Accepted Manuscript

Experiments and Finite Element Analysis of GFRP Reinforced

Geopolymer Concrete Rectangular Columns Subjected to
Concentric and Eccentric Axial Loading

Mohamed Elchalakani, Ali Karrech, Minhao Dong, M.S.

Mohamed Ali, Bo Yang

PII: S2352-0124(18)30042-0
DOI: doi:10.1016/j.istruc.2018.04.001
Reference: ISTRUC 273
To appear in: Structures
Received date: 10 February 2018
Revised date: 3 April 2018
Accepted date: 5 April 2018

Please cite this article as: Mohamed Elchalakani, Ali Karrech, Minhao Dong, M.S.
Mohamed Ali, Bo Yang , Experiments and Finite Element Analysis of GFRP Reinforced
Geopolymer Concrete Rectangular Columns Subjected to Concentric and Eccentric Axial
Loading. The address for the corresponding author was captured as affiliation for all
authors. Please check if appropriate. Istruc(2017), doi:10.1016/j.istruc.2018.04.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
 ACCEPTED MANUSCRIPT

Experiments and Finite Element Analysis of GFRP Reinforced Geopolymer Concrete

Rectangular Columns Subjected to Concentric and Eccentric Axial Loading

Mohamed Elchalakani1*, Ali Karrech2, Minhao Dong3, M.S. Mohamed Ali4, Bo Yang5

1
Senior Lecturer, School of Civil, Environmental and Mining Engineering, University of Western Australia, WA 6001, Australia
2
Associate Professor, School of Civil, Environmental and Mining Engineering, University of Western Australia, WA 6001, Australia
3
Ph.D. Student, School of Civil, Environmental and Mining Engineering, University of Western Australia, WA 6001, Australia
4
Senior Lecturer, Department of Civil Engineering, The University of Adelaide, SA 5000, Australia

PT
5Chongqing University, Ministry of Education, Chongqing 400045, China

RI
*
Corresponding author: Email: mohamed.elchchalakani@uwa.edu.au

ABSTRACT

SC
Despite a number of recent studies illustrating the effective use of Glass Fibre Reinforced
NU
Polymer (GFRP) rebars as longitudinal reinforcement for concrete compression members.

There is a lack of sufficient experimental and numerical studies on their combined action
MA

response. This paper demonstrates the use of finite element analysis to predict the response of

GFRP reinforced geopolymer and ordinary Portland cement concrete columns under concentric

and eccentric axial load in light of experimental data used for validation. The commercial
D

numerical analysis software ABAQUS was used to carry out a parametric study using different
E

material models, element types and mesh sizes. The concrete was simulated using reduced
PT

integration 8-noded hexahedral elements, C3D8R. The elasto-plastic material behaviour and
CE

the post-peak degradation in strength and stiffness of the concrete was modelled using the

widely used concrete damage plasticity model available in ABAQUS. As for the GFRP bars
AC

and stirrups, a linear elastic behaviour was considered. To validate the model, the measured

load-deflection responses were compared with the predicted curves. It was found that the

experimental curves were in close agreement with the predicted load–displacement responses.

The predicted N-M strength interaction diagrams matched the measured curves from the

experiment, particularly for GFRP RC columns with a large spacing between the ligatures

because they exhibited ductile failure which was well captured by the numerical model.
Keywords: ABAQUS; GFRP; plastic theory; concrete damaged plasticity model.

HIGHLIGHTS

1
 ACCEPTED MANUSCRIPT

1. Geopolymer and OPC concrete structural members were tested under combined loadings

2. Continuum damage mechanics resulted in good agreement with experiment data

3. Peak load and corresponding deflection were predicted at a satisfactory accuracy

4. Premature failures could be avoided by using strong closed stirrups

PT
NOTATION

RI
As Area of tension steel

SC
Asp Area of the stirrups

b Width of cross section of column

NU
bc Width of the concrete core
MA

c Cover to the reinforcement

d Depth of cross section of column

D

dc Depth of the concrete core

E
PT

ds Effective depth to compression rebar

CE

Es Elastic modulus of steel

Ec Elastic modulus of concrete

AC

fsy Yield strength of steel

e Eccentricity

Ef Elastic modulus of the GFRP

f c, R Unconfined compressive strength of N32 concrete at 28 days

2
 ACCEPTED MANUSCRIPT

f cc ,R Confined compressive strength of N32 concrete at 28 days

f fu Rupture stress in GFRP rebar

h Depth of the columns

L Height of the column

Le Effective height of the column

PT
Mu Bending strength

RI
Nu Axial compressive strength (kN)

SC
c Axial strain in concrete

 pl Equivalent plastic strain

NU
 in Inelastic strain
MA

c Proportion of plastic strain

st Stress in tension steel

D

sc Stress in compression steel

E
PT

s Clear vertical spacing between the hoop bars,

wi' i th clear distance between adjacent longitudinal bars

CE
AC

3
 ACCEPTED MANUSCRIPT

1. INTRODUCTION

1.1. General

Over the last two decades, engineers have increasingly looked to GFRP composite materials

as a means of overcoming corrosion and other problems commonly associated with steel

reinforcements. The use of Glass Fibre-Reinforced Polymer (GFRP) reinforcement as an

PT
alternative to steel for use in Reinforced Concrete (RC) structures has recently developed

RI
significantly. GFRP’s excellent corrosion resistance, high tensile-strength-to-weight ratio, non-

SC
magnetic, and non-conductive make it an excellent solution for projects requiring improved

corrosion resistance or reduced maintenance costs. However, GFRP bars have low elastic
NU
modulus and tendency to rupture without yielding means they require different treatment in the
MA

design codes to steel reinforcement. Thus, it is necessary to carry out experimental and

numerical studies on the mechanical properties of CFRP strengthened reinforced concrete

D

structures to better predict such failure modes. GFRP reinforced concrete members are
E
PT

designed to the international design codes such as [1–5]. There are currently insufficient studies
CE

on their combined action response. A number of these studies are reviewed in the following

sections.
AC

The N-M interaction strength diagram is a graph demonstrates the axial loading capacity of

structural members subjected to bending moment, under different eccentricities. The current

design rules noted above for GFRP reinforced concrete structures do not provide design

guidelines for GFRP RC Columns subjected to axial force and moment. In the past, only few

studies provided design guidelines for GFRP RC columns subjected to axial load and moment

4
 ACCEPTED MANUSCRIPT

[6–8]. In this paper, N-M interaction diagrams suitable for design purposes are provided after

validation with test results.

1.2 Geopolymer

Geopolymer concrete (GPC) utilises a binder that consists of inorganic alumino-silicate

polymers synthesized from the alkaline activation of various source materials rich in alumina-

PT
silicates. These materials are usually of geological origin or are by-product materials of other

RI
processes, such as fly ash, silica fume, red mud and blast furnace slag [9]. It is considered to

SC
be a green alternative to the ordinary Portland cement (OPC) concrete due to the utilisation of

the waste material and no calcination process involved in the production of its binder materials.
NU
The behaviour of fly ash based Geopolymer concrete under compression is similar to that of
MA

OPC concrete. However, the measured values of the modulus of elasticity of fly-ash based

geopolymer concrete is different to conventional concrete [10]. GFRP confined geopolymer

D

concrete also has similar shaped stress-strain curves to those for GFRP confined high strength
E
PT

OPC concrete but existing models must be modified [11].

CE

1.3 Past Experimental Researches

In regards to experimental studies, a good deal of laboratory tests were performed to measure
AC

the response of GFRP reinforced RC columns and beam-columns [6,7,12–20]. The main

conclusions from these studies are: (a) the bond behaviour of GFRP in GPC concrete specimens

was similar to that in OPC concrete specimens; (b) for the design of GFRP-RC columns, steel

rebars cannot be simply replaced with GFRP rebars because of the differences in the

mechanical properties of the steel and GFRP materials; (c) columns reinforced with steel rebars

provided greater load carrying capacity than columns reinforced with an equal amount of GFRP

5
 ACCEPTED MANUSCRIPT

rebars. The longitudinal steel rebars contributed an extra strength of between 10% to 50% that

of the GFRP under concentric loading; (d) GFRP RC columns were found to be slightly more

ductile than steel RC columns. Furthermore, reducing the GFRP helix pitches improved the

load carrying capacity, bending moment, and ductility of GFRP RC columns; (e) the load

carrying and bending moment capacities of GFRP RC columns can be calculated by the same

PT
principals as conventional steel RC columns; (f) the calculated failure loads were higher than

RI
the measured failure loads; (g) ignoring the contributions of GFRP rebars under compression

SC
loading could lead to gross discrepancies between analytical capacity calculations and

experimental results. Theoretical capacities better represent experimental data when the
NU
strength and stiffness of GFRP reinforcement are included; (h) columns reinforced with fibre-
MA

reinforced polymer (FRP) rebars are more susceptible to length effect than their steel RC

column counterparts because of the lower modulus of elasticity of FRP rebars. Thus a reduction
D

in members’ slenderness limits was suggested, by 5% for AFRP, 15% for CFRP, and 22% for
E
PT

GFRP rebars.
CE

Although the above tests on GFRP RC columns covered a wide range of parameters, they have

a number of shortcomings: (a) circular columns received little attention such as (b) have not
AC

examined different corrosion free materials for shear reinforcement such as stainless steel

which are widely used in infrastructure because of the difficulty in bending GFRP bars. Recent

tests by the authors have shown that rectangular shaped GFRP reinforced concrete columns

subjected to eccentric axial loading are more ductile than their steel reinforced counterparts if

they have adequate shear reinforcement with volumetric ratio of the order of 3.5% [12].

1.4 Past Numerical Researches

6
 ACCEPTED MANUSCRIPT

In regards to numerical studies, a limited number of finite element analyses were carried out to

simulate the behaviour of GFRP RC columns and beams [21–30]. Venkatasubramani et al. [26]

presented a finite element model using the commercial software ANSYS that presents the

flexural response of steel reinforced Geopolymer concrete beams. This study found differences

in the value of experimental load-deflection and finite element load-deflection for both

PT
reinforcement ratios of 0.87% and 1.75% of beams. An explanation to why this issue came

RI
about was explained by the meshing techniques used. This paper claims that the discrepancies

SC
in the values can be minimized by modifying the size of elements in meshing.

Nguyen et al. [27] constructed a 3D finite element model in ABAQUS to simulate a four-point
NU
bending test of a Geopolymer concrete beam with steel reinforcement. The finite element
MA

analysis (FEA) models and experimental models produced very similar load-displacement

curves but in the linear elastic range the finite element (FE) model was found to be much stiffer.
D

Amiri et al. [25] presented a model to present the flexural response of a reinforced fly-ash based
E
PT

Geopolymer concrete beam using the commercial FEA software ABAQUS. This paper found
CE

differences in the deflections of the beam when compared to experimental results but found

good agreement between the crack patterns.

AC

Three-dimensional (3D) FEA often involved the modified concrete damage plasticity (CDP)

model in ABAQUS to model the degradation in post peak strength and stiffness of the concrete.

Hany et al. [21] developed a new set of strain hardening/softening constitutive relationships to

model the actively confined concrete and FRP-confined concrete. A new concrete dilation

model was also proposed. The FEA showed good agreements with the test data. Ibrahim et al.

[22] modelled the bond-slip behaviours of basalt FRP bars embedded in concrete using 3D

7
 ACCEPTED MANUSCRIPT

numerical finite element modelling. It was recommended to include the bond-slip behaviour to

avoid inaccuracies in the prediction in the post-yield stages. Teng et al. [24] developed a three-

dimensional approach based on the plastic-damage model proposed by Yu et al. [31] to model

the behaviour of FRP-confined cylinders and columns. The non-uniform deformation in the

axial direction was modelled at a satisfactory accuracy by introducing a local stress-strain

PT
model. Piscesa et al. [23] proposed a flow rule that automatically adjusts the dilation rate of the

RI
concrete based on the lateral modulus EL. The model implemented in a 3D FEA was capable

SC
of simulating the regions that elastically unload, the yielding of the steel rebars, etc. A new

model was recently developed to predict concrete damage evolution to avoid mesh sensitivity
NU
for steel RC frames under cyclic loading [28]. Youssf et al. [29] presented a simple empirical
MA

model predicting the concrete dilation parameter based on the experimental program involving

105 FRP-confined specimens. The model was successful in capturing the confined ultimate
D

strength and axial strains of the specimens. Elchalakani et al. [30] incorporated damaged
E
PT

plasticity along with compression hardening and tension stiffening of the concrete filled tubes
CE

in the FE modelling. A good agreement was found between the FE results and the experiment

results.
AC

1.5 Scope

The above review highlighted the lack of experimental studies as well as finite element

analyses on GFRP-RC columns and beam columns. Therefore, numerical simulations verified

using laboratory tests are needed to understand and provide new design rules for GFRP

reinforced RC columns and beam-columns. Specifically, parametric analysis of ABAQUS was

first carried out to examine the accuracy with the most influential parameters, such as element

8
 ACCEPTED MANUSCRIPT

types and sizes, concrete confinement, as well as constitutive material models. The

experimental program involved testing 11 GFRP-reinforced fly ash and ground granulated

blast furnace slag (GGBS) based GPC specimens, of which 9 rectangular concrete were

columns and 2 were concrete beams. The results were compared against the finding of 7 OPC

concrete columns and 2 OPC concrete beams in an experimental study by Elchalakani et al.

PT
[12]. This paper provides a brief summary of the test results and failure modes the specimens

RI
reinforced with GFRP bars. Also a comparison between the N-M interaction diagram resulted

SC
from the tests and the ones resulted from the finite element analysis is presented at the end of

the paper.
NU
2. EXPERIMENTAL WORK
MA

2.1 Test Program

The full test program was based on a recent relevant paper [12], where all OPC concrete
D

specimens shared the same rectangular cross-section with 260mm x 160mm, and they were
E
PT

1200mm long. The top and bottom 225 mm sections of the columns were strengthened by
CE

reducing the spacing of the ligatures to 50 mm to decrease the possibility of shear failure close

to the ends. The GPC specimens in this study adopted the same arrangement but were enlarged
AC

at the ends to reduce end effects. Fig. 1a and 1b shows the schematics of the concentrically

load GPC columns and the eccentrically loaded GPC columns with enlarged ends, respectively.

Fig. 1c shows the GPC beam reinforcement configuration. The same arrangement shown in

Fig. 1c was applied to all the OPC concrete columns and beams, with the differences being

their corresponding stirrup spacings. The unconfined compression tests on 100mm diameter

200mm high concrete cylinders showed that the GPC has a 28-day compressive strength f’c of

9
 ACCEPTED MANUSCRIPT

26.0 MPa whereas the f’c of OPC concrete was 32.8 MPa [12]. The longitudinal reinforcement

consisted of three 14mm GFRP rebars on top for tension and other three 14mm GFRP rebars

for compression. Ligatures were sufficiently provided and consisted of 8mm GFRP rebar. The

physical properties of GFRP bars are demonstrated in Table 1. The arrangement of the GFRP

reinforcement of the GPC and OPC concrete specimens is shown in Table 2. The closed stirrups

PT
used in this study were lapped to delay the failure of the stirrups. The sixteen specimens were

RI
all tested until failure where the GFRP bars demonstrated a significant ductility when

SC
embedded in concrete [12]. Fig. 2 shows the test setup for the specimens in this study. Fig. 2a

shows the typical setup for eccentrically loaded GFRP reinforced concrete columns. The
NU
specimens were loaded against their minor axes and the eccentricity was controlled by the
MA

loading pin welded to the steel plate. A similar setup was adopted for the concentrically loaded

columns, except no loading pin was used. All the beams were loaded with a four-point bending
D

configuration (Fig. 2b) with a span of 1100 mm, and distance between loads of 366 mm. The
E
PT

beams were tested where the bending was performed about the minor axis. Thus, the
CE

beam/column width b=260 mm (contains 3 bars) and the depth d=160mm.

2.2 Failure modes

AC

The failure mode of each specimen depended on its reinforcement material, reinforcement

arrangement, and loading condition. In general, the failure of concentrically loaded columns

with small volumetric ratio of transverse reinforcement (ie. 150 mm spacing) was controlled

by longitudinal bar buckling. Conversely, well-confined concentrically loaded GFRP RC

columns (ie. 75 mm spacing) failed by transverse reinforcement rupture and crushing of the

concrete core (see Fig. 3). Transverse reinforcement spacing was found to have a greater effect

10
 ACCEPTED MANUSCRIPT

on ductility and confinement than ultimate strength. Columns with smaller, tighter spaced

transverse reinforcement (ie. 75 mm spacing) were more ductile and produced less brittle

failures than those with larger, more widely spaced transverse reinforcement (i.e. 150 mm

spacing).

For the concentrically loaded columns, the well-confined G75-C and O75-C failed due to large

PT
deformation of the ligatures resulted from excessive hoop tension forces, and spalling of the

RI
concrete, followed by buckling and rupture of the longitudinal rebars and finally crushing of

SC
the concrete core. Stirrup opening was seen for the G75-C columns (Fig. 3a) and ligature

rupture was evident in O75-C with tight spacing between the ligatures (Fig. 3b). G150-C,
NU
G250-C, O150-C and O250-C failed by longitudinal bar buckling followed by explosive
MA

concrete core crushing. The concentrically loaded columns had an increasingly more core

concrete remaining intact as the stirrup spacing decreased.

D

Failure of the eccentrically loaded specimens was due to concrete crushing at the compression
E
PT

face. The GFRP RC columns exhibited bending deformation near the end of the test region.
CE

For the GPC columns loaded at an eccentricity (Fig. 4), heavy spalling was observed before

failure in the columns with low eccentricity of 25mm, i.e. G150-25 and G75-25. Due to the
AC

opening-up of the stirrups, the load capacity of the specimens decreased steadily. At medium

eccentricity of 50mm, the deformation in the well confined G75-50 was uniform throughout

the test region. The column failed by concrete crushing in the compression face. Similarly,

residual strength was observed for the specimens, unlike the concentrically loaded columns. A

plastic hinge was formed near the top end cap in G150-50. The column collapsed due to the

slipping failure of the GFRP bars near the top. Excessive bending of the bars was observed.

11
 ACCEPTED MANUSCRIPT

The failure mode of G150-75 and G75-75 loaded at large eccentricity was similar to that of

G150-50. For the eccentrically loaded OPC concrete columns (Fig. 5), the failure of O150-25

involved more global bending deformation than O75-25. Tension cracking was observed on

the tension face of each column. O150-45 was tested at 45 mm eccentricity where failure was

due to concrete crushing near the base of the compression face, causing the specimens to

PT
become unbalanced and slip off the steel loading plate. The eccentricity was therefore reduced

RI
to 35 mm for O75 specimen. O75-35 remained in position and buckled without slipping of the

SC
steel plate. Concrete crushing failure was still observed for O75-35 near the top end.

The flexural failure modes included bending, concrete crushing and shear (see Fig. 6). A two-
NU
peak failure mode was observed for the GPC beams due to concrete crushing in the
MA

compression face and the critical shear cracking initiated from the tension face. The flexural

specimen G80-F failed due to both shear and concrete crushing. Tensile cracking on the G40-
D

F tension faces indicates partial bending failure, however concrete crushing in the compression
E
PT

area was also dominant. The GFRP stirrups opened upon failure and significant bending
CE

occurred in the longitudinal bars at the diagonal cracks. Minor cracking was also observed

around the support rollers. The failure of the OPC concrete beams were also in two separate
AC

stages, similar to the GPC beams.

2.3 Column Strength

The experimental results for the 16 column specimens tested under concentric and eccentric

loading, and the 4 beams tested in flexure from this study and Elchalakani et al. [12] are shown

in Table 3. The first letter in the specimen ID denoted the type of concrete, with “G”

representing GPC and “O” representing the OPC concrete. The number after the initial letter

12
 ACCEPTED MANUSCRIPT

denoted the stirrup spacing. The third letter/number after the dash denoted the test method,

with “C” representing concentric loading, a number representing the load eccentricity in

eccentric loading and “F” representing flexural loading in the four-point bending test. SGL and

SGR are the maximum left and right strains measured at the peak load at two opposite

reinforcement corner bars. The GFRP RC columns exhibited ultimate capacities below that of

PT
the steel RC columns in most cases. Specimen G250-C produced erroneous strain gauge values

RI
and exhibited abnormally high strength. This is probably due to concrete vibration and

SC
compaction variations [12]. The ultimate compression capacities of the plain GPC and OPC

concrete section were 1082 kN and 1362.4 kN, respectively. From the ultimate loads presented
NU
in Table 3, the GFRP-OPC concrete columns exhibited 3.2% average increase in the load
MA

carrying capacity, with respect to the plain concrete section capacity, whereas the steel-OPC

concrete columns achieved an average enhancement of 15.8% [12]. The average axial load
D

carrying capacity of GFRP RC columns was 93.5% of their steel RC columns counterparts
E
PT

[12]. The concentrically loaded GPC columns showed an average increment of 10.8%,
CE

positioned in between the GFRP-OPC concrete columns and steel-OPC concrete columns. In

construction projects such as the coastal zones in Australia where steel reinforcement was not
AC

suitable, the use of GFPR-GPC systems would be more favourable than the GFRP-OPC

concrete systems.

3. CONSTITUTIVE MODELS

The prediction of concrete column behaviour reinforced with GFRP bars is one of the

challenging research areas in particular the selection of constitute models. The commercial

numerical analysis tool ABAQUS 6.14 version was used to carry out sensitivity tests to

13
 ACCEPTED MANUSCRIPT

examine different material models. To conduct the current analysis, we considered that

concrete exhibits a linear elastic behaviour in the reversible regime and a damaged plasticity

behaviour in the irreversible regime.

3.1. Properties in the reversible regime

Due to the differences in the mechanical properties between GPC and OPC concrete, different

PT
equations were used to model the different responses of the specimens. In the linear elastic

RI
regime, the Young’s modulus of the two types of concrete was determined using Eq. 1 from

SC
Hardjito et al. [32] and Eq. 2 from ACI 363R-92 [33].

𝐸𝐺𝑃𝐶 = 2707√𝑓′𝑐𝑜 + 5300 (1)

NU
𝐸𝑂𝑃𝐶 𝑐𝑜𝑛𝑐𝑟𝑒𝑡𝑒 = 3320√𝑓′𝑐𝑜 + 6900 (2)
MA

Where f’co is the 28-day unconfined cylinder compressive strength of the concrete. The
D

Poisson's ratio was taken at 0.15 for GPC [32] and 0.2 for the OPC concrete [34].
E

3.2 Properties in the irreversible regime

PT

The concrete damaged plasticity is commonly accepted to model concrete reinforced structures
CE

in the irreversible or dissipative regime based on the aforementioned literature in the

Introduction section [22,23,28–30]. Damaged plasticity includes five behavioural components,

AC

which are: (1) plasticity; (2) compressive behaviour; (3) confinement; (4) tensile behaviour;

and (5) damage mechanics. These components are individually discussed in the following

subsections.

3.2.1 Plasticity Model

The concrete plasticity model requires a dilation angle  , shape factor for yield surface K C

and a ratio of initial equi-biaxial compressive yield stress to initial uniaxial compressive yield

14
 ACCEPTED MANUSCRIPT

stress ( f b 0 / f c' ). The quantities were selected in the current study as follows [30,35–37] for

both types of concretes:

  40o (3)

fb0
'
 1.5( f ' ) 0.075 (4)
fc

PT
5.5
KC  (5)
5（
2 f c'）0.075

RI
The above choices for the parameters included in the plasticity model are controlled by the

SC
compressive and confinement behaviours as detailed in the following sections.

3.2.2 Compressive behaviour

NU
Customised uniaxial compression strain-stress relationships can be introduced in ABAQUS by
MA

the user. Thus, two main aspects will be investigated in this study, the adjusted concrete

constitutive model and the confinement under uniform constraints. The uniaxial compressive
D

stress–strain model established by Dong et al. [38] in Eq. 6 was used for GPC.
E

0.65𝐸𝑜 𝜀
PT

𝜎𝑐 = 𝜀 4 3 (6)
(1+0.091( ) )
𝜀𝑐𝑜

Where 𝜎𝑐 represents the compressive stress, 𝜀 represents the compressive strain and 𝜀𝑐𝑜 is
CE

the compressive strain at peak load, which is defined in Eq. 7 by Nath and Sarker [39].
AC

𝑓′ n
𝜀co = 𝐸 𝑐 . n−1 (7)
𝑜

Where

𝑓′𝑐
n = 0.8 + (8)
12

Eq. 9 proposed by Desayi and Krishnan [40] was used to model the stress-strain behaviours of

the OPC concrete.

𝐸𝑜 𝜀
𝜎𝑐 = 𝜀 2 (9)
1+( )
𝜀𝑐𝑜

15
 ACCEPTED MANUSCRIPT

Where 𝜀𝑐𝑜 was taken as 0.002 as per the recommendation by Mander et al. [41].

As illustrated in Fig. 7 from Liu et al. [42], it is necessary to assign inelastic strain  in in further

degree to fit the ABAQUS settings. This is because the maximum elastic stress of concrete

ought to be assigned at the first row of compression part in concrete damaged plasticity model.

Thus, all initial strain  c were transformed as inelastic strain  in to demonstrate inelastic

PT
shaping as explained in the following sections.

RI
3.2.3 Confining Models for Concrete

SC
The confining models of Kappos and Konstantinidis [43] and Mander et al. [41] were

introduced to analyse confining effects in light of which the sensibility tests of ABAQUS will
NU
be discussed. Kappos and Konstantinidis [43] model describing the post-peak stress-strain
MA

curves was used to calculate the enhancement in concrete strength due to confinement

especially for stirrups.

D

f cc  0.85 * f c  10.3 *（ * h * f yh）

0.4
(10)
E
PT

 n 2
  Ci  s s
  1 - i 1 (1  )(1  ) (11)
CE

 6 bc d c  2b c 2 d c
 

where f cc is the confined concrete strength, f c is the unconfined concrete strength from the
AC

cylinder [44], the volumetric ratio of the stirrups arrangement shown in Fig. 1 can be

approximated as

2( bc  2dc )Asp
h  (12)
bc dc s

where s is the clear vertical spacing between the hoop bars, bc and dc are the width and depth

of confined concrete core defined by the centreline of the ties, Asp is the cross sectional area of

16
 ACCEPTED MANUSCRIPT

shear reinforcement, Ci is the centre-to-centre distance between laterally supported longitudinal

'
bars, f yh is the tensile strength of GFRP ties and it was determined according to ACI 440R-

15 [1] as

 r 
f yh  f fu  0.05 b  0.3   f fu (13)
 db 

PT
where rb and db are the radius and diameter of the hoop bars, respectively. The remaining

parameters in Eqs. 8 and 9 are defined in Kappos and Konstantinidis [43].

RI
On the other hand, the model introduced by Mander et al. [41] can also be used to describe the

SC
confining effects of steel and FRP stirrups.
NU
 7.94 fl f 
f cc  f c  1.254  2.254 1   2 l (14)
 fc fc 

MA

Where f cc is the maximum stress under lateral stress f l  ; f c is the unconfined concrete

strength from the cylinder test to AS1012-9:1991 [44]:

D

1
f l  ke  s f yh (15)
E

2
PT

n
( wi' ) 2 s' s'
(1   ) (1  )(1  )
i 1 2d s 2bc 2d c
ke  (16)
(1   cc )
CE

6 Abar
cc  (17)
AC

bc dc

 fcc 
 cc   co  1  5(  1)  (18)
 f c 

Where co and cc are the unconfined and confined strains, s is clear vertical spacing between

the hoop bars, wi' is the i th clear distance between adjacent longitudinal bars, bc and dc are the

confined width and depth of concrete, Abar is the area of the longitudinal reinforcement bar, s

is the centre-to-centre spacing between hoops, Asp is the cross sectional area of shear

17
 ACCEPTED MANUSCRIPT

reinforcement, s is the volumetric ratio of the ties given in Eq. 12; Acc= bc.dc is the area of the

'
confined core concrete section enclosed by the centreline of stirrups. f yh is the tensile

strength of GFRP ties and it was given in Eq. 13.

The comparison between Kappos and Konstantinidis [43] confined concrete model (Eq. 10 and

Eq. 18) and Mander et al. [41] confined concrete model (Eq. 16 and Eq. 18) and the initial

PT
unconfined concrete model (Eq. 9) for the OPC concrete specimen O75-C (s=75mm) is shown

RI
in Fig. 8. It is observed that the peak stress given by Eq. 14 is considerably large when

SC
compared to that given by Eq. 8.

3.2.4 Tensile behaviour

NU
To describe the tensile behaviour of concrete, two options were considered, namely the stress-
MA

strain response and mode I fracture energy. Similar to the compressive behaviour, the stress-

strain response in tension can be modelled with an elastic and plastic region. The tensile yield
D

stress was expressed as shown in Eq. 19 [32] for GPC and in Eq. 20 [34] for the OPC concrete.
E
PT

𝑓𝑐𝑡_𝐺𝑃𝐶 = 0.7√𝑓′𝑐𝑜 (19)

𝑓𝑐𝑡_𝑂𝑃𝐶 𝑐𝑜𝑛𝑐𝑟𝑒𝑡𝑒 = 0.4√𝑓′𝑐𝑜

CE

(20)

The tensile stress 𝑓𝑡 was defined using the model proposed by Stoner [45] as shown in Eq. 21.
AC

𝐸𝑐𝑜 𝜀𝑡 , when 𝜀𝑡 ≤ 𝜀𝑡𝑜

𝑓𝑡 = { 𝑓 (𝜀to )0.4 , when 𝜀 < 𝜀 (21)
ct 𝜀𝑡 𝑡𝑜 𝑡

Where 𝜀𝑡 is the tensile strain and 𝜀to is the tensile strain at peak load.

The mode I fracture energy Gf was first proposed by Hillerborg, Modeer and Petersson [46] as

a material property to overcome the mesh sensitivity. It represents the amount of energy Gf

required to open a unit area of crack. It is assumed to be 79J/m2 for a linear loss of strength

18
 ACCEPTED MANUSCRIPT

after cracking. The fracture energy approach was initially adopted in the FE models for its

simplicity.

3.2.5 Damage Mechanics Theory

There are several constitutive models that can be used to describe the behavior of concrete

based on the theories of elasticity, plasticity, fracture, continuum damage mechanics, and

PT
combinations of the above theories. In particular, the theories of continuum damage mechanics

RI
[30,37,47] takes both the existence of cracks in initial period without loads and the expansion

SC
of cracks during loading and thus trigger the degradation of material strength. The relationships

between compression stress, strain, inelastic strain as well as plastic strain are shown in Fig. 7.
NU
The damage parameter d c can be expressed as follow [42].
MA

(1−𝜂 )𝜀𝑐𝑖𝑛 𝐸0
𝑐
𝑑𝑐 = 𝜎 +(1−𝜂 𝑖𝑛 (22)
𝑐 𝑐 )𝜀𝑐 𝐸0

where  c is the proportion of the plastic strain  pl in the inelastic strain  in , and it can be
D

taken as 0.7 according to previous laboratory tests. The relationship between elastic and plastic
E
PT

strains in hardening regime can also be expressed by Liu et al. [42]:

c
CE

 in   c  (23a)
E0

dc  c
AC

 pl   in  (23b)
(1  d c ) E0

Similarly, the damage parameter under tensile loading 𝑑𝑡 is governed by Eq. 24.

𝑡𝜂 𝜀 in 𝐸𝑜
𝑑𝑡 = 𝜎 +𝜂 in 𝐸
(24)
𝑐 𝑡𝜀 𝑜

3.2.6 GFRP Bars Properties

The stress-strain relationship for the GFRP bars were assumed isotropic linear elastic [22] as

illustrated in Fig. 9. No damage criterion was applied to the GFRP bars. Their properties were

19
 ACCEPTED MANUSCRIPT

dependent on the loading mode (tensile or compressive loading). In accordance with previous

test measurements [6,7,13], we considered a Young’s modulus of 50 GPa in tension and 50

GPa in compression and a tensile strength of 1200 MPa (in tension) and 600 MPa in

compression. In both tension and compression regimes, a Poisson’s ratio of 0.25 was

considered (Table 1).

PT
4. FINITE ELEMENT MODEL

RI
4.1 General

SC
The geometry of the FE models (Fig. 10) was designed according to the detailed drawings

shown in Fig. 1. The concrete of the specimens was modelled as a homogenous three-
NU
dimensional solid section. Three-dimensional wire elements were used to simulate the
MA

longitudinal and transverse reinforcement. The six longitudinal bars were defined with a cross-

sectional area of 154 mm2 each and a length of 1.17 m. The ligatures were defined with cross-
D

sectional areas of 50 mm2 and designed such that they had 20 mm cover. The plain concrete
E
PT

elastic-plastic behaviour shown in Fig. 7 was based on the well-established damage plasticity
CE

model by Liu et al. [39]. GFRP bars and stirrups were simulated with linear elastic material up

to failure without damage criterion applied to it, as illustrated in the added Fig. 9. The GFRP
AC

bars and stirrups were embedded in the concrete sections in ABAQUS. Therefore, the interface

between the concrete and GFRP bars was not simulated. The two materials were governed by

their own material models. To simulate the interaction between the reinforcement and the

concrete the ABAQUS built-in constraint “embedment” was used. This constraint restricts the

nodes of the reinforcement to the corresponding degrees of freedom of the host domain. The

20
 ACCEPTED MANUSCRIPT

concentric columns were partitioned based on the confinement model by Kappos and

Konstantinidis [43] to define the regions of unconfined and confined concrete.

The concrete element type for all specimens was defined as three dimensional eight-noded

deformable brick elements (C3D8R) with reduced integration (Fig. 11a). Previous studies

[25,27] have shown that these elements follow the constitutive law of integration very

PT
accurately and are suitable for nonlinear static and dynamic analyses. The reinforcement

RI
element type was defined as deformable two-noded truss elements (T3D2R) (Fig. 11b). The

SC
basis of this decision is that truss elements provide only axial stiffness, which is reflective of
NU
GFRP’s weak strength perpendicular to its grain direction. In this paper, all the Geopolymer

concrete specimens were meshed with an approximate size of 40mm in the longitudinal,
MA

transverse and thickness directions.

D

Induced displacements are used to test the capacity of the FE models. For the concentric
E

columns, the base of the column is fixed in the U3 direction and a displacement of -0.015m is
PT

applied to the top of the specimens. For the eccentric columns, the load and restraints are
CE

provided by external rigid steel plates tied to the top and bottom surfaces. This enables the

models to distribute the loading equally over the surface the loads are applied to. The bottom
AC

plate is fixed in the U3 direction and a displacement of -0.015m is applied to the top plate. For

each eccentric column, the dimensions of the plate are changed in order to provide the required

load eccentricity. To replicate the four-point bending test for the beams, the model was

supported by two linear partitions spaced 1.1m apart and fixed in the U3 direction. The applied

loading was a displacement of 50mm applied on two symmetric lines spaced 0.366m apart and

symmetric with the centre of the specimen.

21
 ACCEPTED MANUSCRIPT

4.2 Convergence and validation of the model

To ensure that the models produce the desirable accuracy, we conducted a convergence study

by varying the size of the elements from relatively coarse to excessively refined meshes. Fig.

12 depicts the results obtained for meshes with element sizes ranging from 0.1m to 0.04m for

G250-C. The results show that 0.04m elements offer an acceptable convergence.

PT
4.3 Confining effect

RI
The predictions of Kappos and Konstantinidis [43] and Mander et al. [41] were compared to

SC
the experimental data in Fig. 13. As shown in the figure, the model proposed by Mander et al.

[41] resulted in inaccurate peak loading capacity and the displacement at peak load was
NU
overestimated significantly. Therefore, Kappos and Konstantinidis [43] model was considered
MA

in all the 3D models of all the GFRP RC columns and beams. In following simulations, the

modified plastic model is used, which combined Kappos and Konstantinidis [43] confined
D

concrete model (Eq. 10) and the equivalent plastic strain (Eq. 22).
E
PT

5. DISCUSSION OF THE RESULTS

CE

5.1 Peak load and deflection at peak load

The FEA results of the 20 GPC and OPC concrete specimens are shown in Table 3. Fig. 14
AC

shows the typical FEA results of the stresses in the GPC columns and beams. The stress

distributions for the OPC concrete specimens were similar to the GPC specimens. For the OPC

concrete specimens, the models achieved on average 8% difference in peak load between the

experimental and FE predictions and 11% in the deflection at peak load. However, the models

had a tendency of overestimating the peak load, which is especially seen for the concentrically

loaded columns (O250-C, O150-C and O75-C). The overestimation only occurred for the

22
 ACCEPTED MANUSCRIPT

eccentrically loaded column with 75mm stirrup spacing. The largest inaccuracy in the

simulation occurred for the OPC concrete beam O150-F with 15% difference from the

experimental result, indicating the premature failure caused by opening of the lapped stirrups.

This also resulted in the large discrepancy between the FE predicted displacement and the

experimental results.

PT
For the GPC specimens, an obvious trend was that as the eccentricity increased, the accuracy

RI
of the FE simulations decreased considerably. The models used in this study successfully

SC
captured the peak load and the corresponding deflection for the concentrically loaded columns

and the eccentrically loaded column with small stirrup spacing. The average inaccuracy for the
NU
5 specimens (G250-C, G150-C, G75-C, G150-25 and G75-25) was 5%, which corresponded
MA

to an 59% improvement in the prediction than the average prediction of the OPC concrete

specimens. The predicted peak load was off by up to 42% for G150-75. However, all the peak
D

loads from FEA favourably underpredicted the load capacity of the columns, except for G75-
E
PT

C and G250-C, making it safe from a design viewpoint. The experimental data of G80-F was
CE

lost, therefore an FEA was not performed for this specimen. The other GPC beam G40-F was

modelled in ABAQUS with the difference being 25% in the peak load and 6% in the deflection
AC

at peak load. This was due to the two-peak failure mode was not predicted by the FE model,

which resulted in a larger predicted peak load and a close deflection. The average difference

between the FE models and the experimental results for the GPC specimens was 15% despite

the large underestimation for the eccentrically loaded columns at large eccentricity. The

prediction in the deflection at peak load achieved a similar 11% to the OPC concrete specimens.

23
 ACCEPTED MANUSCRIPT

This showed the models used in this study produced satisfactory results. However, the

predictions for the GPC specimens were more favoured due to the safe underprediction.

5.2 The load-deflection relationships for the GPC specimens

5.2.1 Concentrically loaded GPC columns

The load-deflection curves of the concentrically loaded GPC columns with 75mm, 150mm and

PT
250mm stirrup spacings were shown in Fig. 15. It can be seen that the FE models are in close

RI
agreement with the behaviour of all of the three concentrically loaded specimens in the elastic

SC
range. The FEA results of G150-C accurately predicted the peak loading capacity of their

respective experiment counterparts, however, the FEA results of G75-C and G250-C predicted
NU
a slightly greater peak loading capacity than the experimental G75-C. Specimen G75-C with
MA

closely spaced stirrups had a more ductile failure while G150-C and G250-C experienced

explosive failures with no residual strengths. This was well captured in the FE models.
D

5.2.2 Eccentrically loaded GPC columns

E
PT

The load-deflection curves of the concentrically loaded GPC columns with 75mm and 150mm
CE

stirrup spacings at 25mm, 50mm and 75mm eccentricity are shown in Fig. 16a, b and c,

respectively. At 25mm eccentricity (Fig. 16a), the initial elastic range of both G150-25 and
AC

G75-25 was well modelled. The peak load of G150-25 was correctly identified at the correct

deflection. The FE load-deflection curves predicted the sudden collapse of the column.

However, the gradual loss of capacity after the collapse was not modelled by the FEA. At

50mm eccentricity (Fig. 16b), the initial elastic range of both G150-50 and G75-50 was again

well modelled. The peak loads of both specimens were favourably slightly underpredicted by

the FEA. The residual strengths of the specimens were identified in the model. However, the

24
 ACCEPTED MANUSCRIPT

experimental results suggested a lower residual value for the specimen with large stirrup

spacing. The opening of the stirrups could not be effectively modelled by the FEA, therefore

the significant loss in capacity after the stirrups opened in the specimens with large stirrup

spacings was not simulated accurately. At 75mm eccentricity (Fig. 16c), the elastic range of

G150-75 and G75-75 was not as accurately predicted as the specimens loaded at a smaller

PT
eccentricity. The peak loads, as suggested in Table 3, were more underestimated by the FEA

RI
than the low eccentricity counterparts. Similar to G150-50 and G75-50, the residual strengths

SC
of the column were overestimated by the FEA due to the different failure mechanisms of the

stirrups between the experiment and the FEA. The load capacity reduced significantly after the
NU
stirrups prematurely opened-up, resulting a much smaller residual strength than that predicted
MA

by the FE model. All the peak loads of the eccentrically loaded GPC columns were

underestimated by the FE model, while the deflections at the peak load were captured at a
D

satisfactory accuracy. This is likely due to the compressive stress-strain model of the GPC was
E
PT

intended for higher strength geopolymer, which was expected to have a more brittle failure
CE

mode and higher residual strength than the GPC used in this study.

5.2.2 GPC beams

AC

The load-deflection curve of G40-F is shown in Fig. 17. The two distinct peaks from G40-F

results obtained in the experiment were immediately seen. The first peak was modelled

reasonably well. The peak load and the deflection were accurately predicted by the model.

After the first peak, the beam experienced the first failure caused by concrete crushing in the

compression face, which was not modelled in the FEA. However, the load from the FE model

continued to increase at a slightly lower rate due to the damage induced by the first failure. The

25
 ACCEPTED MANUSCRIPT

deflection at the second peak was again successfully modelled by the FE model. However, a

25% discrepancy was found between the second peak load of the FEA and experimental results

as the first failure was not identified. The damage by the first failure was more critical in the

actual experiment, indicating a slightly underestimated GPC tensile damage relationship.

5.3 The load-deflection relationships for the OPC concrete specimens

PT
5.2.1 Concentrically loaded OPC concrete columns

RI
The load-deflection curves of the concentrically loaded OPC concrete columns with 75mm,

SC
150mm and 250mm stirrup spacings were shown in Fig. 18. The FE results were in close

agreement with the behaviour of the concentrically loaded OPC concrete columns, especially
NU
in the elastic range. However, the columns failed earlier than the FE predictions in the
MA

experiment, which resulted in overestimation by the FE models. Sudden collapses were seen

in O75-C and O150-C whereas O75-C experiences a more gradual loss of capacity. The higher
D

ductility of O75-C was successfully modelled in the FEA.

E
PT

5.2.2 Eccentrically loaded OPC concrete beams

CE

The load-deflection curves of the concentrically loaded OPC concrete columns with 75mm and

150mm stirrup spacings at low eccentricity (25mm) and high eccentricity (35mm and 45mm)
AC

are shown in Fig. 19a, and b, respectively. The elastic range of the columns was modelled with

high accuracy. However, both columns with 75mm stirrup spacing failed prematurely in terms

of the FEA due to the opening of the lapped stirrups. For the two columns with 150mm stirrup

spacing, the load sustained in the stirrups were lower, therefore the premature failure was not

observed. The elastic range, inelastic range, peak load and deflection at the peak load were

accurately predicted. The residual strengths of the columns O75-25, O150-25 and O75-35 were

26
 ACCEPTED MANUSCRIPT

not identified in the FE models. Similar to the GPC columns loaded at relatively small

eccentricities, the failure mode of those OPC concrete columns in the FE simulation was still

the sudden collapse of the entire column caused by the concrete crushing in the compression

face. It is shown that at smaller eccentricities, the ductility of both types of concrete columns

was not accurately represented. The simulation terminated at a point where excess deformation

PT
in an element of the FE model was detected. Therefore, it is recommended to use a finer mesh

RI
in the future. The model used in this study was still considered suitable as the elastic range,

SC
peak load and the deflection at peak load were simulated at a considerable accuracy.

5.2.3 OPC concrete beams

NU
The load-deflection curves of O75-F and O150-F are shown in Fig. 20. The FE results of O75-
MA

F were in acceptable agreement with the experimental results. Both the peak load and the

deflection at peak load were successfully modelled by the FE model. However, O150-F failed
D

prematurely at approximately half the expected deflection. The loosely spaced stirrups were
E
PT

unable to prevent any excess deformation in the GFRP bars once opened. Therefore, the beam
CE

collapses due to cracking and excessive bending of the longitudinal GFRP bars. The OPC

concrete beams were showing the same two-peak failure mode as the GPC beams. The FE
AC

model successfully captured the general trend of the two stages (before and after the first failure)

and the damage by the first failure.

5.4 Moment interaction diagrams

The moment interaction diagrams of the GPC and OPC concrete specimens with 75mm and

150mm stirrup spacings obtained from FEA and the experiments are shown in Fig. 21a and b,

respectively. It is seen that the GPC specimens were mostly favourably underpredicted by the

27
 ACCEPTED MANUSCRIPT

FE models. For the concentrically loaded GPC columns and the GPC columns loaded at a small

eccentricity (25mm), a satisfactory accuracy was achieved. However, due to the tendency to

open up under high strain of the lapped stirrups, the GPC columns loaded at higher eccentricity

(50mm and 75mm) failed prematurely when compared against the FE predictions. A high level

of accuracy was also achieved for the OPC concrete specimens. However, overestimation was

PT
often observed, especially for the specimens with 75mm stirrup spacing due to the failure in

RI
the stirrups. Therefore, the characteristic equations used to define the two types of concrete

SC
were more suitable for the analysis on the GPC specimens. Alternative models were needed to

ensure the analysis of OPC concrete specimens to be safe from the design viewpoint.
NU
7. CONCLUSIONS
MA

This paper demonstrated the use of finite element analysis validated with experimental data to

predict the response of GFRP reinforced GPC and OPC concrete columns under concentric and
D

eccentric axial loading. Both concretes were simulated using reduced integration 8-noded
E
PT

hexahedral elements, C3D8R and the GFRP bars was simulated using truss-in-solid elements
CE

type T3D2R. However, different models were used to distinguish the two types of concrete

with distinct mechanical behaviours. The following conclusions were drawn based on the
AC

experimental and finite element analysis:

Based on the experimental investigation, it was clearly shown that the specimens with smaller

stirrup spacings had larger areas of concrete intact and higher residual strengths after failure.

The stirrups opened or ruptured after the crushing failure of the concrete cover, initiating the

global failure of the entire specimen. The longitudinal GFRP bars ruptured in well-confined

specimens as opposed to the buckling failure in specimens with larger stirrup spacings. Slip

28
 ACCEPTED MANUSCRIPT

failure was also observed at the plastic hinge in the GPC columns with large stirrup spacing

loaded at high eccentricity. Critical diagonal failures occurred for all the beam specimens with

excessive longitudinal GFRP bar bending.

For the finite element analysis, it was found that an existing confinement model can be used to

accurately predict the elastic and plastic behaviours of the GPC and OPC concrete specimens.

PT
It was shown that the experimental load-deflection curves were closely predicted by utilising

RI
the selected concrete damage plasticity theory. The peak load and the deflection at peak load

SC
were successfully predicted at a satisfactory level of accuracy for both types of concrete

specimens (on average 15% and 11% for GPC specimens and 8% and 11% for OPC concrete
NU
specimens, respectively). The discrepancies mainly came from the premature failures due to
MA

the opening of the lapped stirrups. It is recommended to construct the specimens using stirrups

with the lapped distance being at least half the perimeter length to prevent such unfavourable
D

failure mode. The material characteristics models used for the GPC produced the favourable
E
PT

underestimations while overestimations were often observed for the OPC concrete models. It
CE

is concluded that the models for GPC specimens incorporating the concrete damage plasticity

theory and GFRP bar stress-strain response approach would be suitable in simulating the
AC

behaviours of the GPC columns and beams with sufficiently strong stirrups accurately and

producing safe designs.

ACKNOWLEDGEMENT

The contribution made by Airey Taylor Consulting was vital to the success of this project.

Thanks are given to Pultrall in Canada and V-Rod Australia for their help and support in

29
 ACCEPTED MANUSCRIPT

donating the GFRP material used for the project. Thanks are given to Mr Angus Lynn for their

help with the finite element modelling.

3. REFERENCE

[1] ACI 440. Guide for the design and construction of concrete reinforced with FRP bars

ACI 440.1R-15 2015.

PT
[2] CSA. Design and construction of building structures with fibre reinforced polymers,

RI
Canadian Standards Association 2012:CAN/CSA S806--12.

SC
[3] TR55. Design Guidance for Strengthening Concrete Structures Using Fibre Composite

Materials, 2012.
NU
[4] ISO. Fibre-reinforced polymer (FRP) reinforcement of concrete: Test methods: Part 1:
MA

FRP bars and grids 2015:10406--1:2015.

[5] FIB. FRP Reinforcement in RC Structures 2007:FIB Bulletin 40, Fib Task Group 9.3.
D

[6] Hadi M, Youssef J. Experimental investigation of GFRPreinforced and GFRP-encased

E
PT

square concrete specimens under axial and eccentric load, and four-point bending test. J
CE

Compos Constr 2016;20:4016020.

[7] Hadi M, Karim H, Sheikh N. Experimental investigations on circular concrete columns

AC

reinforced with GFRP bars and helices under different loading conditions. J Compos

Constr 2016;20:4016009.

[8] Mirmiran A, Yuan W, Chen X. Design for slenderness in concrete columns internally

reinforced with fiber-reinforced polymer bars. ACI Struct J 2011;98:116–25.

[9] Davidovits J. High-Alkali Cements for 21st Century Concretes. Am Concr Inst

1994;144:383–98. doi:10.14359/4523.

30
 ACCEPTED MANUSCRIPT

[10] Ganesan N, Abraham R, Deepa Raj S, Sasi D. Stress-strain behaviour of confined

Geopolymer concrete. Constr Build Mater 2014;73:326–31.

doi:10.1016/j.conbuildmat.2014.09.092.

[11] Lokuge W, Karunasena W. Ductility enhancement of geopolymer concrete columns

using fibre-reinforced polymer confinement. J Compos Mater 2016;50:1887–96.

PT
doi:10.1177/0021998315597553.

RI
[12] Elchalakani M, Ma G, Aslani F. Design of GFRP Reinforced Concrete Columns,. Mag

SC
Concr Res 2017.

[13] Karim, H, Sheikh, MN & Hadi MNS. Axial load-axial deformation behaviour of circular
NU
concrete columns reinforced with GFRP bars and helices. Constr Build Mater
MA

2016;112:1147–57.

[14] Afifi MZ. Behavior of circular concrete columns reinforced with FRP bars and stirrups.
D

Univ. of Sherbrooke, Canada, 2013.

E
PT

[15] Afifi MZ, Mohamed HM, Benmokrane B. Axial capacity of circular concrete columns
CE

reinforced with GFRP bars and spirals. J Compos Constr 2014;18:4013011–7.

[16] Afifi MZ, Mohamed HM, Benmokrane B. Strength and 616 axial behavior of circular
AC

concrete columns reinforced with CFRP bars and spirals. J Compos Constr

2014;18:4013010–35.

[17] Zadeh HJ, Nanni A. Design of RC columns using glass FRP reinforcement. J Compos

Constr 2013;17:294–304.

[18] Sreenath S, Balaji S, Saravana Raja Mohan K. Behaviour of axially and eccentrically

loaded short columns reinforced with GFRP bars. IOP Conf Ser Earth Environ Sci

31
 ACCEPTED MANUSCRIPT

2017;80.

[19] Tikka T, Francis M, Teng B. Strength of Concrete Beam Columns Reinforced with

GFRP Bars 2010;3:1–10.

[20] Hailu Tekle B, Khennane A, Kayali O. Bond behaviour of GFRP reinforced geopolymer

cement concrete. MATEC Web Conf 2017;120:1–10.

PT
doi:10.1051/matecconf/201712004002.

RI
[21] Hany NF, Hantouche EG, Harajli MH. Finite element modeling of FRP-confined

SC
concrete using modified concrete damaged plasticity. Eng Struct 2016;125:1–14.

doi:10.1016/j.engstruct.2016.06.047.
NU
[22] Ibrahim AMA, Fahmy MFM, Wu Z. 3D finite element modeling of bond-controlled
MA

behavior of steel and basalt FRP-reinforced concrete square bridge columns under lateral

loading. Compos Struct 2016;143:33–52. doi:10.1016/j.compstruct.2016.01.014.

D

[23] Piscesa B, Attardb MM, Samanic AK. Three-Dimensional Finite Element Analysis of
E
PT

Circular Reinforced Concrete Column Confined with FRP using Plasticity Model.
CE

Procedia Eng 2017;171:847–56. doi:10.1016/j.proeng.2017.01.377.

[24] Teng JG, Xiao QG, Yu T, Lam L. Three-dimensional finite element analysis of
AC

reinforced concrete columns with FRP and/or steel confinement. Eng Struct

2015;97:15–28. doi:10.1016/j.engstruct.2015.03.030.

[25] Amiri AM, Olfati A, Najjar S, Beiranvand P, Naseri Fard MH. The effect of fly ash on

flexural capacity concrete beams. Adv Sci Technol Res J 2016;10:89–95.

[26] Venkatasubramani R, Anuradha R, Uma K. Experimental investigation and analytical

modeling of reinforced Geopolymer concrete beam. Int J Civ Struct Eng 2012;2.

32
 ACCEPTED MANUSCRIPT

doi:10.6088/ijcser.00202030010.

[27] Nguyen KT, Ahn N, Le TA, Lee K. Theoretical and experimental study on mechanical

properties and flexural strength of fly ash-geopolymer concrete. Constr Build Mater

2015;106:65–77.

[28] Alfarah B, López-Almansa F, Oller S. New methodology for calculating damage

PT
variables evolution in Plastic Damage Model for RC structures. Eng Struct

RI
2017;132:70–86. doi:10.1016/j.engstruct.2016.11.022.

SC
[29] Youssf O, ElGawady MA, Mills JE, Ma X. Finite element modelling and dilation of

FRP-confined concrete columns. Eng Struct 2014;79:70–85.

NU
doi:10.1016/j.engstruct.2014.07.045.
MA

[30] Elchalakani M, Karrech A, Hassanein MF, Yang B. Plastic and yield slenderness limits

for circular concrete filled tubes subjected to static pure bending. Thin-Walled Struct
D

2016;109:50–64. doi:10.1016/j.tws.2016.09.012.
E
PT

[31] Yu T, Teng JG, Wong YL, Dong SL. Finite element modeling of confined concrete-II:
CE

Plastic-damage model. Eng Struct 2010;32:680–91.

doi:10.1016/j.engstruct.2009.11.013.
AC

[32] Hardjito D, Wallah SE, Sumajouw DMJ. The stress–strain behaviour of fly ash-based

geopolymer concrete. Dev Mech Struct Mater 2005:831–4.

[33] ACI 363. State-of-the-Art Report on High-Strength Concrete (ACI 363R-92). ACI J

Proc 1992;92.

[34] AS3600. AS3600-2009: Concrete Structures, Australian Standard. Sydney, Australia:

Australian Standard; 2009.

33
 ACCEPTED MANUSCRIPT

[35] Karrech A, Poulet T, Regenauer-Lieb K. A limit analysis approach to derive a

thermodynamic damage potential for non-linear geomaterials. Philos Mag

2012;92:3439–50.

[36] Karabinis AI, Rousakis TC. Concrete confined by FRP material: a plasticity approach.

Eng Struct 2002;24:923–32.

PT
[37] Karrech A, Abbassi F, Attar M, Basarir H. Self-consistent Fractal Damage of Natural

RI
Geo-materials in Finite Strain. Mech Mater 2017;104:107–20.

SC
[38] Dong M, Feng W, Elchalakani M, Karrech A, May EF, Highway S. Development of a

High Strength Geopolymer by Novel Solar Curing. Ceram Int 2017.

NU
doi:10.1016/j.ceramint.2017.05.173.
MA

[39] Nath P, Sarker PK. Geopolymer concrete for ambient curing condition. Proc. Australas.

Struct. Eng. Conf. 2012 (ASEC 2012), Perth, Western Australia: Engineers Australia;
D

2012.
E
PT

[40] Desayi P, Krishnan S. Equation for the Stress-strain Curve of Concrete. J Am Concr Inst
CE

1964;61:345–50. doi:10.1017/CBO9781107415324.004.

[41] Mander JB, Priestley JN, Park R. Theoretical Stress-Strain Model for Confined
AC

Concrete. J Struct Eng 1988;114:1804–26.

[42] Liu W, Xu M, Chen Z. Parameters calibration and verification of concrete damage

plasticity model of Abaqus. Ind Constr 2014;44:167–171, 213. (in Chinese).

[43] Kappos A, Konstantinidis D. Statistical analysis of confined high strength concrete.

Mater Struct 1999;32:734–48.

[44] AS1012-9. AS1012-9: Methods for testing concrete-determination of concrete

34
 ACCEPTED MANUSCRIPT

compressive strength, Australian Standard. Sydney, Australia: Australian Standard;

1991.

[45] Stoner J. Finite Element Modelling of GFRP Reinforced Concrete Beams. UWSpace,

2015.

[46] Hillerborg A, Modeer M, Petersson P. Analysis of crack formation and crack growth in

PT
concrete by means of fracture mechanics and finite elements. Cem Concr Res

RI
1976;6:773–82.

SC
[47] Karrech A, Regenauer-Lieb K, Poulet T. Damaged visco-plasticity model for pressure

and temperature sensitive geomaterials. Int J Eng Sci 2011;49:1141–50.

NU
MA
E D
PT
CE
AC

35
 ACCEPTED MANUSCRIPT

Fig. 1. Schematic of GFRP RC columns, (a) concentrically loaded GPC column G150-C, (b)
eccentrically loaded GPC column G150-75, (c) GPC beam G40-F. (Unit in mm).

Fig. 2. The general test setup for (a) GFRP reinforced concrete columns and (b) GFRP
reinforced concrete beams

Fig. 3. GFRP RC columns with concentric loading (a) left-to-right, G250-C, G150-C, G75-C,
(b) left-to-right, O250-C, O150-C, O75-C.

PT
Fig. 4. Eccentrically loaded GFRP-GPC columns (left-to-right, G150-25; G75-25; G150-50;
G75-50; G150-75; G75-75)

RI
Fig. 5. Eccentrically loaded GFRP-OPC concrete columns (left-to-right O150-25; O75-25;
O150-45, O75-35)

SC
Fig. 6. GFRP RC Beams (a) G80-F, (b) G40-F, (c) O150-F, (d) O75-F

Fig. 7. The plain concrete damage plasticity model [42]

NU
Fig. 8. Comparison of unconfined stress-strain models for O75-C by Desayi and Krishnan
[40] (Eq. 9), Kappos and Konstantinidis [43] (Eqs. 10, 18) and Mander et al. [41] (Eqs. 14,
MA

18).

Fig. 9. The elastic stress-stress relationship of the longitudinal GFRP bars used in this study.
ffu: ultimate tension capacity, ffc: ultimate compression capacity, Ef: the elastic modulus.
D

Fig. 10. The typical geometry of the FE models of, (a) concentrically loaded GPC columns,
E

(b) eccentrically loaded GPC columns, (c) GPC and OPC concrete beams, (d), OPC concrete
PT

columns

Fig. 11. The FE Mesh for eccentrically loaded GPC columns with 75mm stirrup spacing. (a)
CE

Concrete section: G3D8R with 0.04mm mesh size, (b) GFRP: T3D2R with 0.04mm mesh
size, (c) constraints and boundary conditions of the model.
AC

Fig. 12. The comparison between different mesh sizes for G250-C

Fig. 13. The comparison between the confinement models by Kappos and Konstantinidis [43]
and Mander et al. [41] of (a) G250-C, (b) G75-25

Fig. 14. The typical FE results for (a) concentrically loaded GPC columns, (b) eccentrically
loaded GPC columns, (c) GPC beams

Fig. 15. Comparison of experimental and FEA results of the axial load-axial displacement
curves for the concentrically loaded GPC columns G75-C, G150-C and G250-C

Fig. 16. Comparison of experimental and FEA results of the axial load-axial displacement
curves for the eccentrically loaded GPC columns (a) G75-25 and G150-25, (b) G75-50 and
G150-50, (c) G75-75 and G150-75

36
 ACCEPTED MANUSCRIPT

Fig. 17. Comparison of experimental and FEA results of the axial load-displacement curves
for the GFRP reinforced GPC beam G40-F

Fig. 18. Comparison of experimental and FEA results of the axial load-axial displacement
curves for the concentrically loaded OPC concrete columns O75-C, O150-C and O250-C

Fig. 19. Comparison of experimental and FEA results of the axial load-axial displacement
curves for the eccentrically loaded OPC concrete columns (a) O75-25 and O150-25, (b) O75-
35 and O150-45

Fig. 20. Comparison of experimental and FEA results of the axial load-displacement curves

PT
for the GFRP reinforced OPC concrete beams O75-F and O150-F

Fig. 21. Moment interaction diagrams for (a) GPC columns, (b) OPC concrete columns and

RI
beams. (s: stirrup spacing)

SC
NU
MA
E D
PT
CE
AC

37
 ACCEPTED MANUSCRIPT

Table 1: Nominal and measured Properties of GFRP Bars [6,7]

Area
Tensile Elastic Tensile Strength Ultimate Strain in
2
Diameter (mm) (mm ) Modulus (GPa) (MPa) Tension (%)
Bar
Size V-ROD Test V-ROD Test V-Rod Test V-ROD Test V-ROD Test

8mm 6.35 - 31.7 - 46.1 - 784 - 1.90 -

PT
14mm 12.7 14.6 126.7 168 46.3 50 708 1200 1.70 2.4

RI
Table 2: Summary of experimental setup for column and beam specimens from this study

SC
and Elchalakani et al. [12]

OPC concrete columns by GPC columns (this

NU
Elchalakani et al. [12] study)
Concrete column dimensions 160×260×1200 160×260×1200
Stirrup spacings (mm) 75, 150, 250 75, 150, 250
MA

Average cylinder compressive strength, f’c (MPa) 32.8 26.0

GFRP long bars, Diameter (mm) 12a 14b
GFRP stirrups, Diameter (mm) 6a 8b
Volumetric ratio of the transverse reinforcement, ρs at
D

0.74% 0.94%
75mm spacing
E

Volumetric ratio of the longitudinal reinforcement, ρc 2.19% 2.37%

PT

% increase in load carrying capacity over the 3.2% 10.8%

corresponding plain concrete section
a: V-Rod GFRP bars and stirrups from Pultrall, Canada
CE

b: MateenBar GFRP bars and stirrups from Pultron Composites, New Zealand
AC

38
 ACCEPTED MANUSCRIPT

Table 3: Summary of experimental results for column and beam specimens from this
study and Elchalakani et al. [12]

%
Vertical FE Average %
Ultimate Maximu Maximu % % differenc
Peak deflectio deflectio Average % difference
Column vertical m strain m strain residual FE peak differenc e in
load n at peak n at peak difference in
ID deflectio (SGL) (SGR) load to load (kN) e in peak deflectio
(kN) load load in peak load deflection
n (mm) (µε) (µε) peak load load n at peak
(mm) (mm) at peak load

PT
load

O250-C 1402 5.79 8.38 1684 1526 0% 1426 5.77 2% 0%

O150-C 1367 5.57 10.67 1896 - 0% 1458 5.96 6% 7%

RI
O75-C 1449 6.39 15.64 4983 5084 - 1624 6.80 11% 6%

SC
O150-25 880 4.86 10.77 1474 -8607 30% 859 4.63 2% 5%

O75-25 917 4.51 11.18 2632 -2881 48% 1099 6.77 17% 33% 8% 13%

O75-35 788 6.13 9.12 7237 -8384 48% 882 7.45 11% 18%
NU
O150-45 584 5.67 5.71 269 -423 0% 569 5.54 3% 2%

O75-F 189 43.16 54.19 13695 -10043 - 178 46.08 6% 6%

O150-F 144 26.46 34.00 747 -1770 - 170 43.42 15% 39%
MA

G250-C 1041 5.81 6.03 -5630 -4960 0% 1115 6.80 7% 15%

G150-C 1194 7.23 8.35 -5931 -5850 0% 1151 7.09 4% 2%

G75-C 1357 8.12 12.03 -11753 -8945 66% 1438 9.24 6% 12%
D

G150-25 657 6.25 14.87 6866 -9595 31% 648 6.25 1% 0%

G75-25 804 5.28 10.60 6127 -9780 48% 744 5.57 8% 5%

E

G150-50 353 5.65 14.17 7512 -7292 13% 315 4.44 12% 27% 15% 11%
PT

G75-50 454 5.61 18.72 12236 -7860 56% 384 4.63 18% 21%

G150-75 234 10.24 23.47 5720 -4147 16% 165 9.52 42% 8%

G75-75 244 10.67 33.37 7834 -6150 16% 188 9.51 30% 12%
CE

G80-F 124 - - - - - - - - -

G40-F 172 46.30 49.07 10199 -8087 - 228 49.48 25% 6%

AC

+ve strain value refers to tension and –ve value refers to compression

39
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13
Figure 14
Figure 15
Figure 16
Figure 17
Figure 18
Figure 19
Figure 20
Figure 21