Page 1 of 8

NUMERICAL SIMULATION OF THE MECHANICAL BEHAVIOUR OF

FRP RC TENSIONED ELEMENTS
IRENE VILANOVA MARCO
MSc
University of Girona
EPS Campus Montilivi
irene.vilanova@udg.edu*
ALBERT TURON TRAVESA
PhD
University of Girona
EPS Campus Montilivi
albert.turon@udg.edu
LLUS TORRES LLINAS
PhD
University of Girona
EPS Campus Montilivi
lluis.torres@udg.edu
MARTA BAENA MUOZ
PhD
University of Girona
EPS Campus Montilivi
marta.baena@udg.edu
Abstract
The use of fibre reinforced polymer (FRP) bars in reinforced concrete (RC) structures has
emerged as an alternative to traditional RC due to the corrosion of steel in aggressive
environments and the electromagnetic transparency of FRP. Because of their different
mechanical properties, the cracking and deformability behaviour of FRP RC elements is quite
unlike traditional steel reinforced concrete. In recent decades a number of authors and design
codes have proposed several approaches to predict the behaviour of FRP RC elements.
However, using finite element method (FEM) for the advanced modelling of FRP RC
structures has not been investigated in depth. In this paper, a numerical method to model FRP
RC tensioned elements using Abaqus FEM code is presented. The interaction between
reinforcement and concrete is modelled using cohesive elements powered with a bi-linear
bond-slip law obtained directly from experimental tests. Concrete mechanical behaviour is
modelled using a damaged plasticity approach with Drucker-Prager criterion to define the
failure surface. Numerical results are compared with experimental data and Eurocode 2
predictions.
Keywords: Concrete, FRP, tensile members, cohesive models, bond
1. Introduction
Conventional steel reinforced concrete structures can be corroded in aggressive environments,
such as saline, which reduces their service life and increases maintenance and replacement
costs [1, 2]. Fibre reinforced polymers (FRP) as reinforcement for concrete structures have
emerged as an alternative to steel due their good behaviour in corrosive environments and the
electromagnetic transparency.
Page 2 of 8
The mechanical behaviour of FRP RC structures differs from traditional steel RC structures.
FRP bars have a linear behaviour up to failure and a lower elastic modulus leading to larger
deformations and crack widths. Therefore, the design of FRP RC structures may often be
governed by serviceability limit states. The different analytical methods proposed to predict
the behaviour of reinforced concrete structures under tensile loads take into account the
cracking behaviour, and the tension stiffening effect, and allow calculations of deformations,
crack spacing and cracks widths [1-6]. Most of these methods have been developed based on
many simplifications. More recently, methods to model the mechanical behaviour of
reinforced concrete structures under tensile load using advanced numerical tools such as the
finite element method (FEM) have also been presented [7-9]. These models rely on fewer
hypotheses and simplifications than the design codes, but have a much higher computational
cost. Further research in this area is needed to reduce that cost and improve the constitutive
laws used to model the material behaviour.
In this work, a numerical method to study the behaviour of FRP RC tensioned members is
presented. The bond between concrete and rebar is modelled using in-house cohesive
elements. These cohesive elements are fed with the material parameters obtained directly
from experimental tests and reported in a previous work by Baena et al. [10]. The constitutive
behaviour of the concrete is modelled using the concrete damaged plasticity (CDP) model
available in ABAQUS [11]. Results obtained using this method are compared with data from
a previous campaign reported in Baena et al. [6], and with crack spacing predictions from
Eurocode 2 [4].
2. Description of the model
Three different constitutive laws are used in this work to model FRP RC members. It is
assumed that the mechanical behaviour of the FRP is linear elastic until failure. Therefore, the
FRP rebar is modelled using a linear elastic constitutive equation. For the concrete, the DCP
model of ABAQUS/Explicit is used [11]. Finally, the interaction between the FRP rebar and
concrete is modelled using cohesive elements. A brief description of the bond interface and
the concrete damaged plasticity model is introduced in this section.
2.1 Bond interface model
Several analytical laws have been proposed to reproduce the bond behaviour between
concrete and reinforcing bars under tensile loads [12-15]. In this work, the bi-linear cohesive
law described in Figure 1 is used to describe the interaction between concrete and FRP rebar.
The law was described by Turon et al. [16] and adapted by Gonzalez et al. [17] to be used in
ABAQUS/Explicit code. Although the constitutive model was developed for the delamination
of composite structures, it can also be used to model bonds between reinforcing bars and
concrete, assuming that the interaction between concrete and FRP rebar can be collapsed to
zero-thickness surface. This approach has been previously adopted by other authors such as
Serpieri and Alfano [18, 19], who enhanced the cohesive formulation by introducing friction
and dilatancy to model the interaction between concrete and reinforcement. In this work a
simpler constitutive model is used because it is assumed that all the mechanisms taking place
are contained in the fracture toughness, G
c,
used in the constitutive equation.
Page 3 of 8
Figure 1. Bi-linear bond-slip law (Turon et al. [16]).
The first part of the constitutive model is linear elastic up to the maximum bond stress
0
,
involving an initial displacement
0
. At this point the interface starts to become damaged.
Once the maximum bond stress is reached and the crack begins to propagate, the stress is
reduced up to the maximum displacement
f
. G
c
is the fracture toughness needed to propagate
the crack.
The fracture toughness and the maximum bond stress are taken from pull-out test results [10].
The authors tested 88 specimens, changing the type of reinforcement bar (carbon, glass and
steel). The maximum bond stress is defined as the maximum point of the bond stress-slip
curve and the fracture toughness as the area below the curve (Figure 2).
Figure 2. Typical pull-out test response.
2.2 Concrete model
The CDP model from the ABAQUS library [11] is used in the finite element code to define
concrete behaviour. The model considers the total strain as the addition of elastic (
el
) and
plastic (
pl
) strain, =
el
+
pl
.
The stress-strain relation associated with damage corresponds to:
s = 1 d
( )
D
0
el
:
pl
( )
(1)
where d is the damage variable (d=0 no damage, d=1 fully damaged) and is the non-
damaged elastic modulus. Damage is associated with the damaged mechanisms of concrete
cracking in tension and crushing in compression that involve a decrease of the elastic
modulus.
Page 4 of 8
Damage is governed by hardening variables
pl
and the effective stress d=d(s ,
pl
). The
hardening variables at compression (
c
pl
) and tension (
t
pl
) are considered independent.
Microcracks and crushing are proportional to the hardening variables. These variables control
the yield surface and the degradation of the elastic modulus. However, they are also related to
the energy dissipation that generates microcracks. The yield surface F(
s ,
pl
), which
determines the failure state, is the effective tensile zone, based on the formulation proposed
by Lubiner et al. [20] and by Lee and Fenves [21] (Figure 3). A flow rule defines how the
yield surface develops under plastic strain.
Figure 3. Yield surface of DCP model [11].
3. Validation of the presented method
3.1 Description of the model
To validate the model, results from an experimental campaign conducted by the authors and
reported in a previous work (Baena et al. [6]) have been used. Four GFRP RC elements
axially loaded in tension were tested. Each test specimen was a concrete prism 1300 mm long;
the nominal diameter (d
n
) and the cross-section size are summarized in Table 1.
Table 1. Definition, geometric and mechanical properties of concrete and FRP bar specimens.
SPECIMEN
Cross-
section
size (mm)
Nominal
diameter
dn (mm)
Reinforce-
ment ratio
(%)
Concrete
compres-
sive
strength
fcm (MPa)
Concrete
tensile
strength
fct (MPa)
Concrete
elasticity
modulus Ec
(MPa)
Bar
elasticity
modulus
EFRP (MPa)
13_170 170 13.7 0.51 48.1 1.75 27315 37418
16_170 170 16.9 0.71 46.6 2.58 34514 42835
19_170 170 19.1 1.00 56.2 2.10 33275 40595
16_110 110 15.9 1.69 56.2 2.34 33275 38758
The experimental results of the compressive and tensile strength of concrete (f
cm
and f
ct
respectively), and the moduli of elasticity (E
c
for concrete and E
FRP
for reinforced bar) are
also given in Table 1.
Page 5 of 8
The mechanical properties needed for the bond-slip law, fracture toughness and bond strength
are given in Table 2. The bond-slip parameters depend on the bar diameter. Further details of
the bond behavior can be found in [6].
Table 2. Mechanical properties of the bond-slip law.
SPECIMENT
Bond strength

max
(MPa)
Fracture
toughness
GF (mJ)
13_170 17.06 237.26
16_170 17.33 279.91
19_170 14.57 225.54
16_110 17.33 279.91
ABAQUS/Explicit [11] is used to model the RC members. An Abaqus C3D8R element is
employed to model the concrete and the rebar material. On average, 61,000 elements per
specimen are used to model the ties. Each element consists of eight nodes and has an
approximate size of 5 mm. Cohesive elements placed between the concrete and the
reinforcing bar are used to simulate the bond behaviour. Taking advantage of the symmetry,
only an eighth is analysed (Figure 4). A displacement controlled analysis is performed,
increasing the displacement applied at the free end of the rebar.
Figure 4. Meshed eighth part of the concrete specimen.
3.2 Results
In this section the results of the method are presented, compared with the experimental data
and discussed. The comparison between the experimental and the FEM prediction is shown in
Figure 5.
The numerical results have a linear behaviour until the cracking load (P
cr
) is reached. As seen
in Figure 5, the FEM model captures the cracking process when the load increases. Once all
the cracks are formed (stabilized cracking), the trend of the modelling results in a line parallel
to, or slowly approaching, the bare bar response. The sudden changes in the FEM curve are
attributed to the nature of the analysis, an explicit scheme, and the sudden energy release
when a new crack appears.
The FEM simulation can also provide the crack pattern obtained during the analysis. The final
crack patterns obtained for each specimen are shown in Figure 6. Crack spacing, S
m
, can be
predicted using equation 2 from Eurocode 2:
Page 6 of 8
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
0
50
100
150
200
13 170
Member Mean Strain (
m
)
A
x
i
a
l

L
o
a
d
P

(
k
N
)
Experimental
Bare bar
Modelling
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
0
50
100
150
200
16 170
Member Mean Strain (
m
)
A
x
i
a
l

L
o
a
d
P

(
k
N
)
Experimental
Bare bar
Modelling
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
0
50
100
150
200
19 170
Member Mean Strain (
m
)
A
x
i
a
l

L
o
a
d
P

(
k
N
)
Experimental
Bare bar
Modelling
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
0
50
100
150
200
16 110
Member Mean Strain (
m
)
A
x
i
a
l

L
o
a
d
P

(
k
N
)
Experimental
Bare bar
Modelling
Figure 5. Comparison of numerical and experimental results.
S
rm
= 50+0, 25k
1
k
2
d
n

eff
(2)
where k
1
is 0.8 for ribbed and 1.6 for smooth steel bars, k
2
is 0.5 for flexural and 1.0 for
tensile loading, and
eff
is the ratio of internal reinforced to the effective area of the concrete
in tension.
The crack spacing obtained from the FEM simulations, together with the experimental results
and Eurocode 2 provisions (Eq. 2), are given in Table 3.
For lower reinforcement ratios, <1, the FEM predictions underestimate crack spacing, while
for <1 the crack spacing is overestimated in the FEM simulations. Although the random
effects associated with the cracking process always have to be taken into consideration when
analysing crack patterns, the FEM predictions of mean crack spacing are more accurate than
predictions made using Eurocode 2 equations, at least for the specimens with higher
reinforcement ratios.
Table 3. Comparison between numerical and experimental results for crack spacing.
SPECIMEN
Experimental
S
rm,exp
(mm)
Eurocode 2
S
rm,EC2
(mm)
S
rm,EC2
/S
rm,exp
FEM model
S
rm,FEM
(mm)
S
rm,FEM
/S
rm,exp
13_170 (=0.51%) 264.96 586.00 2.2 188.30 0.7
16_170 (=0.71%) 227.75 486.00 2.1 145.00 0.6
19_170 ( =1.00%) 113.27 434.50 3.8 116.00 1.0
16_110 ( =1.69%) 123.34 241.26 1.9 138.50 1.1
Page 7 of 8
Figure 6. Ties crack pattern and crack spacing
4. Conclusions
A numerical method to predict the response of GFRP RC members under tensile load is
presented. Cohesive elements are used to model the interaction between concrete and
reinforcement and a concrete damaged plasticity model is used to define the concrete
behaviour. The method is validated by comparing its predictions with experimental data. It
reproduces satisfactorily the cracking process and the experimental results in terms of load-
deformation. The method also provides satisfactory results for crack spacing when compared
with those obtained using Eurocode 2 equations.
5. Acknowledgements
The authors acknowledge the support provided by the Spanish Government (Ministerio de
Educacin y Ciencia), Project BIA2010-20234-C03-02.
6. References
[1] FIB (2007). FRP reinforcement in RC structures, prepared by Task Group 9.3, International Federation
for Structural Concrete, fib bulletin 40, 2007, Lausanne, Switzerland.
Page 8 of 8
[2] ACI 440.3R-04. Guide test methods for fiber-reinforced polymers (FRPs) for reinforcing or
strengthening concrete structures ACI committee 440. Farmington Hills, Michigan (USA), American
Concrete Institute. 2004.
[3] CEB-FIB Model Code. Comit Euro-International du Bton. Fdration Internationale de la
Prcontrainte, ThomasTelford House, London, England, 1990.
[4] EUROCODE 2: Design of concrete structures-part 1: general rules and rules for buildings, DD ENV
1992-1-1:1992. CEN (Comit Europen de Normalisation). Brussels; 1992. p. 274.
[5] CSA S806-02 Design and construction of building components with fiber reinforced polymers Canada:
CSA Standard, Canadian Standards Association, 2002.
[6] BAENA M., TORRES LL., TURON A., MIAS C., Experimental study and code predictions of fibre
reinforced polymer reinforced concrete (FRP RC) tensile members, Composite Structures, Vol. 93, April
2011, pp. 2511-2520.
[7] DIAS-DA-COSTA D., ALFAIATE J., JLIO E.N.B.S., FE modelling of the interfacial behaviour of
composite concrete members. Construction and Building Materials. Vol. 26. July 2011. pp. 233-243.
[8] MACORINI L., FRAGIACOMO M., AMADIO C., IZZUDIN B.A., Long-term analysis of steel-
concrete composites beams: FE modelling for effective width evaluation. Engineering Structures. Vol.
28, 2006, pp. 1110-1121.
[9] WU, H.Q., GILBERT R.I., Modelling short-term tension stiffening in reinforced concrete prisms using a
continuum-based finite element model, Engineering Structures, Vol. 31, June 2009, pp. 2380-2391.
[10] BAENA M., TORRES LL., TURON A., BARRIS C., Experimental study of bond behaviour between
concrete and FRP bars using a pull-out test. Composites: Part B. Vol. 40. 2009, pp. 784-797.
[11] ABAQUS version 6.10 Users manual. RI: Hibbitt, Karlsson & Sorensen Inc, 2010.
[12] MALVAR LJ., Bond stress-slip characteristics of FRP rebars. Techinical Report TR2013-SHR, Naval
Facilities Engineering Service Center, Port Huneme, California, February 1994, p. 45.
[13] COSENZA, E., MANFREDI G., REALFONZO R., Analytical modelling of bond between FRP
reinforcing bars and concrete Proceedings of second international RILEM symposium (FRPRCS-2)
London, 1995.
[14] COSENZA, E., MANFREDI G., REALFONZO R., Development length of FRP straight rebars
Composites B, Vol 33, 2002, pp. 493-504.
[15] ELIGEHAUSEN R., POPOV EP., BERTERO VV., Local bond stress-slip relationships of deformed bars
under generalized excitations. Report No 83/23, EERC, University of California, 1983.
[16] TURON A., CAMANHO P.P., COSTA J., DVILA C.G., A damage model for the simulation of
delamination in advanced composites under variable-model loading, Mechanics of Materials, Vol. 38,
2006, pp. 1072-1089.
[17] GONZALEZ E.V., MAIMI P., TURON A. CAMANHO P.P., Simulation of delamination by means of
cohesive elements using an explicit finite element code, Computers Materials & Continua, Vol. 9, 2009,
pp. 51-59.
[18] SERPERI R., ALFANO G., Bond-slip analysis via a thermodynamically consistent interface model
combining interlocking, damage and friction Int. J. Numer. Meth. Engng, Vol. 85, 2011, pp. 164-186.
[19] ALFANO G., MARFIA S., SACCO E., A cohesive damage-friction interface model accounting for water
pressure on crack propagation Computer methods in applied mechanics and engineering. Vol. 196, 2006,
pp. 192-209.
[20] LUBLINER J., OLIVER J., OLLER S., OATE E., A plastic-damage model for concrete, J. Solid
structures, Vol. 25, 1989, pp. 299-326.
[21] LEE J., FENVES G.L., A Plastic-Damage Concrete Model for Earthquake Analysis of Dams,
Earthquake Engineering and Structural Dynamics, Vol. 27, 1998, pp. 937-956.