Nonlinear nite element analysis of reinforced concrete

beams strengthened by ber-reinforced plastics

Hsuan-Teh Hu
*
, Fu-Ming Lin, Yih-Yuan Jan
Department of Civil Engineering, National Cheng Kung University, Tainan 701, Taiwan
Abstract
Numerical analyses are performed using the ABAQUS nite element program to predict the ultimate loading capacity of
rectangular reinforced concrete beams strengthened by ber-reinforced plastics applied at the bottom or on both sides of these
beams. Nonlinear material behavior, as it relates to steel reinforcing bars, plain concrete, and ber-reinforced plastics is simulated
using appropriate constitutive models. The inuences of ber orientation, beam length and reinforcement ratios on the ultimate
strength of the beams are investigated. It has been shown that the use of ber-reinforced plastics can signicantly increase the
stinesses as well as the ultimate strengths of reinforced concrete beams. In addition, with the same ber-reinforced plastics layer
numbers, the ultimate strengths of beams strengthened by ber-reinforced plastics at the bottom of the beams are much higher than
those strengthened by ber-reinforced plastics on both sides of the beams.
2003 Elsevier Ltd. All rights reserved.
Keywords: Reinforced concrete beams; Strengthened; Fiber-reinforced plastics
1. Introduction
The traditional material used in the strengthening of
concrete structures is steel. Because of its drawbacks of
low corrosion resistance and of handling problems in-
volving excessive size and weight, there is a need for the
engineering community to look for alternatives. Due to
lightweight, high strength and good fatigue and corro-
sion properties, ber-reinforced plastics (FRP) have
been intensively used in the repair and strengthening of
aerospace structures [14]. Though the study of using
FRP to strengthen reinforced concrete structures just
started in the 1990s [515], the technology is currently
widely used.
To study the behavior of reinforced concrete struc-
tures strengthened by FRP, the fundamental step is to
understand the nonlinear behavior of the constitutive
materials, reinforced concrete and FRP, separately. The
nonlinear behavior of reinforced concrete such as con-
crete cracking, tension stiening, shear retention, con-
crete plasticity and yielding of reinforcing steel have
been extensively studied by various researchers and
numerous proper constitutive laws have been proposed
[1621]. However, in the literature, most studies of re-
inforced concrete structures strengthened by FRP have
assumed that the behavior of FRP is linear. It is well
known that unidirectional brous composites exhibit
severe nonlinearity in their in-plane shear stressstrain
relations [22]. In addition, deviation from linearity is
also observed with in-plane transverse loading but the
degree of nonlinearity is not comparable to that ob-
served with the in-plane shear [22,23]. Therefore, ap-
propriate modeling of the nonlinear behavior of FRP
becomes crucial.
In this investigation, proper constitutive models are
introduced to simulate the nonlinear behavior of rein-
forced concrete and FRP. Then the nite element pro-
gram ABAQUS [24] is used to perform a failure analysis
of rectangular reinforced concrete beams strengthened
by FRP. In the numerical analyses, two types of beams,
i.e. a short beam and a long beam, and two types of
reinforcement ratios, i.e. a low and a high steel per-
centage ratio, are considered. The aim of this research is
to study the inuence of beam length, reinforcement
ratio and ber orientation on the globally nonlinear
behavior of rectangular reinforced concrete beams sub-
jected to uniformly distributed load and strengthened by
FRP. In addition, the improvement in stiness and
*
Corresponding author. Tel.: +886-6-2757575x63168; fax: +886-6-
2358542.
E-mail address: [email protected] (H.-T. Hu).
0263-8223/$ - see front matter 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0263-8223(03)00174-0
Composite Structures 63 (2004) 271281
www.elsevier.com/locate/compstruct
strength as well as the alteration of the failure mecha-
nism of the reinforced concrete beams due to FRP are
investigated.
2. Material properties and constitutive models
The materials used in the analysis involve steel rein-
forcing bars, concrete and FRP. Reliable constitutive
models applicable to steel reinforcing bars and concrete
are available in the ABAQUS material library. Thus,
their input material properties and associated constitu-
tive models are only briey discussed. The ABAQUS
program does not have a nonlinear material library for
FRP. Hence, its nonlinear constitutive model is dis-
cussed here in detail. The resulting nonlinear constitu-
tive equations for the FRP are coded in FORTRAN
language as a subroutine and linked to the ABAQUS
program.
2.1. Steel reinforcing bar
The steel reinforcement used in the beam is assumed
to have the yielding stress
r
y
344:7 MPa 50 ksi 1
while its elastic modulus is assumed to be
E
s
199:9 GPa 29000 ksi 2
The stressstrain curve of the reinforcing bar is assumed
to be elastic perfectly plastic as shown in Fig. 1. In
ABAQUS, the steel reinforcement is treated as an
equivalent uniaxial material smeared through out the
element section and the bondslip eect between con-
crete and steel is not considered. In order to properly
model the constitutive behavior of the reinforcement,
the cross sectional area, spacing, position and orienta-
tion of each layer of steel bar within each element needs
to be specied.
2.2. Concrete
The concrete has an uniaxial compressive strength f
0
c
selected as
f
0
c
34:47 MPa 5 ksi 3
Under uniaxial compression, the concrete strain e
o
cor-
responding to the peak stress f
0
c
is usually around the
range of 0.0020.003. A representative value suggested
by ACI Committee 318 [25] and used in the analysis is
e
o
0:003 4
The Poissons ratio m
c
of concrete under uniaxial
compressive stress ranges from about 0.150.22, with a
representative value of 0.19 or 0.20 [16]. In this study,
the Poissons ratio of concrete is assumed to be
m
c
0:2 5
The uniaxial tensile strength f
0
t
of concrete is dicult
to measure. For this study the value is taken as [16]
f
0
t
0:33

f
0
c
_
MPa 6
The initial modulus of elasticity of concrete E
c
is
highly correlated to its compressive strength and can be
calculated with reasonable accuracy from the empirical
equation [25]
E
c
4700

f
0
c
_
MPa 7
Under multiaxial combinations of loading, the failure
strengths of concrete are dierent from those observed
under uniaxial condition. However, the maximum
strength envelope under multiple stress conditions seems
to be largely independent of load path [26]. In ABA-
QUS, a MohrCoulomb type compression surface to-
gether with a crack detection surface is used to model
the failure surface of concrete (Fig. 2). When the prin-
cipal stress components of concrete are predominantly
compressive, the response of the concrete is modeled by
Fig. 1. Elastic perfectly plastic model for steel reinforcing bar. Fig. 2. Concrete failure surface in plane stress.
272 H.-T. Hu et al. / Composite Structures 63 (2004) 271281
an elasticplastic theory with an associated ow and an
isotropic hardening rule. In tension, once cracking is
dened to occur (by the crack detection surface), the
orientation of the crack is stored. Damaged elasticity is
then used to model the existing crack [24].
When plastic deformation occurs, there should be a
certain parameter to guide the expansion of the yield
surface. A commonly used approach is to relate the
multidimensional stress and strain conditions to a pair
of quantities, namely, the eective stress r
c
and eective
strain e
c
, such that results obtained following dierent
loading paths can all be correlated by means of the
equivalent uniaxial stressstrain curve. The stressstrain
relationship proposed by Saenz [27] has been widely
adopted as the uniaxial stressstrain curve for concrete
and it has the following form:
r
c

E
c
e
c
1 R R
E
2
e
c
eo
_ _
2R 1
e
c
eo
_ _
2
R
e
c
eo
_ _
3
8
where
R
R
E
R
r
1
R
e
1
2

1
R
e
; R
E

E
c
E
o
; E
o

f
0
c
e
o
and R
r
4, R
e
4 may be used [20]. In the analysis, Eq.
(8) is taken as the equivalent uniaxial stressstrain curve
for concrete and approximated by several piecewise
linear segments as shown in Fig. 3.
When cracking of concrete takes place, a smeared
model is used to represent the discontinuous macrocrack
behavior. It is known that the cracked concrete of a
reinforced concrete element can still carry some tensile
stress in the direction normal to the crack, which is
termed tension stiening [16]. In this study, a simple
descending line is used to model this tension stiening
phenomenon (Fig. 4). The default value of the strain e

at which the tension stiening stress reduced to zero is

[24]
e

0:001 9
During the postcracking stage, the cracked reinforced
concrete can still transfer shear forces through aggregate
interlock or shear friction, which is termed shear reten-
tion. Assuming that the shear modulus of intact con-
crete is G
c
, then the reduced shear modulus

GG of cracked
concrete can be expressed as

GG lG
c
10
l 1 e=e
max
11
where e is the strain normal to the crack direction and
e
max
is the strain at which the parameter l reduces to
zero (Fig. 5). Numerous analytical results have demon-
strated that the particular value chosen for l (between 0
and 1) does not appear to be critical but values greater
than zero are necessary to prevent numerical instabilities
[16,21]. In ABAQUS, e
max
is usually assumed to be a
very large value, i.e., l 1 (full shear retention). In this
investigation, the default values for tension stiening
parameter e

0:001 and for shear retention parameter

l 1 are used.
2.3. Fiber-reinforced plastics
For ber-reinforced plastics (Fig. 6), each lamina can
be considered as an orthotropic layer in a plane stress
condition. It is well known that unidirectional brous
composites exhibit severe nonlinearity in their in-plane
shear stressstrain relation. In addition, deviation from
linearity is also observed with in-plane transverse load-
ing but the degree of nonlinearity is not comparable to
0
10
20
30
40
0 0.001 0.002 0.003 0.004 0.005

c

(
M
P
a
)

c
Fig. 3. Equivalent uniaxial stressstrain curve for concrete.
Fig. 4. Tension stiening model.
Fig. 5. Shear retention parameter.
H.-T. Hu et al. / Composite Structures 63 (2004) 271281 273
that in the in-plane shear [22]. Usually, this nonlinearity
associated with the transverse loading can be ignored
[23]. To model the nonlinear in-plane shear behavior,
the nonlinear strainstress relation for a composite
lamina suggested by Hahn and Tsai [22] is adopted.
Values are given as follows:
e
1
e
2
c
12
_
_
_
_
_
_

1
E
11

m
21
E
22
0

m
12
E
11
1
E
22
0
0 0
1
G
12
_

_
_

_
r
1
r
2
s
12
_
_
_
_
_
_
S
6666
s
2
12
0
0
s
12
_
_
_
_
_
_
12
In this model only one constant S
6666
is required to ac-
count for the in-plane shear nonlinearity. The value of
S
6666
can be determined by a curve t to various o-axis
tension test data [22]. Let us dene Dfr
0
g
Dfr
1
; r
2
; s
12
g
T
and Dfe
0
g Dfe
1
; e
2
; c
12
g
T
. Inverting and
dierentiating Eq. (12), the incremental stressstrain
relations are established
Dfr
0
g Q
0
1
Dfe
0
g 13
Q
0
1

E
11
1m
12
m
21
m
12
E
22
1m
12
m
21
0
m
21
E
11
1m
12
m
21
E
22
1m
12
m
21
0
0 0
1
1=G
12
3S
6666
s
2
12
_

_
_

_ 14
Furthermore, it is assumed that the transverse shear
stresses always behave linearly and do not aect the
nonlinear behavior of any in-plane shear. If we dene
Dfs
0
t
g Dfs
13
; s
23
g
T
and Dfc
0
t
g Dfc
13
; c
23
g
T
, the con-
stitutive equations for transverse shear stresses become
Dfs
0
t
g Q
0
2
Dfc
0
t
g 15
Q
0
2

a
1
G
13
0
0 a
2
G
23
_ _
16
where a
1
and a
2
are the shear correction factors and are
taken to be 0.83 in this study.
Among existing failure criteria, the TsaiWu criterion
[28] has been extensively used in the literature and is
adopted in this analysis. Under plane stress conditions,
this failure criterion has the following form:
F
1
r
1
F
2
r
2
F
11
r
2
1
2F
12
r
1
r
2
F
22
r
2
2
F
66
s
2
12
1
17
with
F
1

1
X

1
X
0
; F
2

1
Y

1
Y
0
; F
11

1
XX
0
; F
22

1
Y Y
0
;
F
66

1
S
2
The X, Y and X
0
, Y
0
are the lamina longitudinal and
transverse strengths in tension and compression, re-
spectively, and S is the shear strength of the lamina.
Though the stress interaction term F
12
in Eq. (17) is
dicult to be determined, it has been suggested that F
12
can be set equal to zero for practical engineering ap-
plications [29]. Therefore, F
12
0 is used in this inves-
tigation.
During the numerical calculation, incremental load-
ing is applied to composite plates until failures in one or
more of individual plies are indicated according to Eq.
(17). Since the TsaiWu criterion does not distinguish
failure modes, the following two rules are used to de-
termine whether the ply failure is caused by resin frac-
ture or ber breakage [30]:
(1) If a ply fails but the stress in the ber direction re-
mains less than the uniaxial strength of the lamina
in the ber direction, i.e. X
0
< r
1
< X, the ply failure
is assumed to be resin induced. Consequently, the
laminate loses its capability to support transverse
and shear stresses, but remains to carry longitudinal
stress. In this case, the constitutive matrix of the
lamina becomes
Q
0
1

E
11
0 0
0 0 0
0 0 0
_
_
_
_
18
(2) If a ply fails with r
1
exceeding the uniaxial strength
of the lamina, the ply failure is caused by the ber
breakage and a total ply rupture is assumed. In this
case, the constitutive matrix of the lamina becomes
Q
0
1

0 0 0
0 0 0
0 0 0
_
_
_
_
19
The material properties for FRP used in the analysis
are E
11
138 GPa, E
22
14:5 GPa, G
12
G
13

5:86 GPa, G
23
3:52 GPa, S
6666
7:32 (GPa)
3
,
X 1450 MPa, X
0
1450 MPa, Y 52 MPa,
Y
0
206 MPa, S 93 MPa, m
12
0:21.
Fig. 6. Material, element and structure coordinates of ber reinforced
plastics.
274 H.-T. Hu et al. / Composite Structures 63 (2004) 271281
During a nite element analysis, the constitutive
matrix of composite materials at the integration points
of shell elements must be calculated before the stiness
matrices are assembled from the element level to the
structural level. For composite materials, the incre-
mental constitutive equations of a lamina in the element
coordinates (x; y; z) can be written as
Dfrg Q
1
Dfeg 20
Dfs
t
g Q
2
Dfc
t
g 21
where Dfrg Dfr
x
; r
y
; s
xy
g
T
, Dfs
t
g Dfs
xz
; s
yz
g
T
,
Dfeg Dfe
x
; e
y
; c
xy
g
T
, Dfc
t
g Dfc
xz
, c
yz
g
T
, and
Q
1
T
1

T
Q
0
1
T
1
22
Q
2
T
2

T
Q
0
2
T
2
23
T
1

cos
2
h sin
2
h sin h cos h
sin
2
h cos
2
h sin h cos h
2 sin h cos h 2 sin h cos h cos
2
h sin
2
h
_
_
_
_
24
T
2

cos h sin h
sin h cos h
_ _
25
The h is measured counterclockwise from the element
local x-axis to the material 1-axis (Fig. 6). Assume
Dfe
o
g Dfe
xo
; e
yo
; c
xyo
g
T
are the incremental in-plane
strains at the midsurface of the shell section and Dfjg
Dfj
x
; j
y
; j
xy
g
T
are its incremental curvatures. The in-
cremental in-plane strains at a distance z from the
midsurface of the shell section become
Dfeg Dfe
o
g zDfjg 26
Let h be the total thickness of the composite shell
section, the incremental stress resultants, DfNg
DfN
x
; N
y
; N
xy
g
T
, DM DfM
x
; M
y
; M
xy
g
T
and DfV g
DfV
x
; V
y
g, can be dened as
DfNg
DfMg
DfV g
_
_
_
_
_
_

_
h=2
h=2
Dfrg
zDfrg
Dfs
t
g
_
_
_
_
_
_
dz 27
Substituting Eqs. (20), (21) and (26) into the above ex-
pression, one can obtain the stiness matrix for the ber
composite laminate shell at the integration point as
DfNg
DfMg
DfV g
_
_
_
_
_
_

_
h=2
h=2
Q
1
zQ
1
0
zQ
1
z
2
Q
1
0
0
T
0
T
Q
2

_
_
_
_
Dfe
o
g
Dfjg
Dfc
t
g
_
_
_
_
_
_
dz
28
where [0] is a 3 2 null matrix.
3. Verication of the proposed material constitutive
models
The validity of the material models for steel, concrete
and FRP has been veried individually by testing
against experimental data [24,31] and is not duplicated
here. The validity of the these material models to sim-
ulate the composite behavior of reinforced concrete
beam strengthened by FRP is examined in this section
by comparing with the result of beam experiment per-
formed by Shahawy et al. [10]. The dimensions of the
test beam are given in Fig. 7. The beam is subjected to
four-point static load up to failure. The exural rein-
forcement is composed of two 13 mm diameter steel bars
in tension zone and two 3 mm diameter steel bars in
compression zone. The yielding strength and the elastic
modulus of the reinforcing steel are r
y
468:8 MPa and
E
s
199:9 GPa. The compressive strength and the
Poissons ratio of concrete are f
0
c
41:37 MPa and
m
c
0:2. Three FRP layers with their ber directions
oriented in the axial direction of the beam are adhered
to the bottom face of the beam. Each FRP layer is
0.1702 mm in thickness with tensile strength X 2758
MPa and modulus E
11
141:3 GPa. To take the Tsai
Wu criterion into account, the following parameters are
assumed: X
0
2758 MPa, Y 52 MPa, Y
0
206
MPa, S 93 MPa, E
22
14:5 GPa, G
12
G
13
5:86
GPa, G
23
3:52 GPa, S
6666
7:32 (GPa)
3
, m
12
0:21.
Since the FRP layers are subjected to uniaxial tension in
ber direction only, these assumed parameters would
not aect the uniaxial tensile behavior of the FRP.
The beam has two planes of symmetry. One plane of
symmetry is the xy plane cutting beam in half longi-
tudinally. The other plane of symmetry is the yz plane
cutting beam in half transversely. Due to symmetry,
only 1/4 portion of the beam is analyzed and symmetric
boundary conditions are placed along the two symmet-
ric planes. In the nite element analysis, 8-node solid
elements (three degrees of freedom per node) are used to
model the reinforced concrete beams. The 1/4 beam
mesh has 78 solid elements in total (26 rows in x-direc-
tion, 3 rows in y-direction and 1 row in z-direction).
Because the ber-reinforced plastics are relatively thin
compared to the concrete beam, they are modeled by the
Fig. 7. Details of test beam.
H.-T. Hu et al. / Composite Structures 63 (2004) 271281 275
4-node shell elements (six degrees of freedom per node).
The FRP shell elements are attached to the bottom
surface of the concrete beam directly and perfect
bonding between FRP and the concrete is assumed.
Fig. 8 shows the moment versus deection curves of
the beam at the midspan. It can be observed that the
correlation is quite good between the numerical result
and the experimental data. The predicted ultimate mo-
ment 60.9 kNm is in good agreement with the experi-
mental ultimate moment 60.4 kNm. The error is only
about 0.8%. Hence, the proposed material constitutive
models are proved to be able to simulate the composite
behavior of reinforced concrete beam strengthened by
FRP correctly.
4. Numerical analysis
4.1. Beam geometry and nite element model
In the numerical analyses, simply supported rein-
forced concrete beams with two types of lengths, i.e.,
short beam and long beam, are considered (Fig. 9).
While the deection of the long beam is primary caused
by bending, the deection of the short beam is due to
both bending and shear [32]. To study the inuence of
reinforcement ratio, two types of reinforcement ratios,
i.e. low reinforcement ratio and high reinforcement
ratio, are considered. Two #4 steel bars q 0:0066 are
used for beams with low reinforcement ratio and two #8
steel bars q 0:0264 are used for beams with high
reinforcement ratio. Both high and low reinforcement
ratios satisfy the requirement of ACI code [25], i.e.
1:4=r
y
6q60:75q
b
, where q
b
0:108 is the reinforce-
ment ratio for the balanced strain condition. These
beams are subjected to a uniformly distributed load p
(force per unit area) at the top surface of the beam and
the weights of the beams are neglected. The material
properties for steel, concrete and FRP discussed in
Section 2 are used in the numerical analyses.
These beams again have two planes of symmetry.
Therefore, only 1/4 portion of each beam is analyzed
and symmetric boundary conditions are placed along
the two symmetric planes. In the nite element analysis,
27-node solid elements (three degrees of freedom per
node) are used to model the reinforced concrete beams.
Based on the results of convergent studies [32], it was
decided to use 72 elements (18 rows in x-direction, 4
rows in y-direction and 1 row in z-direction) for long
beams and 36 elements (9 rows in x-direction, 4 rows in
y-direction and 1 row in z-direction) for short beams.
The FRP are modeled by the 8-node shell elements (six
degrees of freedom per node) and attached to the outer
surface of the concrete beams directly.
4.2. Ultimate analysis of reinforced concrete beams
without strengthening FRP
In order to provide a base to make a comparison or
show how the FRP changes the beam, ultimate analyses
of ordinary reinforced concrete beams without any FRP
are carried out. Fig. 10 shows the uniformly distributed
load p versus the midspan deection of the beams. The
rst character L or S in the gure represents long beam
or short beam, respectively. The following numbers 4 or
8 stand for beams with #4 or #8 steel bars. From the
gure one can observe that the stiness and the ultimate
load of the long beams (L4 and L8) are much lower than
those of the short beams (S4 and S8). This is because the
long beams are weaker in bending than the short beams.
Generally, the reinforcement ratio does not inuence the
ultimate load of beams signicantly. For example, the
ultimate load p
u
of L8 beam (71.02 kPa) is higher than
that of L4 beam (68.53 kPa) by 3.63% and the ultimate
load of S8 beam (151.68 kPa) is higher than that of S4
beam (146.17 kPa) by 3.77%. For the long beams, the
beam with low reinforcement ratio (L4) has more ductile
behavior near the ultimate loading stage than that with
high reinforcement ratio (L8). However, for the short
beams, the reinforcement ratio does not inuence their
0
10
20
30
40
50
60
70
0 5 10 15 20 25
Experimental data
Numerical result
M
i
d
s
p
a
n

m
o
m
e
n
t

(
k
N
-
m
)
Midspan deflection (mm)
Fig. 8. Comparison of numerical and experimental results.
Fig. 9. Details of beams in numerical analysis.
276 H.-T. Hu et al. / Composite Structures 63 (2004) 271281
behaviors prior to the ultimate loading stage signi-
cantly.
Fig. 11 shows the crack patterns of all four types of
beams under ultimate loads. The black dots in the gure
indicate that the integration points of the concrete ele-
ments have cracks. It can be seen that the long beams
fail by bending and numerous cracks take place in the
bottom of the central region of the beams. In addition,
the beam with low reinforcement ratio (L4) would have
more cracks than that with high reinforcement ratio
(L8). The short beams fail by shear and cracks take
place near the bottom of the support area. Unlike the
long beams, the crack patterns for S4 and S8 beams are
very similar. Hence, it can be conrmed again that the
reinforcement ratio does not inuence the behaviors of
short beams prior to the ultimate loading stage signi-
cantly.
4.3. Ultimate analysis of reinforced concrete beams
strengthened by FRP at the bottom
To increase the bending resistance of the reinforced
concrete beams, we consider attaching the FRP to the
bottom of the beams in this section. The thickness of
each FRP layer is 1 mm (0.04 in) and the laminate
lay-ups are [0]
n
, where n 1, 2, 3, 4. The ber angle of
the lamina is measured counterclockwise (through
outward normal direction) from the longitudinal di-
rection of the beams. The reason that all the bers are
placed in the axial direction of beam is because FRP
has the highest stiness and strength in its ber di-
rection.
Figs. 12 and 13 show the uniformly distributed load p
versus the midspan deection of reinforced concrete
beams strengthened by FRP. Generally, the stinesses of
the beams increase when the numbers of FRP layers are
Fig. 11. Crack patterns of reinforced concrete beams without
strengthening FRP and under ultimate loads.
0
50
100
150
200
250
300
350
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
without FRP
n = 1
n = 2
n = 3
n = 4
p

(
k
P
a
)
Midspan deflection (cm)
(a) L4
0
100
200
300
400
500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
without FRP
n = 1
n = 2
n = 3
n = 4
p

(
k
P
a
)
Midspan deflection (cm)
(b) L8
Fig. 12. Loaddeection curves of long reinforced concrete beams
strengthened by [0]
n
FRP at the bottom.
0
20
40
60
80
100
120
140
160
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
L4
L8
S4
S8
p

(
k
P
a
)
Midspan deflection (cm)
Fig. 10. Loaddeection curves of reinforced concrete beams without
strengthening FRP.
H.-T. Hu et al. / Composite Structures 63 (2004) 271281 277
increased. Fig. 14 shows the increasing of the ultimate
load p
u
versus the numbers of FRP layers at the bottom
of beams. For the long beams with low reinforcement
ratio (L4), p
u
seems to increase linearly with the number
of FRP layers (for n < 4). For the other three types of
beams, L8, S4 and S8, the use of one FRP layer would
have the most signicant eect in increasing the p
u
.
When the numbers of FRP layers are increased, this
increase in p
u
seems to approach constant values (say
500% for L8 and S8 beams; 350% for S4 beams) and
becomes less signicant than for the rst FRP layer. It
can be seen that the curves of L8 and S8 in Fig. 14 are
almost identical and that the trends of the loaddeec-
tion curves in Figs. 12(b) and 13(b) are similar. This may
indicate that the behaviors of the beams with high re-
inforcement ratio and strengthened with FRP are not
inuenced by the length of beam signicantly. However,
for beams with low reinforcement ratio and strength-
ened with FRP, the beam lengths do aect their be-
haviors signicantly, as shown by Figs. 12(a), 13(a) and
14.
Fig. 15 shows the crack patterns of reinforced con-
crete beams strengthened by [0]
4
FRP at the bottom
and under ultimate loads. Comparing Fig. 15 with Fig.
11, one could see that after FRP is employed at the
bottom of the beams, these beams are failed in a
combination of bending and shear modes, i.e., severe
cracks occur at the bottom of the beam from the
central region through out the support area. Generally,
the beams with high reinforcement ratios and
strengthened with FRP would have more cracks at the
central region than those with low reinforcement ra-
tios. On the other hand, the beams with low rein-
forcement ratios and strengthened with FRP would
have more cracks at the support area than those with
high reinforcement ratios.
0
100
200
300
400
500
600
0 1 2 3 4
L4
L8
S4
S8
i
n
c
r
e
a
s
e

i
n

p
u

(
%
)
Numbers of FRP layers
Fig. 14. Increase of p
u
versus numbers of FRP layers for reinforced
concrete beams strengthened by [0]
n
FRP at the bottom.
Fig. 15. Crack patterns of reinforced concrete beams strengthened by
[0]
4
FRP at the bottom and under ultimate loads.
0
100
200
300
400
500
600
700
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
without FRP
n = 1
n = 2
n = 3
n = 4
p

(
k
P
a
)
Midspan deflection (cm)
(a) S4
0
200
400
600
800
1000
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
without FRP
n = 1
n = 2
n = 3
n = 4
p

(
k
P
a
)
Midspan deflection (cm)
(b) S8
Fig. 13. Loaddeection curves of short reinforced concrete beams
strengthened by [0]
n
FRP at the bottom.
278 H.-T. Hu et al. / Composite Structures 63 (2004) 271281
4.4. Ultimate analysis of reinforced concrete beams
strengthened by FRP on both sides
To increase the shear resistance of the reinforced
concrete beams, we consider attaching the FRP to both
sides of the beams in this session. The thickness of each
FRP layer is the same as before and the laminate lay-ups
are h
n
, where n 1, 2, 3. The ber angle of the
lamina is measured counterclockwise from the midsur-
face of the beams.
Figs. 16 and 17 show the typical loaddeection
curves of long beams (L4 and L8) and short beams (S4
and S8) strengthened by h
3
FRP on both sides, re-
spectively. From these gures one can observe that when
h angle is close to 0, the beams have the strongest
stinesses. When h angle is close to 90, the beams are
prone to have the weakest stinesses.
Figs. 18 and 19 show the increase of the ultimate load
p
u
versus ber angle h for beams with h
n
FRP on both
sides. Generally, the ultimate load p
u
increases with the
increasing of FRP layer numbers. For long beams with
low reinforcement ratio as shown in Fig. 18(a), when
n 1 and 2, the increasing in p
u
seems to be less inde-
pendent on the ber angle h. However, when n 3, the
ber angle does have signicant inuence on the ulti-
mate load p
u
and the optimal angle seems to be around
0
30
60
90
120
150
0 0.1 0.2 0.3 0.4 0.5
without FRP
= 0
= 30
= 60
= 90
p

(
k
P
a
)
Midspan deflection (cm)
(a) L4
0
30
60
90
120
150
0 0.1 0.2 0.3 0.4 0.5
without FRP
= 0
= 30
= 60
= 90
p

(
k
P
a
)
Midspan deflection (cm)
(b) L8
Fig. 16. Loaddeection curves of long reinforced concrete beams
strengthened by h
3
FRP on both sides.
0
50
100
150
200
250
300
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
without FRP
= 0
= 30
= 60
= 90
p

(
k
P
a
)
Midspan deflection (cm)
(a) S4
0
50
100
150
200
250
300
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
without FRP
= 0
= 30
= 60
= 90
p

(
k
P
a
)
Midspan deflection (cm)
(b) S8
Fig. 17. Loaddeection curves of short reinforced concrete beams
strengthened by h
3
FRP on both sides.
0
20
40
60
80
100
0 15 30 45 60 75 90
n = 1
n = 2
n = 3
i
n
c
r
e
a
s
e

i
n

p
u

(
%
)
(degrees)
(a) L4
0
20
40
60
80
100
0 15 30 45 60 75 90
n = 1
n = 2
n = 3
i
n
c
r
e
a
s
e

i
n

p
u

(
%
)
(degrees)
(b) L8
Fig. 18. Increase of p
u
versus h for long reinforced concrete beams
strengthened by h
n
FRP on both sides.
H.-T. Hu et al. / Composite Structures 63 (2004) 271281 279
60. For long beams with high reinforcement ratio as
shown in Fig. 18(b), the increase in p
u
is less dependent
on the ber angle h only for the case with n 1. For
n 2 and 3, the ultimate loads are highly dependent on
the ber angles. For short beams with n 2 and 3 as
shown in Fig. 19(a) and (b), the ultimate loads are also
highly dependent on the ber angles. For short beams
with n 1, the ultimate loads are less dependent on the
ber angles when the ber angles are large, say h > 15
for short beams with low reinforcement ratio (S4) and
h > 30 for short beams with high reinforcement ratio
(S8). No matter of the reinforcement ratio and the FRP
layer numbers, the optimal ber angle of short beams
seems to close to 0.
Comparing Figs. 18 and 19 with Fig. 14, one can
observe that with the same numbers of FRP layers, the
ultimate strengths of beams strengthened by FRP on
both sides of beams are much less than those strength-
ened by FRP at the bottom of beams. This indicates that
to increase the bending resistance of the reinforced
concrete beams is more crucial than to increase the shear
resistance of the beams.
Fig. 20 shows the crack patterns of reinforced con-
crete beams strengthened by 45
3
FRP on both sides
and under ultimate loads. Comparing Fig. 20 with Fig.
11, one can see that after FRP is employed on both sides
of the beams, the long beams develop more cracks from
central region toward the support area. For short
beams, they start to develop cracks at the central-bot-
tom region of the beams. Comparing Fig. 20 with Fig.
15, it can be seen that the beams with FRP on both sides
have less cracks under the ultimate loads that those with
FRP at the bottom. This is because that the ultimate
strengths of the former beams are less than the latter
ones.
5. Conclusions
In this paper, nonlinear nite element analyses of
rectangular reinforced concrete beams strengthened by
FRP are performed. Based on the numerical results, the
following conclusions may be drawn:
(1) The behaviors of the beams with high reinforcement
ratio and strengthened with FRP at the bottom are
not inuenced by the length of beam signicantly.
(2) For beams with low reinforcement ratio and
strengthened with FRP at the bottom, the beam
lengths do aect their behaviors signicantly.
(3) The beams with high reinforcement ratios and
strengthened with FRP at the bottom would have
more cracks at the central region than those with
Fig. 20. Crack patterns of reinforced concrete beams strengthened by
45
3
FRP on both sides and under ultimate loads.
0
20
40
60
80
100
120
0 15 30 45 60 75 90
n = 1
n = 2
n = 3
i
n
c
r
e
a
s
e

i
n

p
u

(
%
)
(degrees)
(a) S4
0
20
40
60
80
100
120
0 15 30 45 60 75 90
n = 1
n = 2
n = 3
i
n
c
r
e
a
s
e

i
n

p
u

(
%
)
(degrees)
(b) S8
Fig. 19. Increase of p
u
versus h for short reinforced concrete beams
strengthened by h
n
FRP on both sides.
280 H.-T. Hu et al. / Composite Structures 63 (2004) 271281
low reinforcement ratios. On the other hand, the
beams with low reinforcement ratios and strength-
ened with FRP at the bottom would have more
cracks at the support area than those with high rein-
forcement ratios.
(4) For long beams strengthened by h
n
FRP on both
sides, when the FRP layer numbers is small, the in-
crease in the ultimate load p
u
seems to be less depen-
dent on the ber angle h.
(5) For short beams strengthened by h
n
FRP on both
sides, the optimal ber angle seems to be 0 no mat-
ter of the reinforcement ratio and the numbers of
FRP layers.
(6) With the same FRP layer numbers, the ultimate
strengths and the numbers of cracks of beams
strengthened by FRP on both sides are much less
than those strengthened by FRP at the bottom.
Thus, to increase the bending resistance of the rein-
forced concrete beams is more crucial than to in-
crease the transverse shear resistance of the beams.
Acknowledgements
This research work was nancially supported by the
National Science Council, Republic of China under
Grant NSC 88-2211-E-006-014.
References
[1] Baker AA. Fiber composite repair of cracked metallic aircraft
componentspractical and basic aspects. Composite
1987;18(4):293308.
[2] Ong CL, Chu RC, Ko TC, Shen SB. Composite patch reinforce-
ment of cracked aircraft upper longeron: analysis and specimen
simulation. Theor Appl Fract Mech 1990;14:1326.
[3] Naboulsi S, Mall S. Modeling of a cracked metallic structure with
bonded composite patch using the three layer technique. Compos
Struct 1996;35:295308.
[4] Schubbe JJ, Mall S. Investigation of a cracked thick aluminum
panel repaired with a bonded composite patch. Eng Fract Mech
1999;63:30523.
[5] Ritchie PA, Thomas DA, Lu L-W, Connelly GM. External
reinforcement of concrete beams using ber reinforced plastics.
ACI Struct J 1991;88(4):490500.
[6] Saadatmanesh H, Ehsani MR. RC beams strengthened with FRP
plates II: analysis and parametric study. J Struct Eng (ASCE)
1991;117(11):343455.
[7] Nanni A. Flexural behavior and design of RC members using
FRP reinforcement. J Struct Eng (ASCE) 1993;119(11):334459.
[8] Sharif A, Al-Sulaimani GJ, Basunbul IA, Baluch MH, Ghaleb
BN. Strengthening of initially loaded reinforced concrete beams
using FRP plates. ACI Struct J 1994;91(2):1607.
[9] Chajes MJ, Januszka TF, Mertz DR, Thomson TA, Finch WW.
Shear strengthening of reinforced concrete beams using externally
applied composite fabrics. ACI Struct J 1995;92(3):295303.
[10] Shahawy MA, Arockiasamy M, Beitelman T, Sowrirajan R.
Reinforced concrete rectangular beams strengthened with CFRP
laminates. Compos Part B: Eng 1996;27(34):22533.
[11] Chambers RE. ASCE design standard for pultruded ber-
reinforced-plastic (FRP) structures. J Compos Construct (ASCE)
1997;1(1):2638.
[12] Malek AM, Saadatmanesh H. Analytical study of reinforced
concrete beams strengthened with web-bonded ber reinforced
plastic plates or fabrics. ACI Struct J 1998;95(3):34352.
[13] Mukhopadhyaya P, Swamy N, Lynsdale C. Optimizing structural
response of beams strengthened with GFRP plates. J Compos
Construct (ASCE) 1998;2(2):8795.
[14] Tedesco JW, Stallings JM, El-Mihilmy M. Finite element method
analysis of a concrete bridge repaired with ber reinforced plastic
laminates. Comput Struct 1999;72:379407.
[15] Vecchio FJ, Bucci F. Analysis of repaired reinforced concrete
structures. J Struct Eng (ASCE) 1999;125(6):64452.
[16] ASCE Task Committee on Concrete and Masonry Structure.
State of the art report on nite element analysis of reinforced
concrete, ASCE, 1982.
[17] Chen WF. Plasticity in reinforced concrete. McGraw-Hill; 1982.
[18] Meyer C, Okamura H. Finite element analysis of reinforced
concrete structures. ASCE; 1985.
[19] Vecchio FJ, Collins MP. The modied compression-eld theory
for reinforced concrete elements subjected to shear. ACI J
1986;83:21931.
[20] Hu H-T, Schnobrich WC. Constitutive modelling of concrete by
using nonassociated plasticity. J Mater Civil Eng (ASCE)
1989;1(4):199216.
[21] Hu H-T, Schnobrich WC. Nonlinear analysis of cracked rein-
forced concrete. ACI Struct J 1990;87(2):199207.
[22] Hahn HT, Tsai SW. Nonlinear elastic behavior of unidirectional
composite laminae. J Compos Mater 1973;7:10218.
[23] Jones RM, Morgan HS. Analysis of nonlinear stressstrain
behavior of ber-reinforced composite materials. AIAA J
1977;15:166976.
[24] Hibbitt, Karlsson, and Sorensen, Inc. ABAQUS Theory Manual,
User Manual and Example Manual, Version 5.8, Providence, RI,
2000.
[25] ACI Committee 318. Building Code Requirements for Structural
Concrete and Commentary (ACI 318-99), American Concrete
Institute, Detroit, MI, 1999.
[26] Kupfer H, Hilsdorf HK, Rusch H. Behavior of concrete under
biaxial stresses. ACI J 1969;66:65666.
[27] Saenz LP. Discussion of Equation for the stressstrain curve of
concrete by Desayi P, Krishnan S. ACI J 1964;61:122935.
[28] Tsai SW, Wu EM. A general theory of strength for anisotropic
materials. J Compos Mater 1971;5:5880.
[29] Narayanaswami R, Adelman HM. Evaluation of the tensor
polynomial and Homan strength theories for composite mate-
rials. J Compos Mater 1977;11:36677.
[30] Rowlands RE. Strength (failure) theories and their experimental
correlation. In: Sih GC, Skudra AM, editors. Failure mechanics
of composites. The Netherlands: Elsevier Science Publishers; 1985.
p. 71125.
[31] Lin W-P, Hu H-T. Nonlinear analysis of ber-reinforced com-
posite laminates subjected to uniaxial tensile load. J Compos
Mater 2002;36(12):142950.
[32] Jan Y-Y. Strengthening of rectangular reinforced concrete beams
with ber-reinforced composite laminated materials. MS thesis,
Department of Civil Engineering, National Cheng Kung Univer-
sity, Tainan, Taiwan, ROC, 1997.
H.-T. Hu et al. / Composite Structures 63 (2004) 271281 281