50.2 6.6E04 0.086 36.9 4.2E04 0.097 28.8 [1]
36 C2n2 0.133 0.061 54.1 5.9E04 0.077 41.9 3.7E04 0.085 35.9 [1]
37 C2n3 0.137 0.056 59.4 5.4E04 0.070 48.8 3.3E04 0.076 44.4 [1]
A (%) 32.2 26.6 22.4
SD (%) 27.4 15.3 14.7
Max. error (%) 103.2 66.8 56.1

;y H ð;u  ;y Þlp ðH  0:5lp Þ

du ¼ þ ð15aÞ
3 H

2:45esy
;y ¼ ð15bÞ
D

0:013 þ 0:05kl
;u ¼ ð15cÞ
0:18D þ PP0 ð0:7  0:25kl ÞD

where du is the ultimate drift ratio of the FRP-confined column, H is

the column shear span, ;y is the yield curvature, ;u is the ultimate
curvature, lp is the plastic hinge length of the FRP-confined column,
calculated using Eq. (12), E sy is the yield strain of the longitudinal
reinforcement, D is the column diameter, kl is the column confine-
ment ratio, P is the column axial load, and P0 is the column axial
Fig. 8. A, SD, and max error comparisons for all du models.
load capacity.
Compared to the experimental results, the A, SD, and max. error
were 22.4% 14.7%, and 56.1%, respectively in the proposed du model models. As shown in the figure, the scatter in the proposed model
predictions, 26.6%, 15.3%, and 66.8%, respectively in the Paulay and predictions (R2 = 0.36) is less than that of the previous models and
Priestly model predictions, and 32.2%, 27.4%, and 103.2%, respec- the linear trend of the present model predictions is closer to the
tively in the Gu et al. model prediction, as shown in Fig. 8. Thus, neutral line than the previous models, indicating better predictions
the proposed model reduced A, SD, and max. error by 15.1%, for du .
3.9%, and 16.0%, respectively compared to the Paulay and Priestly The effect of kl on du for both experimental and model predic-
model, and by 30.4%, 46.3%, and 45.6%, respectively compared to tions can be seen in Fig. 10 which indicates that du increases with
the Gu et al. model. Fig. 9 shows the experimental results versus increasing column kl . This is attributed to the ductility that the FRP
the predicted ones using the proposed model and the previous confinement provides for the RC column, which allows the column
474 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
Fig. 9. Experimental versus model predictions: (a) Gu et al. model, (b) Paulay and Priestly model, and (c) Proposed model.
Fig. 10. Effect of kl on du .
to undergo higher lateral movement without failure [11,14]. The
figure also illustrates that the linear trend of the model predictions
is close to that of the experimental results, especially for confine-
ment ratios up to 0.3. Beyond that and due to the lack of data
results in that range, the accuracy of the model prediction is less.
More experimental studies are recommended using confinement
ratios higher than 0.3 to further improve the model. In Fig. 11,
the effect of column aspect ratio (H/D), reinforcement ratio (q),
and axial load ratio (P/P0) on the proposed du model error is pre-
sented. As shown in the figure, there are no strong correlations
between these column parameters and the error variation in the
predicted du (low R2 values). This indicates that the error in the
proposed model predictions is independent of these column
parameters.
7. Summary and conclusions
This paper presents a finite element (FE) model that was devel-
oped using the LS-DYNA program to model the plastic hinge length
of FRP-confined reinforced concrete (RC) columns subjected to lat-
eral loading. The proposed FE model was calibrated using experi-
mental results of six different FRP-confined RC columns. A FE
parametric study on the effect of FRP-confinement on plastic hinge
length was conducted using the calibrated specimens. Empirical
models were proposed to predict the plastic hinge lengths and ulti-
mate drifts for FRP-confined RC columns. The proposed ultimate
drift ratio model predictions were compared to the predictions of
previous models by utilising the experimental results of 37 differ-
ent FRP-confined RC columns. This FE study applies to bridge
columns/piers that allow plastic deformations under a significant
earthquake event and cantilever columns that deform in the form
of a single curvature under lateral load. The main findings of this
study are summarized in the following points:
 The FE model predicted ultimate displacements with absolute
average of errors (A) and standard deviation (SD) of 4.4% and
1.7%, respectively and predicted ultimate forces with A and SD
of 6.9% and 3.4%, respectively.
 90% of the backbone curves predicted by the FE model were
within 10% of the experimental curves when compared for
every 1% drift.
 Increasing the confinement ratio increases the plastic hinge
length. Up to a confinement ratio of 0.4 the rate of increase in
lp is higher than the rate of increase beyond the confinement
ratio of 0.4.
 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
Fig. 11. Effect of column parameters on the proposed du model prediction: (a) aspect ratio, (b) reinforcement ratio, and (c) axial load ratio.
 The proposed lp model resulted in A and SD of 29.1% and 17.9%,
respectively, compared to A and SD of 57.6% and 9.5%, respec-
tively, in the Paulay and Priestley model predictions for the fifty
four lp FE results.
 Paulay and Priestly model was able to decrease the A and the SD
in the predict du by 17.4% and 44.2%, respectively, compared to
the A and the SD in the predict du by Gu et al. model.
 Compared to the experimental results, the A, SD, and max.
errors were 22.4% 14.7%, and 56.1%, respectively in the pro-
posed du model predictions, representing significant improve-
ments compared with previous models.
 The proposed du model decreased A, SD, and max. error by
15.1%, 3.9%, and 16.0%, respectively compared to Paulay and
Priestly model, and by 30.4%, 46.3%, and 45.6%, respectively
compared to Gu et al. model.
 The proposed du model decreased the scatter in the drift ratio
predictions compared to previous models.
 The proposed du model demonstrated an increase in the ulti-
mate drift ratio as the confinement increased as evidenced by
the experimental results. The linear trend of the model predic-
tions was close to the linear trend of the corresponding experi-
mental results, especially for confinement ratios up to 0.3.
Research recommendations
For further improvement of the model, the following points are
recommended:
 More experimental studies are strongly recommended for con-
finement ratios higher than 0.3.
 Short columns that displayed pure shear mode of failure are
recommended to be considered in the applications of the pro-
posed model.
475
References
[1] Wang Z, Wang D, Sheikh S, Liu J. Seismic performance of FRP-confined circular
RC columns. Advances in FRP composites in civil engineering. Springer; 2011.
p. 810–4.
[2] Zhou C, Lu X, Li H, Tian T. Experimental study on seismic behavior of circular
RC columns strengthened with pre-stressed FRP strips. Earthq Eng Eng Vib
2013;12:625–42.
[3] Liu J-X, Zhao G-H. Typical bridge damage analysis in’’ 5 12’’ Wenchuan
earthquake. J Arch Civ Eng 2009;2:019.
[4] Zhuang W, Liu Z, Jiang J. Earthquake-induced damage analysis of highway
bridges in Wenchuan earthquake and countermeasures. Chin J Rock Mech Eng
2009;7:013.
[5] Mortezaei A, Ronagh HR. Plastic hinge length of FRP strengthened reinforced
concrete columns subjected to both far-fault and near-fault ground motions.
Sci Iranica 2012;19:1365–78.
[6] Youssf O, ElGawady MA, Mills JE. Experimental investigation of crumb rubber
concrete columns under seismic loading. Structures 2015;3:13–27.
[7] ElGawady MA, Endeshaw M, McLean D, Sack R. Retrofitting of
rectangular columns with deficient lap splices. J Compos Constr
2010;14:22–35.
[8] ElGawady MA, Booker AJ, Dawood HM. Seismic behavior of posttensioned
concrete-filled fiber tubes. J Compos Constr 2011;14:616–28.
[9] ElGawady MA, Sha’lan A. Seismic behavior of self-centering precast segmental
bridge bents. J Bridge Eng 2011;16:328–39.
[10] Ozbakkaloglu T, Idris Y. Seismic behavior of FRP-high-strength concrete–steel
double-skin tubular columns. J Struct Eng 2014:140.
[11] Youssf O, ElGawady MA, Mills JE, Ma X. An experimental investigation of
crumb rubber concrete confined by fibre reinforced polymer tubes. Constr
Build Mater 2014;53:522–32.
[12] ElGawady MA, Dawood HM. Analysis of segmental piers consisted of concrete
filled FRP tubes. Eng Struct 2012;38:142–52.
[13] Desprez C, Mazars J, Kotronis P, Paultre P. Damage model for FRP-confined
concrete columns under cyclic loading. Eng Struct 2013;48:519–31.
[14] Youssf O, ElGawady MA, Mills JE, Ma X. Finite element modelling and dilation
of FRP-confined concrete columns. Eng Struct 2014;79:70–85.
[15] Ozbakkaloglu T, Saatcioglu M. Seismic behavior of high-strength concrete
columns confined by fiber-reinforced polymer tubes. J Compos Constr
2006;10:538–49.
[16] Gu D-S, Wu G, Wu Z-S, Wu Y-F. Confinement effectiveness of FRP in
retrofitting circular concrete columns under simulated seismic load. J
Compos Constr 2010;14:531–40.
476 O. Youssf et al. / Engineering Structures 101 (2015) 465–476
[17] Hashemi SH, Rahgozar R, Maghsoudi A. Flexural testing of high strength
reinforced concrete beams strengthened with CFRP sheets. Int J Eng, Trans B
2009;22:131–46.
[18] Ozbakkaloglu T, Saatcioglu M. Seismic performance of square high-strength
concrete columns in FRP stay-in-place formwork. J Struct Eng
2007;133:44–56.
[19] Dawood H, ElGawady MA, Hewes J. Behavior of segmental precast
posttensioned bridge piers under lateral loads. J Bridge Eng 2011;17:
735–46.
[20] Sheikh SA, Jaffry SA, Cui C. Investigation of glass-fibre-reinforced-polymer
shells as formwork and reinforcement for concrete columns. Can J Civ Eng
2007;34:389–402.
[21] Bae S, Bayrak O. Plastic hinge length of reinforced concrete columns. ACI Struct
J 2008:105.
[22] Fakharifar M, Chen G, Arezoumandi M, ElGawady MA. Hybrid jacketing for
rapid repair of seismically damaged reinforced concrete columns.
Transportation research board 94th annual meeting; 2015.
[23] Youssf O, ElGawady MA, Mills JE. Static cyclic behaviour of FRP-confined
crumb rubber concrete columns. Eng Struct 2015 [in preparation].
[24] Priestley M, Seible F, Calvi G. Seismic design and retrofit of bridges. John Wiley
& Sons; 1996.
[25] Elsanadedy HM, Haroun MA. Seismic design guidelines for squat composite-
jacketed circular and rectangular reinforced concrete bridge columns. ACI
Struct J 2005:102.
[26] Binici B. Design of FRPs in circular bridge column retrofits for ductility
enhancement. Eng Struct 2008;30:766–76.
[27] Monti G, Nisticò N, Santini S. Design of FRP jackets for upgrade of circular
bridge piers. J Compos Constr 2001;5:94–101.
[28] Zhao X-M, Wu Y-F, Leung A. Analyses of plastic hinge regions in
reinforced concrete beams under monotonic loading. Eng Struct
2012;34:466–82.
[29] Paulay T, Priestly MJN. Seismic design of reinforced concrete and masonry
buildings. New York: John Wiley & Sons, Inc.; 1992.
[30] Berry MP, Lehman DE, Lowes LN. Lumped-plasticity models for performance
simulation of bridge columns. ACI Struct J 2008:105.
[31] Gu D-S, Wu Y-F, Wu G, Wu Z-s. Plastic hinge analysis of FRP confined circular
concrete columns. Constr Build Mater 2012;27:223–33.
[32] Priestley M, Park R. Strength and ductility of concrete bridge columns under
seismic loading. ACI Struct J 1987:84.
[33] Lu Y, Gu X, Guan J. Probabilistic drift limits and performance evaluation of
reinforced concrete columns. J Struct Eng 2005;131:966–78.
[34] Pam H, Ho J. Length of critical region for confinement steel in limited
ductility high-strength reinforced concrete columns. Eng Struct 2009;31:
2896–908.
[35] Liu J. Seismic behaviour of reinforced concrete columns. University of Toronto;
2013.
[36] Idris Y, Ozbakkaloglu T. Seismic behavior of high-strength concrete-filled FRP
tube columns. J Compos Constr 2013:17.
[37] LS-DYNA_support. <http://www.dynasupport.com/howtos/element/hourglass>.
[38] Belytschko T, Lin JI, Chen-Shyh T. Explicit algorithms for the
nonlinear dynamics of shells. Comput Methods Appl Mech Eng 1984;42:
225–51.
[39] LS-DYNA Keyword User’s Manual. Version 971. Livermore Software
Technology Corp, Livermore, CA, USA; 2007.
[40] Elsanadedy HM, Al-Salloum YA, Alsayed SH, Iqbal RA. Experimental and
numerical investigation of size effects in FRP-wrapped concrete columns.
Constr Build Mater 2012;29:56–72.
[41] Malvar L, Crawford J, Morrill K. K&C concrete material model Release III –
Automated generation of material model input. Karagozian and Case
Structural Engineers, Technical Report TR-99-2432000.
[42] Markovich N, Kochavi E, Ben-Dor G. An improved calibration of the concrete
damage model. Finite Elem Anal Des 2011;47:1280–90.
[43] Simons D. Recent modifications to DYNA3D model 16 for concrete. In:
Proceedings, DNACWE structural analysis meeting, Logicon RDA, Albuquerque,
NM; 1995. p. 141–57.
[44] Malvar L, Crawford J, Wesevich J, Simons D. A new concrete material model for
DYNA3D-Release II: shear dilation and directional rate enhancements. A report
to defense nuclear agency under contract no DNA001-91-C-00591996.
[45] Hallquist JO. LS-DYNA3D theoretical manual. Livermore software technology
corporation; 1993.
[46] Mohammed TA. Reinforced concrete structural members under impact
loading. The University of Toledo; 2012.
[47] Abdelkarim OI, ElGawady MA. Analytical and finite-element modeling of FRP–
concrete–steel double-skin tubular columns. J Bridge Eng 2014;20. SPECIAL
ISSUE: Design, Analysis, and Construction of Segmental Bridges, B4014005.
[48] Park R, Paulay T. Reinforced concrete structures. John Wiley & Sons; 1975. p. C-
53–4.
[49] Fardis MN. Seismic design, assessment and retrofitting of concrete buildings:
based on EN-Eurocode 8. Springer Science & Business Media; 2009.
[50] Lampropoulos AP, Dritsos SE. Concrete shrinkage effect on the behavior of RC
columns under monotonic and cyclic loading. Constr Build Mater
2011;25:1596–602.
[51] Abbasnia R, Hosseinpour F, Rostamian M, Ziaadiny H. Cyclic and monotonic
behavior of FRP confined concrete rectangular prisms with different aspect
ratios. Constr Build Mater 2013;40:118–25.
[52] Sharbatdar MK. Monotonic and cyclic loading of new FRP reinforced concrete
cantilever beams. Int J Civ Eng 2008;6:58–71.
[53] ACI-318. Building code requirements for structural concrete and
commentary. American Concrete Institute; 2011. p. 503.
[54] Li Y-F, Sung Y-Y. A study on the shear-failure of circular sectioned bridge
column retrofitted by using CFRP jacketing. J Reinf Plast Compos
2004;23:811–30.
[55] Sheikh SA, Yau G. Seismic behavior of concrete columns confined with steel
and fiber-reinforced polymers. ACI Struct J 2002:99.
[56] Xiao Y, Wu H, Martin G. Prefabricated composite jacketing of RC columns for
enhanced shear strength. J Struct Eng 1999;125:255–64.
[57] Mirmiran A, Shahawy M, Samaan M, Echary HE, Mastrapa JC, Pico O. Effect of
column parameters on FRP-confined concrete. J Compos Constr
1998;2:175–85.
[58] Lam L, Teng J. Strength models for fiber-reinforced plastic-confined concrete. J
Struct Eng 2002;128:612–23.
[59] ACI-440. Guide for the design and construction of externally bonded FRP
systems for strengthening concrete structures. Reported by ACI Committee;
2008. p. 440.
[60] Yazici V, Hadi MN. Normalized confinement stiffness approach for modeling
FRP-confined concrete. J Compos Constr 2012;16:520–8.
[61] Sheikh SA, Li Y. Design of FRP confinement for square concrete columns. Eng
Struct 2007;29:1074–83.