Engineering Structures 24 (2002) 385395

www.elsevier.com/locate/engstruct
FRP-strengthened RC beams. I: review of debonding strength
models
S.T. Smith, J.G. Teng
*
Department of Civil and Structural Engineering, The Hong Kong Polytechnic University, Hong Kong, Peoples Republic of China
Received 11 April 2001; received in revised form 5 September 2001; accepted 12 September 2001
Abstract
Bonding of a bre-reinforced polymer (FRP) plate to the tension face of a beam has become a popular exural strengthening
method in recent years. As a result, a large number of studies have been carried out in the last decade on the behaviour of these
FRP-strengthened beams. Many of these studies reported premature failures by debonding of the FRP plate with or without the
concrete cover attached. The most commonly reported debonding failure occurs at or near the plate end, by either separation of
the concrete cover or interfacial debonding of the FRP plate from the RC beam. This and the companion paper are concerned with
strength models for such plate end debonding failures. In this paper, a comprehensive review of existing plate debonding strength
models is presented. Each model is summarised and classied into one of the three categories based on the approach taken, and
its theoretical basis claried. The review not only brings together for the rst time all existing plate end debonding strength models
into a unied framework for future reference, but also provides the necessary background information for them to be assessed in
the companion paper using a large test database assembled by the authors from the published literature. 2002 Published by
Elsevier Science Ltd.
Keywords: Debonding; FRP; RC beams; Retrotting; Strengthening; Strength models
1. Introduction
The exural strength of a reinforced concrete (RC)
beam can be increased by bonding a bre reinforced
polymer (FRP) plate to the tension face, or to the soft
in the more restricted but commonly considered case of
simply supported beams (Fig. 1). This paper is also
explicitly concerned only with soft plating of simply-
Fig. 1. RC beam bonded with an FRP soft plate.
* Corresponding author. Tel.: +852-2334-6389; fax: +852-2766-
6012.
E-mail address: cejgteng@pdyu.edu.hk (J.G. Teng).
0141-0296/02/$ - see front matter 2002 Published by Elsevier Science Ltd.
PII: S0141- 0296( 01) 00105- 5
supported beams for simplicity in description, although
the conclusions are applicable to tension face plating of
beams in general.
This technique of exural strengthening has numerous
advantages such as minimum increases in structural size
and weight, ease of site handling and good corrosion
resistance, and has been widely applied in practice in
recent years. As a result, a great amount of research has
been carried out in recent years on the behaviour and
strength of these FRP-strengthened RC beams, with the
vast majority of this research being undertaken in the
past decade (e.g. [124]). This research has identied a
386 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
number of failure modes. A schematic representation of
the six main failure modes observed in tests is shown in
Fig. 2. These are herein termed (a) exural failure by
FRP rupture, (b) exural failure by crushing of com-
pressive concrete (c) shear failure, (d) concrete cover
separation, (e) plate end interfacial debonding, and (f)
intermediate crack induced interfacial debonding.
Among the six failure modes shown in Fig. 2, the rst
three are not totally different from those in conventional
RC beams, although there are some important differ-
ences [25]. The three failure modes shown on the right
are not found in conventional RC beams and are instead
modes unique to beams bonded with a soft plate. These
modes have often been referred to as premature debond-
ing failures modes, as they occur before the exural fail-
ure of the section in mode (a) or (b) or the shear failure
in mode (c) occurs.
The three modes of debonding shown in Figs. 2df
can be broadly classied into two types [25]: (a) those
that initiate at or near one of the plate ends (simply
referred to as the plate end hereafter) and then propagate
away from the plate end; and (b) those that initiate at
an intermediate exural or exural-shear crack and then
propagate from such a crack towards the plate end. The
rst type of debonding is referred to as plate end
debonding here and the second is referred to as inter-
mediate crack induced interfacial debonding. Of these
two failure modes, plate end debonding is by far the
more commonly reported failure mode. Although less
commonly reported, failures by intermediate crack
induced debonding are likely to control the strength of
a signicant portion of FRP-strengthened beams. This
paper and the companion paper [26] are concerned with
strength models for FRP-strengthened RC beams which
fail by plate end debonding.
Fig. 2. Failure modes of FRP-strengthened RC beams. (a) FRP rup-
ture; (b) Crushing of compressive concrete; (c) Shear failure; (d) Con-
crete cover separation; (e) Plate end interfacial debonding; (f) Inter-
mediate crack induced interfacial debonding.
2. Plate end debonding failures
Of the two plate end debonding failure modes, failure
by separation of the concrete cover has been far more
commonly reported (e.g. [1,5,12,20,23]). It is generally
believed that failure of the concrete cover is initiated by
the formation of a crack at or near the plate end, due to
high interfacial shear and normal stresses caused by the
abrupt termination of the plate here (e.g. [14,20,22,27]).
Detailed discussions of these interfacial stresses are
given in Smith and Teng [28] and Teng et al. [25]. Once
a crack forms in the concrete at or near the plate end,
the crack propagates to the level of the tension reinforce-
ment and then progresses horizontally, along the level
of the reinforcement, resulting in the separation of the
concrete cover [16] (Fig. 3a). Fig. 3b shows a close-
up view of the detached plate end, where the tension
reinforcement of the beam can be clearly seen. This
mode of failure has been referred to elsewhere as end-
of-plate failure through concrete [1], concrete rip-off
failure [5,23], debond at rebar layer [15], concrete
cover delamination [20], and local shear failure [27].
As the failure occurs away from the bondline, this is not
Fig. 3. FRP-plated RC beam: concrete cover separation. (a) Overall
view; (b) Close-up view.
387 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
a debonding failure mode in strict terms, although it does
stem from stress concentration near the plate end. The
term concrete cover separation adopted here has pre-
viously been used by Garden and Hollaway [16] and
Juvandes et al. [29] and appears to describe this mode
of failure most accurately (Fig. 2d).
Debonding between the FRP plate and the RC beam
(Fig. 2e), that propagates away from the plate end, has
also been observed in experiments (e.g. [2,8,30]) and is
herein referred to as plate end interfacial debonding. The
general consensus among researchers is that debonding
failures of this form are initiated by high interfacial shear
and normal stresses near the plate end that exceed the
strength of the weakest element, generally the concrete
(e.g. [14,30]). Upon debonding, a thin layer of concrete
generally remains attached to the plate [30]. This sug-
gests that failure generally occurs in the concrete adjac-
ent to the concrete-to-adhesive interface.
As both plate end debonding failure modes are due to
the same cause (i.e. high interfacial shear and normal
stresses near the plate end), plate end interfacial debond-
ing has not been differentiated from concrete cover sep-
aration in many of the existing plate end debonding
strength models.
It should be noted that plate end debonding failures
were rst observed in RC beams bonded with a steel
plate on which considerable research had already
accumulated before FRP plates became popular for plate
bonding (e.g. [3136]) and this effort continued through
the 1990s (e.g. [3741]). Much of what was learnt in
research on steel plate bonding is relevant to FRP
plate bonding.
3. Overview of existing debonding strength models
An extensive search of the existing literature has
found twelve debonding strength models. Seven of them
have been developed for FRP-plated RC beams in the
past few years [14,27,4245]. Another ve models have
been developed for plate end debonding in steel plated
beams [39,4648]. Plate end debonding occurs either by
concrete cover separation or in the concrete adjacent to
the adhesive-to-concrete interface. Although the differ-
ent mechanical properties of the bonded plate are
expected to play a signicant role, strength models can
be expected to be applicable to different types of plates
if the geometrical and material properties of plates are
properly included. It is thus not unreasonable to expect
that models developed for steel plated beams may well
be applicable to FRP-plated beams, with modications
if necessary. Therefore, all the above-mentioned models
are reviewed in this paper.
The existing debonding strength models can be classi-
ed into three categories based on their approaches,
namely (a) shear capacity based models [39,43,47], (b)
concrete tooth models [42,45,48], and (c) interfacial
stress based models (models I and II of Ziraba et al. [46],
[14,27,44]). Model II of Ziraba et al. [46] in fact com-
bines approaches (a) and (c). In addition, it should be
mentioned that Oehlers model [39] is not purely based
on the shear capacity, as it also takes into account the
interaction between shear and bending. It is worth noting
that all the models developed specically for FRP-
strengthened RC beams [14,27,44] follow the interfacial
stress based approach. On the other hand, three of the
ve models specically developed for steel plated beams
follow either the shear capacity approach [39,47] or a
combined approach (model II of Ziraba et al. [46]).
Each of the existing models is summarized and
reviewed in the rest of the paper. All models are
presented here using a consistent set of notation and in
a form that best facilitates subsequent comparisons in
the companion paper [26]. It should be noted that some
of the models are dimensionally inconsistent, and that
all models are presented in this paper based on the fol-
lowing set of consistent units: forces in Newtons (N),
lengths in millimetres (mm) and hence stresses in Mega-
pascals (MPa).
4. Shear capacity based models
The common feature of these models is that the
debonding failure strength is related to the shear strength
of the concrete with no or only partial contribution of
the steel shear reinforcement. The debonding strength is
generally given as the shear force acting at the plate end,
with or without taking into account the effect of any co-
existent moment. As interfacial stresses between the
plate and the beam need not be evaluated, the required
calculations are generally simple.
4.1. Oehlers model
Oehlers and Moran [38] and Oehlers [39] investigated
simply supported steel plated beams subjected to either
four or three point bending and developed a strength
model by rst considering two extreme positions of soft
plate termination. For a plate terminated in the constant
moment region, the following expression for the exural
debonding moment M
db,f
at the end of the plate was pro-
posed and calibrated with test results of steel plated
beams:
M
db,f

E
c
I
tr,c
f
ct
0.901E
frp
t
frp
(1)
where E
c
and E
frp
are the moduli of elasticity of the con-
crete and the FRP respectively, I
trc,c
the cracked second
moment of area of the plated section transformed to con-
crete, f
ct
the cylinder splitting tensile strength of con-
388 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
crete, and t
frp
the plate thickness. If the splitting tensile
strength is not determined from tests, f
ct
(in MPa) is
taken as 0.5f

c
where f

c
is the concrete cylinder com-
pressive strength in MPa [39]. It was suggested that in
design, this moment needs only to be compared to the
additional moment applied to the beam at the plate end
after the plate is bonded.
For a plate terminated near the support, it was pro-
posed that debonding occurs when the shear force at the
plate end V
db,s
reaches the shear capacity of the concrete
in the RC beam alone, without the contribution from the
steel shear reinforcement, based on experimental obser-
vations [39]. That is,
V
db,s
V
c
[1.4(d/2000)]b
c
d[r
s
f

c
]
1/3
(2)
where r
s
=A
s
/b
c
d is the ratio of steel tension reinforce-
ment, A
s
the area of steel tension reinforcement, b
c
the
section width and d the effective depth of the section.
The above expression for V
c
, the concrete shear capacity
in the beam, is that given by the Australian concrete
code [49] and it is a requirement that 1.4(d/2000)1.1.
For general cases where both the shear force and the
moment at the plate end are important, the following
interaction equation was suggested based on test
results [39]:
M
db,end
M
db,f

V
db,end
V
db,s
1.17 (3)
and
M
db,end
M
db,f
,V
db,end
V
db,s
(4)
In terms of the shear force at the plate end, Eq. (3) can
be re-arranged to read
V
db,end

1.17
a
M
db,f
+
1
V
db,s

(5)
and
V
db,end
aM
db,f
,V
db,end
V
db,s
(6)
where a is the distance from the support to the nearer
end of the soft plate.
4.2. Janszes model
Jansze [47] proposed a plate end debonding strength
model for steel plated beams but the background infor-
mation to the model [47] is not available to the authors.
Only the nal form of the model with general comments
given in Ahmed and van Gemert [43] is available where
it is stated that Janszes model [47] was based on the
initiation of shear cracking in an RC beam without the
contribution of shear reinforcement. The critical shear
force in the RC beam at the plate end to cause debonding
V
db,end
is given as follows
V
db,end
t
PES
b
c
d (7)
where
t
PES
0.18
3

3
d
B
mod

200
d

3
100r
s
f

c
(8)
B
mod

(1r
s
)
2
r
s
da
3
(9)
Here, B is the shear span and B
mod
a modied shear
span as given by Eq. (9). If B
mod
as dened by Eq. (9)
is greater than the actual shear span B of the beam then
the modied shear span should be given by (B
mod
+B)/2.
Janszes model [47] appears to be invalid for soft plates
terminated at the support as B
mod
becomes zero and Eq.
(8) predicts that debonding is never possible.
4.2.1. Ahmed and van Gemerts model
Ahmed and van Gemert [43] modied Janszes model
[47] to be suitable for use with FRP-plated RC beams.
Ahmed and van Gemerts model [43] is given as follows
V
db,end
(t
PES
t
mod
)b
c
d (10)
where
t
MOD
t
PES
b
c
d

S
s
I
s
b
frp

S
frp
I
frp
b
a

6188.5

t4.121
b
c
d

(11)
t

0.15776f

17.2366r
s
d
B

0.9
A
sv
f
yv
sb
c
(12)
where t
PES
is given by Eq. (8), S
frp
and S
s
are the rst
moment of area of the FRP plate, and that of an equival-
ent steel plate about the neutral axis of a cracked plated
section transformed to concrete, where the equivalent
steel plate is one that has the same total tensile capacity
and width as that of the FRP plate, but with an equival-
ent thickness determined assuming that the yield stress
of steel is 550 MPa, I
frp
and I
s
are the second moments
of area of a cracked plated section transformed to con-
crete with an FRP plate and an equivalent steel plate
respectively, and b
frp
and b
a
are the widths of the FRP
and adhesive respectively. The stirrup spacing is denoted
by s, while A
sv
and f
yv
are the cross sectional area and
the yield stress of the steel stirrups respectively. For all
practical purposes b
a
is equal to b
frp
.
389 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
The modications given in Eqs. (11) and (12) appear
to rely on determining the difference in the interfacial
shear stress between the RC beam and the plate as a
result of replacing a steel plate with an FRP plate. This
difference t
mod
is then added to the shear stress t
PES
in
Eq. (8) where the shear stress is determined over a c-
titious shear span of length B
mod
. The increase in shear
strength offered by the shear reinforcement is also
included as seen in Eq. (12).
5. Concrete tooth models
Concrete tooth models make use of the concept of a
concrete tooth between two adjacent cracks deforming
like a cantilever under the action of horizontal shear
stresses at the base of the beam (Fig. 4). Debonding is
deemed to occur when these shear stresses lead to tensile
stresses at the root of the tooth that exceed the tensile
strength of the concrete. The stress in the soft plate at
debonding can then be determined by dening an effec-
tive length for the plate for end anchorage over which
a uniform shear stress is assumed.
5.1. Raoof and Zhangs model
The concept of a concrete tooth was rst described in
Zhang et al. [50]. Based on this concept, Zhang et al.
[50] and Raoof and Zhang [48] developed a strength
model to predict concrete cover separation failures in
steel plated beams. Minimum and maximum crack
widths (l
min
and l
max
respectively) are determined which
lead to a lower and an upper bound stress in the plate
to cause debonding respectively and the width of these
cracks differ by a factor of two (l
max
=2l
min
). The mini-
mum width of a crack l
min
, termed the minimum stabil-
ized crack spacing [50], is given by
l
min

A
e
f
ct
u(O
bars
+b
frp
)
(13)
where A
e
is the area of concrete in tension, u the steel-
to-concrete average bond strength, O
bars
the total per-
Fig. 4. Concrete tooth model [50].
imeter of the tension reinforcing bars. It is assumed that
u=0.28f
cu
(in MPa) and f
ct
=0.36f
cu
(in MPa) where f
cu
is the concrete cube compressive strength. For the case
of an RC beam with a single layer of steel tension
reinforcement, A
e
is twice the distance from the centroid
of the tension reinforcement to the base of the RC beam
multiplied by the width of the RC beam.
Ignoring interaction between adjacent teeth and
assuming elastic behaviour, the required shear stress to
cause failure of a tooth can be determined. Failure of
the tooth occurs when the stress at point A (Fig. 4b)
exceeds the tensile strength of the concrete. The tensile
stress at point A, s
A
, can be determined from
s
A

M
A
I
A

l
2

(14)
where M
A
=tlb
frp
h and I
A
=b
frp
l
3
/12. Here, l is the crack
spacing (minimum or maximum), h the net height of
concrete cover measured from the base of the steel ten-
sion reinforcement to the base of the concrete beam, t
the shear stress at the interface between the concrete and
the soft plate, I
A
the second moment of area of the
tooth, and M
A
the moment at the base of the tooth.
Substitution of M
A
and I
A
into Eq. (14) and assuming
that at the instant of plate debonding s
A
=f
ct
, the shear
stress at the interface between the concrete and the steel
plate based on a minimum stabilized crack spacing can
be determined as follows. In this tooth theory, all teeth
in the end anchorage zone are assumed to fail simul-
taneously at debonding.
t
min

f
ct
l
min
6h
b
c
b
frp
(15)
The minimum stress in the soft plate s
s,min
required to
cause exural cracking and failure of a tooth covering
the minimum stabilized crack spacing can then be
determined as follows [50]:
s
s(min)
0.154
L
p
h
1
b
2
c
f
cu
hb
frp
t
frp
(O
bars
+b
frp
)
(16)
where L
p
is an effective length of soft plate for end
390 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
anchorage, and h
1
is the distance from the centroid of
the tension reinforcement to the base of the RC beam.
This model has been applied only to simply-supported
beams subject to three or four point bending, for which
the limiting stress of Eq. (16) was compared to the stress
in the soft plate directly under a point load to check
the possibility of debonding. It is not clear how this
model should be applied to beams subject to other load-
ing conditions (e.g. uniformly distributed load), but it
is not unreasonable to choose the section of maximum
moment in the beam for an assessment.
In Zhang et al. [50], the effective length for end
anchorage was taken as the length of the soft plate in
the shear span (in Fig. 4a this is indicated as L
p1
). In
Raoof and Zhang [48], it was recommended that the
effective length for end anchorage be the smaller of the
length of the soft plate in the shear span L
p1
and the
following lengths L
p2
which were calibrated against test
data of steel plated RC beams that failed by plate end
debonding:
L
p2
l
min
(210.25l
min
), l
min
72 mm (17a)
L
p2
3l
min
, l
min
72 mm (17b)
In the present and the companion paper [26], only the
modied effective length for plate end anchorage as pro-
posed by Raoof and Zhang [48] is considered as this is
the more recent tooth model derived by this research
group for steel plated RC beams.
Once the stress in the plate is known, the moment to
cause separation of the plate, at the same location as
the stress, can be calculated using a conventional section
analysis with the assumption of plane section bending
in accordance with Zhang et al. [50] with the tensile
strength of concrete taken into account. A lower and an
upper bound bending moment are determined as a conse-
quence of the assumed minimum (l
min
) and maximum
(l
max
) stabilized crack spacings differing by a factor of
two.
For a simply supported beam under four point bending
with the plate positioned in the constant moment region,
Raoof and Zhang [48] specied that the effective length
should be obtained from Eq. (17) as the length of the
plate in the shear span L
p1
is zero.
5.2. Wang and Lings model
Wang and Ling [42] proposed a modication to Zhang
et al.s tooth model [50] to make it suitable for FRP-
strengthened beams. This involved modifying the aver-
age bond strength between the concrete and the plate,
leading to the following formula for the minimum
crack spacing:
l
min

A
e
f
ct
u
s
O
bars
+u
frp
b
frp
(18)
where u
s
=0.313f

c
is the average bond strength between
the steel tension reinforcement and concrete and is the
same as that used by Zhang et al. [50] assuming
f

c
=0.8f
cu
, while u
frp
is the average bond shear strength
between the FRP and the concrete and was taken as
1.96 MPa.
Once the minimum stabilised crack spacing is determ-
ined the remainder of the analysis is essentially the same
as that given in Zhang et al. [50] although the effective
length for end anchorage was taken by Wang and Ling
[42] to be the total plate length in the shear span. In
Wang and Lings model [42], the tensile strength of con-
crete was not included in the section analysis for relating
the bending moment in the beam to the stress in the
plate. Similar to the model of Zhang et al. [50], an upper
bound solution of Wang and Lings model is obtainable
by replacing their minimum crack spacing (Eq. (18))
with a maximum crack spacing which is twice the for-
mer.
5.3. Raoof and Hassanens models
Raoof and Hassanen [45] also modied Zhang et al.s
model [50] for application to FRP-plated beams. Two
expressions were presented by them for the effective
length of the FRP plate for end anchorage with two cor-
responding values for the bond strength between the FRP
and the concrete. Two new models thus resulted from
these modications. They are referred to herein as Raoof
and Hassanens model I and model II [45].
In Raoof and Hassanens model I [45], Eq. (13) is
retained to calculate the minimum stabilised crack spac-
ing. The effective length of the plate for end anchorage
is taken to be the smaller of the plate length in the shear
span (L
p1
in Fig. 4a) and the following lengths which
were calibrated with test data of FRP-plated RC beams
reported to have failed by plate end debonding [45]:
L
p2
l
min
(240.5l
min
), l
min
40 mm (19a)
L
p2
4l
min
, l
min
40 mm (19b)
In Raoof and Hassanens model II [45], the bond
strength between the FRP plate and the concrete is speci-
ed as 0.8 MPa, while the bond strength between the
steel tension reinforcement and the concrete is still the
same as that originally specied in Zhang et al. [50].
The effective length of the FRP plate for end anchorage
is dened as the smaller of the plate length in the shear
span and the following lengths which were calibrated
with test data of FRP-plated RC beams reported to have
failed by plate end debonding [45] using the new value
of FRP-to-concrete bond strength:
L
p2
l
min
(11.60.17l
min
), l
min
56.5 mm (20a)
L
p2
2l
min
, l
min
56.5 mm (20b)
Once the stabilised crack spacing and the effective
391 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
length of FRP plate for end anchorage are determined,
the remainder of the analysis is the same as that given
in Zhang et al. [50].
6. Interfacial stress based models
A popular and logical assumption is that concrete
cover separation or plate end interfacial debonding is due
to high interfacial stresses at the end of the soft plate.
Fig. 5 shows the stresses acting on an element of con-
crete adjacent to the end of the adhesive layer where t
and s
y
denote the shear and transverse normal (peeling)
stresses respectively while s
x
is the longitudinal stress.
Interfacial stress based debonding strength models
generally make use of interfacial stresses from an exist-
ing closed-form solution [e.g. 51] and a concrete failure
criterion. Models of this type include model I of Ziraba
et al. [46], and those of Varastehpour and Hamelin [14],
Saadatmanesh and Malek [27] and Tumialan et al. [44].
An alternative use of the interfacial stresses was made
in developing model II of Ziraba et al. [46]. In this
model, the ACI concrete code [52] prediction of the
shear capacity of an RC beam is modied by adjusting
the contribution of the shear stirrups to the shear capacity
of the plated RC beam. The stirrup efciency factor is
related to the peak interfacial stresses at the plate end
and was calibrated with experimental data. This model
therefore combines the shear capacity approach dis-
cussed above with interfacial stresses.
6.1. Ziraba et al.s models
Ziraba et al. [46] proposed two debonding strength
models for steel plated RC beams, one for predicting
plate end interfacial debonding and the other for pre-
dicting concrete cover separation which are termed Zir-
aba et al.s model I and II [46] respectively in this paper.
6.1.1. Model I: plate end interfacial debonding
In this model, the MohrCoulomb failure criterion is
used to dene the critical stress state at plate end interfa-
cial debonding:
ts
y
tanfC (21)
Fig. 5. Stresses acting on a concrete element adjacent to the plate end.
where t and s
y
are the peak interfacial shear and normal
stresses at the plate end, C the coefcient of cohesion,
and f the angle of internal friction. The peak interfacial
shear and normal stresses are given by
ta
1
f
ct
C
R1
V
0
f

5/4
(22)
s
y
a
2
C
R2
t (23)
where
C
R1

K
s
E
frp
b
frp
t
frp

1/2
M
0
V
0

b
frp
t
frp
I
trc,frp
b
a
(d
frp
x
trc,frp
) (24)
C
R2
t
frp
K
n
4E
frp
I
frp

1/4
(25)
where C
R1
and C
R2
are obtained from Roberts analytical
solution [51] for interfacial shear and normal stresses,
while a
1
and a
2
are empirical multipliers calibrated from
numerical studies [46] for RC beams retrotted with
steel plates. The shear stiffness K
s
and the normal stiff-
ness K
n
of the adhesive layer are given by
K
s

G
a
b
a
t
a
(26)
K
n

E
a
b
a
t
a
(27)
with E
a
, G
a
, b
a
and t
a
being the modulus of elasticity,
shear modulus, width and thickness of the adhesive layer
respectively. I
trc,frp
is the second moment of area of the
cracked plated section transformed into FRP, x
trc,frp
is
the neutral axis depth of this transformed cracked section
(distance from the compression face to the neutral axis),
I
frp
the second moment of area of the FRP plate alone,
d
frp
the distance from the compression face of the RC
beam to the centroid of the FRP plate, and M
0
and V
0
the bending moment and shear force respectively at the
plate end.
Substitution of Eqs. (22) and (23) into Eq. (21) gives
an expression for the shear force in the beam, at the plate
end, to cause plate end interfacial debonding:
V
db,end

c
C
R1

C
a
1
f
ct
(1+a
2
C
R2
tanf)

4/5
(28)
392 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
This relationship is subject to the constraint of a/h3
[46] where h is the depth of the RC beam. The following
values for a
1
, a
2
and f were specied in Ziraba et al.
[46]: a
1
=35, a
2
=1.1 and f=28. In Ziraba et al. [46] two
values of C were used for specic numerical examples,
namely 2.68 MPa and 5.36 MPa although the criteria
used for the selection of these values were not made
clear. Ziraba et al. [53] suggested that C lie between 4.80
MPa and 9.50 MPa based on experimental and numerical
ndings. For the present study [26] C has been taken as
7.15 MPa which is an average of the values given in
Ziraba et al. [53].
6.1.2. Model II: concrete cover separation
Ziraba et al. [46] modied the ACI code [52] predic-
tion of the shear capacity of an RC beam into the follow-
ing equation where k is the efciency factor of steel
shear reinforcement:
V
db,end
(V
c
kV
s
) (29)
where V
c
and V
s
are the contributions of concrete and
steel shear reinforcement to the shear capacity of an RC
beam respectively and are given in their paper by
V
c
1/6(f

c
100r
s
)b
c
d (30)
V
s
(A
sv
f
yv
d)/s (31)
In Eq. (29), k is an empirically derived stirrup efciency
factor to adjust the contribution of the steel stirrups and
is related to the peak interfacial normal stress at the plate
end. Based on a regression analysis of the test results of
fourteen steel plated RC beams which failed by concrete
cover separation, Ziraba et al. [46] proposed that
k2.4e
n
and n0.08C
R1
C
R2
10
6
(32)
6.2. Varastehpour and Hamelins model
Varastehpour and Hamelin [14] also developed a plate
end interfacial debonding strength model based on the
MohrCoulomb failure criterion (Eq. (21)). In their
model, the coefcient of cohesion C was calibrated from
single lap shear tests and the angle of internal friction f
determined from small scale FRP-plated beam tests that
failed by debonding. Details of the particular mode of
debonding exhibited by these test beams were not speci-
ed by them. An average value of 5.4 MPa was sug-
gested for C, while a value of 33 was suggested for f.
The shear stress required in the MohrCoulomb equ-
ation is given by
t
1
2
b(lV
0
)
3/2
(33)
This equation was modied from the following
expression for the shear stress between a steel soft plate
and an RC beam originally proposed by Jones et al. [32]:
tlV
0
(34)
where l is the rigidity of the section dened by
l
t
frp
E
frp
I
trc,c
E
c
(d
frp
x
trc,c
) (35)
where x
trc,c
is the neutral axis depth of a cracked section
transformed to concrete. In Eq. (33), the factor b was
introduced to account for the effect of various variables
that have a signicant inuence on the interfacial shear
stress between the RC beam and the soft plate, such as
the rigidity and thickness of the plate, geometry of the
section, and nature of loading based on a parametric
study using a numerical model which considered slip
between the plate and the RC beam. The following
expression was developed by them for b:
b
1.2610
5
B
h
0.7
t
frp
E
frp
(36)
The normal stress s
y
is related to the shear stress as fol-
lows as derived by Roberts [51]
s
y
C
R2
t (37)
The shear force in the beam, at the plate end to cause
debonding, can then be determined by
V
db,end

1.6t
2/3
max
lb
1/3
(38)
where
t
max

5.4
1+C
R2
tan33
(39)
6.3. Saadatmanesh and Maleks model
Saadatmanesh and Malek [27] developed a debonding
strength model to predict concrete cover separation in
FRP-plated beams based on the assumption that concrete
cover failure is related to high stresses at the plate end
(Fig. 5). Of the three stresses present at the end of the
soft plate, the shear stress t and the normal stress s
y
are from a closed-form solution derived by Malek et al.
[54] while the longitudinal stress is from a section bend-
ing analysis, all based on an uncracked section. This
closed-form solution was derived assuming the follow-
ing quadratic distribution for the bending moment which
can accommodate both point and uniformly-distributed
loads:
Ma
1
(xa)
2
a
2
(xa)a
3
(40)
where x is the distance along the soft plate from its left
end. For a simply supported beam subjected to three or
393 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
four point bending, the bending moment at the plate end
M
0
, for a soft plate terminated in the shear span, is
given by M
0
=a
2
a+a
3
.
Under this bending moment distribution (Eq. (40)),
the interfacial shear stress t at the plate end is given
as [54]
tt
frp
(b
3
Ab
2
) (41)
where
A
G
a
t
a
t
frp
E
frp
(42)
b
2

E
frp
I
tru,c
E
c
(d
frp
x
tru,c
)(2a
1
aa
2
) (43)
b
3
E
frp
1
I
tru,c
E
c
(d
frp
x
tru,c
)(a
1
a
2
a
2
aa
3
) (44)
2
a
1
E
frp
I
tru,c
E
c
(d
frp
x
tru,c
)
t
a
t
frp
G
a

where I
tru,c
is the uncracked second moment of area of
the plated section transformed to concrete, and x
tru,c
is
the neutral axis depth of this transformed section. The
normal stress s
y
at the plate end is [54]
s
y

K
n
2b
a
b
3

V
frp
E
frp
I
frp

0
+b

M
0
E
c
I
c

qE
frp
I
frp
b
frp
E
c
I
c
(45)
where
V

0
V
0
0.5hb
frp
t
frp
(b
3
Ab
2
) (46)
V
frp
0.5b
frp
t
2
frp
(b
3
Ab
2
) (47)
and
b

K
n
b
frp
4b
a
E
frp
I
frp

1/4
(48)
where I
c
is the second moment of area of the beam and
q a uniformly distributed load if such a load exists.
The longitudinal stress s
x
at the base of the RC beam,
at the end of the soft plate, due to a bending moment
M
0
can be determined from a bending analysis of an
uncracked section. The bending moment in the concrete
beam at the plate end is increased by an amount M
m
,
dened as follows and attributed to the peak interfacial
shear stress:
M
m
0.5hab
frp
t (49)
Once all three stress components at the plate end are
determined, the maximum principal stress can be found
as follows
s
1

s
x
+s
y
2

s
x
s
y
2

2
+t
2
(50)
At the plate end, the concrete is generally subjected to
biaxial tension. Concrete cover separation is deemed to
occur when the maximum principal stress (in MPa)
reaches the concrete splitting tensile strength, that is
when
s
1
f
ct
0.295(f

c
)
2/3
(51)
In generating numerical results using this model for
comparison in the companion paper [26], the present
authors adopted a trial and error procedure. The pro-
cedure involves iterating on a load until Eq. (51) is satis-
ed.
6.4. Tumialan et al.s model
Tumialan et al. [44] developed a concrete cover separ-
ation debonding strength model for FRP-plated beams.
Their model is in principle similar to the Saadatmanesh
and Malek model [27] reviewed above, in that the
stresses acting on a concrete element at the plate end
(Fig. 5) are determined and checked against a failure
criterion for the concrete. The differences lie in how
these stresses are evaluated. In Tumialan et al.s model
[44], the peak interfacial shear and normal stresses at the
plate end are determined from the solution of Roberts
[51] as follows:
tC

R1
E
frp
E
c
V
0
(52)
s
y
C
R2
t (53)
where C
R2
is dened by Eq. (25) while C

R1
is given by
C

R1

K
s
E
frp
b
frp
t
frp

1/2
M
0
V
0

b
frp
t
frp
I
tru,c
b
a
(d
frp
x
tru,c
) (54)
The factor C

R1
here differs from C
R1
in Reference [51]
in that the neutral axis and corresponding second
moment of area are here based on an uncracked section
transformed to concrete as opposed to a cracked section
transformed to the plate material. As a result, the modu-
lar ratio of E
frp
/E
c
is included in Eq. (52). The longitudi-
nal stress at the base of the RC beam s
x
, at the plate
end, is also determined from a bending analysis of an
uncracked section. However, the additional component
induced by the interfacial shear stress as considered by
Saadatmanesh and Malek [27] in Eq. (49) is not
included here.
Once the stresses at the plate end are determined, the
maximum principal stress s
1
can be easily determined
using Eq. (50). Failure by concrete cover separation is
deemed to occur when this maximum principal stress s
1
reaches the modulus of rupture of the concrete f
r
which
is taken as 0.689f

c
(MPa) (8.3f

c
psi, [55]).
In generating numerical results using this model for
comparison in the companion paper [26], the present
394 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
authors adopted the same trial and error procedure as
was used with Saadatmanesh and Maleks model [27].
7. Conclusions
Plate end debonding by either the separation of con-
crete cover or by interfacial debonding in the concrete
adjacent to the adhesive-to-concrete interface has been
the most commonly reported debonding failure mode in
tests on RC beams strengthened with a bonded FRP sof-
t plate. As a result, twelve strength models have been
developed to predict plate end debonding failure loads,
of which seven were developed for FRP-plated RC
beams and the other ve were developed for steel plated
RC beams. All these twelve models have been reviewed,
and classied into the following three categories based
on their approaches: (a) shear capacity based models; (b)
concrete tooth models; and (c) interfacial stress based
models. Each model has been concisely presented using
a consistent set of notation for ease of future reference
and cross comparison. In addition, the theoretical basis
of each model was explained and any relationship
between it and other models established. In the com-
panion paper [26], the performance of each of these
twelve models will be assessed using a large experi-
mental database of RC beams that failed by plate end
debonding assembled from an extensive literature sur-
vey.
Acknowledgements
Both authors wish to thank The Hong Kong Polytech-
nic University for the provision of a postdoctoral fellow-
ship to the rst author.
References
[1] Ritchie PA, Thomas DA, Lu LW, Connely GM. External
reinforcement of concrete beams using ber reinforced plastics.
ACI Structural Journal 1991;88(4):490500.
[2] Saadatmanesh H, Ehsani MR. RC Beams strengthened with
GFRP plates. I: Experimental study. Journal of Structural Engin-
eering 1991;117(11):341733.
[3] Triantallou TC, Plevris N. Strengthening of RC beams with
epoxy-bonded bre-composite materials. Materials and Structures
1992;25:20111.
[4] Chajes MJ, Thomson TA, Januszka TF, Finch WW Jr. Flexural
strengthening of concrete beams using externally bonded com-
posite materials. Construction and Building Materials
1994;8(3):191201.
[5] Sharif A, Al-Sulaimani GJ, Basunbul IA, Baluch MH, Ghaleb
BN. Strengthening of initially loaded reinforced concrete beams
using FRP plates. ACI Structural Journal 1994;91(2):1608.
[6] Meier U. Strengthening of structures using carbon bre/epoxy
composites. Construction and Building Materials
1995;9(6):34151.
[7] Heffernan PJ, Erki MA. Equivalent capacity and efciency of
reinforced concrete beams strengthened with carbon bre
reinforced plastic sheets. Canadian Journal of Civil Engineering
1996;23:219.
[8] Quantrill RJ, Hollaway LC, Thorne AM. Experimental and ana-
lytical investigation of FRP strengthened beam response: Part I.
Magazine of Concrete Research 1996;48(177):33142.
[9] Shahawy MA, Arockiasamy M, Beitelman T, Sowrirajan R.
Reinforced concrete rectangular beams strengthened with CFRP
laminates. Composites: Part B 1996;27B:22533.
[10] Takeda K, Mitsui Y, Murakami K, Sakai H, Nakamura M. Flex-
ural behaviour of reinforced concrete beams strengthened with
carbon bre sheets. Composites Part A 1996;27A:9817.
[11] Arduini M, Nanni A. Behavior of pecracked RC beams strength-
ened with carbon FRP sheets. Journal of Composites for Con-
struction, ASCE 1997;1(2):6370.
[12] Garden HN, Hollaway LC, Thorne AM. A preliminary evaluation
of carbon bre reinforced polymer plates for strengthening
reinforced concrete members. Proceedings of the Institution of
Civil Engineers: Structures and Buildings 1997;123(May):127
42.
[13] Garden HN, Quantrill RJ, Hollaway LC, Thorne AM, Parke
GAR. An experimental study of the anchorage length of carbon
bre composite plates used to strengthen reinforced concrete
beams. Construction and Building Materials 1998;12:20319.
[14] Varastehpour H, Hamelin P. Strengthening of concrete beams
using ber-reinforced plastics. Materials and Structures
1997;30:1606.
[15] Buyukozturk O, Hearing B. Failure behavior of precracked con-
crete beams retrotted with FRP. Journal of Composites for Con-
struction, ASCE 1998;2(3):13844.
[16] Garden HN, Hollaway LC. An experimental study of the inu-
ence of plate end anchorage of carbon bre composite plates used
to strengthen reinforced concrete beams. Composite Structures
1998;42:17588.
[17] Ross CA, Jerome DM, Tedesco JW, Hughes ML. Strengthening
of reinforced concrete beams with externally bonded composite
laminates. ACI Structural Journal 1999;96(2):21220.
[18] Grace NF, Sayed GA, Soliman AK, Saleh KR. Strengthening
reinforced concrete beams using ber reinforced polymer (FRP)
laminates. ACI Structural Journal 1999;96(5):86574.
[19] Swamy RN, Mukhopadhyaya P. Debonding of carbon-bre-
reinforced polymer plate from concrete beams. Proceedings of
the Institution of Civil Engineers: Structures and Buildings
1999;134(November):30117.
[20] Tumialan G, Serra P, Nanni A, Belarbi A. Concrete cover delami-
nation in reinforced concrete beams strengthened with carbon
ber reinforced polymer sheets. In: Dolan CW, Rizkalla SH,
Nanni A, editors. Proceedings of the Fourth International Sym-
posium on Fiber Reinforced Polymer Reinforcement for
Reinforced Concrete Structures. Maryland (USA), 1999:725-735.
[21] Ramana VPV, Kant T, Morton SE, Dutta PK, Mukherjee A,
Desai YM. Behavior of CFRPC strengthened reinforced concrete
beams with varying degrees of strengthening. Composites: Part
B 2000;31:46170.
[22] Mukhopadhyaya P, Swamy N. Interface shear stress: a new
design criterion for plate debonding. Journal of Composites for
Construction, ASCE 2001;5(1):3543.
[23] Nguyen DM, Chan TK, Cheong HK. Brittle failure and bond
development length of CFRP-concrete beams. Journal of Com-
posites for Construction, ASCE 2001;5(1):127.
[24] Rahimi H, Hutchinson A. Concrete beams strengthened with
externally bonded FRP plates. Journal of Composites for Con-
struction, ASCE 2001;5(1):4456.
[25] Teng JG, Chen JF, Smith ST, Lam L. RC structures strengthened
with FRP composites. Research Centre for Advanced Technology
in Structural Engineering, Department of Civil and Structural
395 S.T. Smith, J.G. Teng / Engineering Structures 24 (2002) 385395
Engineering, The Hong Kong Polytechnic University, Hong
Kong, China, December 2000.
[26] Smith ST, Teng JG. FRP-strengthened RC beams. II: assessment
of debonding strength models. Engineering Structures,
2002;24(4):397417.
[27] Saadatmanesh H, Malek AM. Design guidelines for exural
strengthening of RC beams with FRP plates. Journal of Com-
posites for Construction, ASCE 1998;2(4):15864.
[28] Smith ST, Teng JG. Interfacial stresses in plated RC beams.
Engineering Structures 2001;23:85771.
[29] Juvandes L, Figueiras JA, Marques AT. Performance of concrete
beams strengthened with CFRP laminates. In: Saadatmanesh H,
Ehsani MR, editors. Proceedings of the Second International Con-
ference on Composites in Infrastructure (ICCI). Arizona, USA,
1998.
[30] Taljsten B. Concrete beams strengthened for bending using
CFRP-sheets. In: Forde MC, editor. Proceedings of the Eighth
International Conference on Advanced Composites for Concrete
Repair. London, UK, 1999.
[31] Jones R, Swamy RN, Bloxham J, Bouderbalah A. Composite
behaviour of concrete beams with epoxy bonded external
reinforcement. The International Journal of Cement Composites
2 1980;2:91107.
[32] Jones R, Swamy RN, Charif A. Plate separation and anchorage of
reinforced concrete beams strengthened with epoxy-bonded steel
plates. The Structural Engineer 1988;66(5):8594.
[33] Macdonald MD, Calder AJJ. Bonded steel plating for strengthen-
ing concrete structures. International Journal of Adhesion and
Adhesives 1982;2(2):11927.
[34] Swamy RN, Jones R, Ang TH. Under- and over-reinforced con-
crete beams with glued steel plates. The International Journal of
Cement Composites and Lightweight Concrete 4 1982;4(1):19
32.
[35] Swamy RN, Jones R, Bloxham JW. Structural behaviour of
reinforced concrete beams strengthened by epoxy-bonded steel
plates. The Structural Engineer 1987;65A(2):5968.
[36] Swamy RN, Jones R, Charif A. The effect of external plate
reinforcement on the strengthening of structurally damaged RC
beams. The Structural Engineer 1989;67(3):4556.
[37] Basunbul IA, Gubati AA, Al-Sulaimani GJ, Baluch MH.
Repaired reinforced concrete beams. ACI Materials Journal
1990;87(4):34854.
[38] Oehlers DJ, Moran JP. Premature failure of externally plated
reinforced concrete beams. Journal of Structural Engineering,
ASCE 1990;116(4):97895.
[39] Oehlers DJ. Reinforced concrete beams with plates glued to their
softs. Journal of Structural Engineering, ASCE
1992;118(8):202338.
[40] Hussain M, Sharif A, Basunbul IA, Baluch MH, Al-Sulaimani
GJ. Flexural behavior of precracked reinforced concrete beams
strengthened externally by steel plates. ACI Structural Journal
1995;92(1):1422.
[41] Ashrafuddin M, Baluch MH, Sharif A, Al-Sulaimani GJ, Azad
AK, Khan AR. Peeling and diagonal tension failures in steel
plated R/C beams. Construction and Building Materials
1999;13:45967.
[42] Wang CY, Ling FS. Prediction model for the debonding failure
of cracked RC beams with externally bonded FRP sheets. In:
Proceedings of the Second International Conference of Com-
posites in Infrastructure (ICCI), Arizona, USA, 1998:548562.
[43] Ahmed O, van Gemert D. Effect of longitudinal carbon ber
reinforced plastic laminates on shear capacity of reinforced con-
crete beams. In: Dolan CW, Rizkalla SH, Nanni A, editors. Pro-
ceedings of the Fourth International Symposium on Fiber
Reinforced Polymer Reinforcement for Reinforced Concrete
Structures. Maryland, USA, 1999:933943.
[44] Tumialan G, Belarbi A, Nanni A. Reinforced concrete beams
strengthened with CFRP composites: failure due to concrete
cover delamination. Department of Civil Engineering, Center for
Infrastructure Engineering Studies, Report No. CIES-99/01, Uni-
versity of Missouri-Rolla, USA: 1999.
[45] Raoof M, Hassanen MAH. Peeling failure of reinforced concrete
beams with bre-reinforced plastic or steel plates glued to their
softs. Proceedings of the Institution of Civil Engineers: Struc-
tures and Buildings 2000;140(August):291305.
[46] Ziraba YN, Baluch MH, Basunbul IA, Sharif AM, Azad AK, Al-
Sulaimani GJ. Guidelines towards the design of reinforced con-
crete beams with external plates. ACI Structural Journal
1994;91(6):63946.
[47] Jansze W. Strengthening of RC Members in Bending by Exter-
nally Bonded Steel Plates. PhD Thesis, Delft University of Tech-
nology, Delft, 1997.
[48] Raoof M, Zhang S. An insight into the structural behaviour of
reinforced concrete beams with externally bonded plates. Pro-
ceedings of the Institution of Civil Engineers: Structures and
Buildings 1997;122(November):47792.
[49] AS 3600. Concrete Structures, AS 3600, Standards Australia,
Sydney, Australia, 1988.
[50] Zhang S, Raoof M, Wood LA. Prediction of peeling failure of
reinforced concrete beams with externally bonded steel plates.
Proceedings of the Institution of Civil Engineers: Structures and
Buildings 1995;110(August):25768.
[51] Roberts TM. Approximate analysis of shear and normal stress
concentrations in the adhesive layer of plated RC beams. The
Structural Engineer 1989;67(12):22933.
[52] ACI 318-95. Building Code Requirements for Structural Concrete
(318-95) and Commentary (318R-95). American Concrete Insti-
tute (ACI), Fifth Printing, Michigan, USA, 1999.
[53] Ziraba YN, Baluch MH, Basunbul IA, Azad AK, Al-Sulaimani
GJ, Sharif AM. Combined experimental-numerical approach to
characterization of steel-concrete interface. Materials and Struc-
tures 1995;28:51825.
[54] Malek AM, Saadatmanesh H, Ehsani MR. Prediction of failure
load of R/C beams strengthened with FRP plate due to stress
concentration at the plate end. ACI Structural Journal
1998;95(1):14252.
[55] Mirza S, Hatzinikolas M, MacGregor J. Statistical descriptions
of the strength of concrete. Journal of Structural Engineering,
ASCE 1979;105(ST6):102137.