ORIGINAL ARTICLE

Finite element analysis of FRP-strengthened

RC beams
Teeraphot Supaviriyakit1, Phuwanat Pornpongsaroj1
and Amorn Pimanmas2

Abstract
Supaviriyakit, T., Pornpongsaroj, P. and Pimanmas, A.

Finite element analysis of FRP-strengthened RC beams

Songklanakarin J. Sci. Technol., 2004, 26(4) : 497-507
This paper presents a non-linear finite element analysis of reinforced concrete beam strengthened
with externally bonded FRP plates. The finite element modeling of FRP-strengthened beams is demonstrated.
Concrete and reinforcing bars are modeled together as 8-node isoparametric 2D RC element. The FRP plate
is modeled as 8-node isoparametric 2D elastic element. The glue is modeled as perfect compatibility by directly
connecting the nodes of FRP with those of concrete since there is no failure at the glue layer. The key to the
analysis is the correct material models of concrete, steel and FRP. Cracks and steel bars are modeled as
smeared over the entire element. Stress-strain properties of cracked concrete consist of tensile stress model
normal to crack, compressive stress model parallel to crack and shear stress model tangential to crack. Stressstrain property of reinforcement is assumed to be elastic-hardening to account for the bond between concrete and steel bars. FRP is modeled as elastic-brittle material. From the analysis, it is found that FEM can
predict the load-displacement relation, ultimate load and failure mode of the beam correctly. It can also
capture the cracking process for both shear-flexural peeling and end peeling modes similar to the experiment.

Key words : FRP, strengthening, FEM, nonlinear analysis, peeling failure

1

Graduate Student in Civil Engineering 2Ph.D.(Structural Engineering), Asst. Prof., Sirindhorn International
Institute of Technology, Thammasat University, Rangsit Center, Khlong Luang, Pathum Thani, 12121
Thailand.
Corresponding e-mail: amorn@siit.tu.ac.th
Received, 9 December 2003
Accepted, 2 February 2004

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
498

. . 2547 26(4) : 497-507

...
(FRP)
8 FRP
8 FRP
FRP

- FRP
(smeared) -

-
- FRP



FRP

12121
The need to increase the capacity of the
existing structures has recently become one of
the most active areas in civil engineering. Several
strengthening methods have been used in the past
with varying degree of success. These include
1) enlargement of the cross section, 2) addition
of new steel members, 3) steel plate bonding,
4) external post-tensioning, 5) reduction of span
length, and recently 6) external bonding with FRP
plate. FRP is a composite material consisting of
numerous high-strength fibers impregnated in
resin. It is gaining popularity for structural
strengthening because of its high tensile strength,
low weight and high durability (no corrosion)
(Hollaway and Leeming,1999)
FRP has been successfully used to increase
the flexural and shear capacities of RC beams
(Saadatmanesh and Ehsani,1990). However, many
experimental results have shown the premature

failure due to peeling of FRP plate before the

enhanced capacity is reached (Ritchie et al,1991).
Arduini and Nanni (1997) described two common
types of peeling failure, that is, end peeling and
shear-flexural peeling as shown in Figure 1. The
end peeling results from the combination of shear
and normal tensile stress localized in the vicinity
of the plate end. When the principal tensile stress
reaches the tensile strength of concrete, a crack
initiates and propagates horizontally at the level of
tension steel, ripping off the concrete cover. The
shear-flexural peeling initiates at the base of the
flexural or shear-flexural crack, and propagates
towards the support.
These failures are highly undesirable due to
their brittleness. In order to ensure the safe design
of structures, the load capacity associated with
these failures must be accurately evaluated. However, presently there are no commonly acceptable

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
499

Figure 1. Typical peeling failures of FRP-strengthened beam.

design methods to predict the peeling failure load

yet. In this paper, the authors employ the nonlinear finite element method to predict the peeling
failure load.
Significance of research
Currently, FRP materials have been widely
used to increase the flexural and shear capacity of
RC members. FRP plate is advantageous over
steel plate due to its low weight, high strength
and non-corrosive property. However, there is no
standard reliable design method for preventing
the peeling failure due to lack of sufficient experimental data. Since the peeling failure is very brittle,
it is imperative to assess the load capacity associated with these failures accurately to ensure the
structural safety for design. FEM analysis using
non-linear constitute models of cracked concrete
and reinforcing bars is proposed to predict the
stress concentration at the plate end as well as

at the base of diagonal cracks. This leads to the

understanding in the failure mechanism, damage
zone and load capacity.
Finite element modeling of FRP-strengthened
beam
The typical finite element mesh of the FRPstrengthened beam is shown in Figure 2. The mesh
consists of concrete, steel bars and FRP plates.
The concrete and reinforcing steel are modeled
together by 8-node 2-D isoparametric plane stress
RC element. The RC element considers the effect
of cracks and reinforcing steel as being smeared
over the entire element.
Assuming perfect compatibility between
reinforcing steel and cracked concrete, the 2-D
stress vector of the entire element is the direct
addition of concrete according to eq. (1).

{ }rc = { }c + { }r

Figure 2. Finite element modeling and typical finite element mesh.

(1)

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
500

where {}rc, {}c and {}r are stresses of

reinforced concrete, cracked concrete and reinforcing bar, respectively.
Concrete cracks and reinforcing steel are
treated in a smeared manner. The four-way fixed
crack model developed by Okamura and Maekawa
(1991) and Fukuura and Maekawa (1999) is used
to deal with cracks in concrete. This four-way
cracking model can deal with up to four cracks in
arbitrary directions. As will be shown later, the
rotation of a crack in any element hardly occurs
due to the localized failure nature. Hence, the fourway fixed crack model is considered sufficient for
the problem concerned.
The FRP plate is modeled by 2D elastic
element. The epoxy layer can also be modeled
by 2D elastic elements. However, the epoxy is
usually much stronger than concrete and is therefore not a weak link in the FRP strengthened beam.
As will be shown in the experiment, no failure
takes place at the epoxy-bonding interface, so the
full compatibility between FRP and beam soffit
may be assumed. This is done by directly joining
the node of FRP element with concrete element
without using the elastic elements representing
epoxy.
Local Constitutive Models of Cracked Concrete,
Reinforcing Bars and FRP Plates
The local constitutive models of concrete
(cracked and uncracked state), steel bars and FRP
plates are essential to deal with concrete cracking,

reinforcement yielding, crushing failure and

elastic-brittle characteristics of FRP plate. The
following sections explain the constitutive models
of cracked concrete, reinforcing bars and FRP plate.
1. Cracked Concrete
The local strain vector (with respect to the
crack axis) of cracked concrete consists of tensile
strain normal to a crack, n, compressive strain
parallel to a crack, t, and the shear strain along
crack interface, cr as shown in Figure 3. The
constitutive laws can be defined as the uni-axial
relations between the average stress and average
strain as follows.

n = ( n , t , 1,2,...,n )

(2)

t = ( t , n , n+1,...,2n )

(3)

c = ( cr , n , t , 2n+1,... )

(4)

where n, t and c represent the constitutive

laws for computing normal tensile, compressive
and shear stresses, respectively, represents the
set of path-dependent parameters. The formulation of the above constitutive laws and the pathdependent parameters were already reported
(Maekawa et al, 2003). In the following, only the
formula for envelope will be given.
The above constitutive laws correspond to
three degrees of freedom of a cracked element,

Figure 3. Degree of freedom for cracked concrete.

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
501

namely, tension stiffening/softening model normal

to the crack, compression model parallel to the crack
and shear transfer model along the crack interface.
The coupling between shear and normal action is
not explicitly considered in the formulation of
stiffness matrix. The shear dilatancy is implicitly
considered by adding the induced compressive
normal stress due to shear to the computed normal
stress orthogonal to the crack axis.
1.1 Coupled compression-tension model
for normal stresses parallel and perpendicular
to crack
The constitutive law for computing
normal stresses (denoted by n and t in eq. (2)
and (3) above) is the coupled compression-tension
model as shown in Figure 4. For concrete in
tension, the model covers both the softening due
to aggregate bridging at the crack plane as well as
the tension stiffening effect due to bond stress
transfer between concrete and steel bar. For the
envelope in tension, the model can be expressed
as follows
c
tu
n = f t
(5)
n
where n is the tensile stress normal to
crack, tu is the cracking strain, c is softening/

stiffening parameter. The parameter "c" indicates

the rate of drop in tensile stress of concrete after
cracking. It is set to be 0.4 for concrete element
containing reinforcement and 2.0 for plain concrete element. The higher value of "c" indicates
the more sudden decrease in concrete tensile stress
after cracking. By using the appropriate value of
"c" for each concrete element, the localized failure
can be covered according to the fracture mechanics
theory.
For compression parallel to the crack
direction, the normal stress is computed using the
elasto-plastic fracture model as follows

t = K0 Ec0 ( t p )

(6)

Where t is the compressive stress

parallel to crack, K0 is the fracture parameter
representing the continuum damage as a result of
dispersed cracking in concrete, Ec0 is the initial
stiffness and p is the compressive plastic strain.
The model combines the non-linearity of plasticity
and fracturing damages to account for the permanent deformation and loss of elastic strain energy
capacity, respectively. The reduced compressive
stress due to transverse tensile strain is considered
in the model as additional damage.

Figure 4. Coupled compression-tension model for normal stresses parallel and perpendicular
to a crack.

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
502

1.2 Shear transfer model

For computing shear stress along the
crack interface and the shear stiffness of the crack,
the contact density model (Okamura and Maekawa,
1991; and Maekawa et al, 2003 ) is used for the
envelope (Figure 5). The following equations give
shear stress and shear stiffness along the crack,
2
= f st
(7)
1+ 2
Gcr =

( )
cr

(8)

where fst is shear strength transfer along

the crack. is the normalized shear strain defined
as follows.

= cr
(9)

For the cracked RC element, The

imposed total shear strain () compatible with the
assumed nodal displacement can be separated into
the shear strain due to un-cracked concrete part
(0) and shear strain due to cracks (cr) as below

= 0 + cr cr

(10)

Generally, shear strain due to un-cracked

concrete part is relatively small compared with
shear strain to crack, thus it can be ignored

2. Modeling of reinforcing bar

Local strain vector of reinforcing bar is
defined along the bar axis. Reinforcing bar is
modeled in the smeared concept, that is, it is
assumed uniformly distributed over the whole
element. This representation is adopted since it
does not require additional nodes and elements.
For reinforcing steel, compressive strains are
usually smaller than tensile strains, thus reinforcement hardly yields in compression before spalling
of concrete cover has occurred. At the tension side,
however, the highly localized plasticity at the
vicinity of cracks has to be taken into account.
Due to bond effect, the average yield stress is
lower than that of bare bar.
Thus the modeling of reinforcing bar in
compression and tension can be different. For
compression, the bilinear bare bar model is assumed
in the analysis. For tension, the average stressstrain relationship considering the effect of bond
between concrete and steel bar is adopted. Figure 6
shows stress-strain model of reinforcing bar. Since
cracks in reinforced concrete element need not be
orthogonal with the reinforcement direction, the
bond effect may not be fully functional in such
case. Reinforcing bar orthogonal to crack is
supposed to have full bond effect. On the contrary,
reinforcing bar parallel to crack is supposed to
follow bare bar behavior. Therefore, the computation of mean yield strength has to take into account

Figure 5. Shear transfer model.

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
503

Figure 6. Reinforcing bar model.

the angular deviation of normal to crack from

reinforcing bar direction (Maekawa et al, 2003).
3. Modeling of FRP plate
The FRP plate is modeled as elasto-brittle
material. The stress-strain model is shown in
Figure 7. The stress is linear up to the tensile
strength and drops sharply to zero, representing
the fracture of FRP plate.
Failure Criterion
In the finite element program, the failure is
assumed to occur when one of the strain compo-

nents n, t and cr attains a certain value specified

as limiting failure strain. For the RC element based
on RC smeared crack model as shown in Figure 3,
these failure strains are set as follows, (1) for
tension failure, n = 0.03, (2) for compression
failure, t = -0.01, and (3) for shear failure, cr =
0.02.
Numerical Verification
Recently, the authors have conducted the
test on peeling characteristics of FRP-strengthened
beams (Pornpongsaroj and Pimanmas, 2003). The
objective of the test was to investigate the effect

Figure 7. FRP stress-strain model.

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
504

Table 1. Material Properties (obtained from laboratory testing and used in finite
element analysis)
Material

Yield Strength (MPa)

Ultimate Strength (MPa)

DB 16
532.98
DB 12
461.66
RB 6
383.77
FRP-plate
(thickness 1.2 mm.)
Concrete
Compressive strength = 45 MPa

of bonded length on peeling mode of FRP. Here,

the authors selected two FRP-strengthened beam
specimens (A-200-P and B-200-P) and one control
(non-strengthened) beam (specimen A) for verification with FEM. The cross section and reinforcement arrangement of the beams are shown in
Figure 8. The load set-sup and FRP installation are

Elastic Modulus (MPa)

606.94
615.77
493.05
2200

2.0105
2.0105
2.0105
1.5105

given in Figure 9. The material properties of this

experiment are given in Table 1. In the experiment,
it was found that A-200-P fails by shear-flexural
peeling and B-200-P by end peeling. These failures
are target of FEM verification.
Figure 10 shows the comparison between
FEM and experimental results for all beams tested.

Figure 8. Cross section and reinforcement detailing.

Figure 9. load method applied to the specimen.

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
505

Figure 10. comparison of non linear finite element analysis and experiment for each beam.

As shown in the figure, the analytical load-displacement relations perfectly match the experimental results for all beams. As shown in Figure
10(a), in case of control beam (specimen A), the
FEM predicts large flexural cracks at the bottom
side of the beam near mid-span region, indicating
the flexural yielding observed in the experiment.
The FEM can also capture horizontal cracks near
loading point, signifying concrete crushing. As for
A-200-P (Figure 10(b)), FEM predicts the shearflexural peeling indicated by concentration of
diagonal cracks at the location between plate end
and loading point, similar to the experiment. In
case of B-200-P (Figure 10(c)), the FEM can also
predict the localization of diagonal crack at the
plate end similar to the experiment.

The development of cracks in A-200-P is

shown in Figure 11. When the load is low (Figure
11(a)), the crack and strain distribution are similar
to those of control beam. As the load increases
(Figure 11(b)), a concentration of strain appears at
the plate end, but does not actively propagate far
enough to bring about failure. As further load
is applied (Figure. 11(c), the active concentration
of strain appears at a new location, which is
approximately half way between plate end and
loading point. This strain concentration is
equivalent to the shear-flexural peeling observed
in the experiment.
The development of cracks in B-200-P is
shown in Figure 12. When the load is low (Figure
12(a)), the crack and strain distribution are similar

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
506

Figure 11. Development of cracks and strain corresponding to shear-flexural peeling (beam
A-200-P) .

Figure 12. Development of cracks and strain corresponding to end-peeling failure (beam B200-P).

Songklanakarin J. Sci. Technol.

Vol. 26 No. 4 Jul.-Aug. 2004

Finite element analysis of FRP-strengthened RC beams

Supaviriyakit, T., et al.
507

to those of control beam. As the load increases

(Figure 12(b)), a concentration of strain appears
at the plate end and actively propagates into the
beam, indicating the end peeling failure observed
in the experiment. From these results, it is seen
that FEM with nonlinear constitutive models can
be used to predict the load-displacement relation,
the load capacity and failure mode of FRPstrengthened beam accurately. It can also capture
the local development of cracking process for both
shear-flexural peeling and end peeling failure
modes.
Conclusion
Finite element analysis using non-linear
constitutive models of cracked concrete, steel bars
and FRP is used to predict the behavior of FRPstrengthened beam. It is verified that the finite
element analysis can accurately predict the loaddeformation, load capacity and failure mode of the
beam. It can also capture cracking process for the
shear-flexural peeling and end peeling failures,
similar to the experiment.
Reference
Arduini, M. and Nanni, A., 1997. Parametric Study of
Beams with Externally Bonded FRP Reinforcement, ACI Struct. J., 94(5): 493-501.

Fukuura, N., Maekawa, K.,1999 Spatially averaged

constitutive law for RC in-plane elements with
non-orthogonal cracking as far as 4-way directions, Proc. of JSCE, Vol.45: 177-195
Hollaway, L.C., Leeming, M.B., 1999. Strengthening of
reinforced concrete structures using externallybonded FRP composites in structural and civil
engineering, CRC Press Ehsani, M.R.
Maekawa, K, Pimanmas, A. and Okamura, H., 2003.
Nonlinear mechanics of reinforced concrete,
Spon Press, London
Okamura, H., Maekawa, K., 1991. Nonlinear analysis
and constitutive models of reinforced concrete,
Gihodo-Shuppan Co., Tokyo
Pornpongsaroj, P., Pimanmas, A, 2003. A. Effect of
end wrapping on peeling behavior of FRPstrengthened beams. Proceedings of the Sixth
International Symposium on Fibre-Reinforced
Polymer Reinforcement for Concrete Structures
(FRPRCS-6)., Singapore, July 8-10, Vol.1: 277286.
Ritchie P.A., Thomas D.A., Lu L.W. and Connelly G.M.,
1991. External reinforcement of concrete beams
using fiber reinforced plastics, ACI Struct. J.,
88(4): 490-500.
Saadatmanesh, H., Ehsani, M.R., 1990. Fiber composite plates can strengthen beams, Conc. Int.,
12(3): 65-71.