Engineering Structures: Yefei Wu, Zhiqiang Zhou, Qingda Yang, Weiqiu Chen

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Engineering Structures 32 (2010) 897–905

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

On shear bond strength of FRP-concrete structures


Yefei Wu a , Zhiqiang Zhou b , Qingda Yang b , Weiqiu Chen c,∗
a
Department of Civil Engineering, Zijingang Campus, Zhejiang University, Hangzhou 310058, PR China
b
Department of Mechanical and Aerospace Engineering, University of Miami, FL 33124, USA
c
Department of Engineering Mechanics, Yuquan Campus, Zhejiang University, Hangzhou 310027, PR China

article info abstract


Article history: The shear bond strength of FRP-to-concrete bonded joints is generally obtained through single shear tests,
Received 2 July 2009 for which the specimen can be viewed as a layered structure. Based on the fracture mechanics theory for
Received in revised form laminated structures, two simple formulae to estimate shear bond strength are suggested respectively
7 December 2009
for two main failure modes encountered in the testing. The formulae exclude all empirical parameters
Accepted 7 December 2009
Available online 29 December 2009
except the one accounting for the width effect, therefore bearing a more rational and physically strict
sense. Finite element analysis is performed to determine the fracture toughness which is the most
Keywords:
important parameter in the formulae. The cohesive zone model (CZM) is employed in the simulation. A
Fiber reinforced polymer comparison with experimental observations and results predicted by existing formulations is made and
Bond strength good agreement is achieved.
Single shear test © 2009 Elsevier Ltd. All rights reserved.
Fracture mechanics
Cohesive zone model

1. Introduction dominating failure mode, estimate the bond strength, and under-
stand the debonding mechanism.
Fiber reinforced polymer (also referred as fiber reinforced plas- In order to find a simple and effective analytic expression for
tic, both abbreviated as FRP) plates have been increasingly em- bond strength, empirical formulations based on regression of ex-
ployed to strengthen deficient reinforced concrete (RC) structures, perimental data and some simplifying assumptions were used in
because FRP has superior properties such as a high strength-to- the beginning. The method of strength of materials was then em-
weight ratio and a high corrosion resistance. The former property ployed to develop simple theoretical models [7–9]. However, it
leads to a great ease in site handling, reducing labor cost and inter- is difficult to use the method to tackle the stress singularity at
ruption of existing service, while the latter ensures durable perfor- the end of FRP plates where the interfacial stresses attain an in-
mance [1]. finite magnitude. Thus, linear elastic fracture mechanics (LEFM)
The FRP plate bonding technology was first investigated by was adopted to overcome the singularity problem. Rabinovitch
Swiss Federal Laboratory for Material Testing and Research (EMPA) and Frostig [10] obtained the energy release rate at the debond-
where tests on RC beams strengthened with CFRP plates started ing tip by using a high order beam theory. Suo and Hutchinson
in 1984 [2]. In engineering practice, the bond strength between presented theoretical formulations for a crack in a brittle substrate
FRP and concrete has been found to be very crucial to the behav- parallel to the film/substrate interface [11] and a semi-infinite in-
ior of FRP strengthened RC structures. FRP usually cannot attain its terface crack between two infinite isotropic elastic layers [12],
ultimate strength when the structure breaks down due to the respectively. Those models can be used to analyze FRP-concrete
debonding failure occurring at or near the FRP-concrete inter- structures, as will be shown in this paper. Wang and Qiao [13]
face. Therefore plenty of experimental and theoretical studies have extended Suo and Hutchinson’s analysis [12] by accounting for the
been carried out on the bond strength of FRP-concrete structures. transverse shear deformation effect and mode mixity of the inter-
Several effective testing methods have been proposed, including face crack. Au and Büyüköztürk [14] developed a tri-layer fracture
the single shear test and the double shear test. Among these, the model to facilitate the quantification of various debonding modes
single shear test has become the standard experiment for mea- in FRP/epoxy/concrete plated systems. It can be seen from the liter-
suring the shear bond strength because of its simplicity and reli- ature that the energy release rate is the main aim of the calculation.
ability [3–6]. By the single shear test, researchers can observe the In recent studies, nonlinear fracture mechanics (NLFM) became
popular, and more and more researchers adopted nonlinear stress-
deformation laws to describe the behavior of the FRP-concrete
∗ Corresponding author. Tel.: +86 571 87951866; fax: +86 571 87952570. interface. Holzenkämper [15] investigated the bond strength be-
E-mail address: chenwq@zju.edu.cn (W. Chen). tween a steel plate and concrete using NLFM. Täljsten [16] derived
0141-0296/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2009.12.017
898 Y. Wu et al. / Engineering Structures 32 (2010) 897–905

3. Existing models for bond strength prediction

For the sake of later comparison, we review in this section


several bond strength models available in the literature. In 1994,
Holzenkämper [15] used a two-layered model and deduced the fol-
lowing expression for the maximum bond capacity
p
P = bfrp 2Gc Efrp tfrp (1)
where Gc is the interface fracture energy (or fracture toughness),
Efrp tfrp is the FRP stiffness, and bfrp is the width. Based on a series of
tests, Neubauer and Rostásy [24] found that the fracture toughness
could be calculated by

Fig. 1. Single shear test: (a) Elevation; (b) Plan form. Gc = cf ft (2)
where ft is the tensile strength of concrete, and cf was determined
the formula for mode II fracture for symmetric and asymmetric to be 0.204 mm by a regression analysis of the test data. They fur-
overlapping joints using a simplified nonlinear shear-deformation ther considered the effect of geometry of the specimen, and mod-
relation and the energy criterion. Wu et al. [17] introduced two ified Holzenkämper’s formula (1) as
kinds of nonlinear interfacial constitutive laws to describe the pre-
0.64kp bfrp Efrp tfrp ft 
 p
and post-interfacial debonding behavior, and solved the nonlinear  for L ≥ Le
interfacial stress transfer and fracture propagation problems for P = L L (3)
0.64kp bfrp Efrp tfrp ft for L < Le
p
different kinds of adhesive joints in FRP/steel-strengthened con- 2−
Le Le
crete or steel structures. Chen et al. [18] analyzed the debonding
where Le = Efrp tfrp /(2ft ) is the effective bond length, and the ge-
p
process of an FRP-concrete bonded joint where the FRP plate is
subjected to tension at both ends; a bilinear local bond-slip model ometric factor kp is related to the width of the FRP plate bfrp and
was employed to describe the behavior of the FRP-concrete inter- that of concrete bc as
face. Some models in fracture mechanics have also been used in the s
study of FRP-concrete structures. The cohesive zone model (CZM), 2 − bfrp /bc
kp = 1.125 .
which was first suggested by Dugdale [19], plays a vital role in the 1 + bfrp /400
analysis of crack propagating and is a powerful and useful tool to
deal with large-scale fracture processing zones [20–23]. Monti et al. [25] conducted a parametric study using FEA
Two new formulations for calculating the shear bond strength in conjunction with experiments, and suggested the following
will be presented in this paper by making use of the theoretical re- formula for the bond capacity
sults in Refs. [11,12]. The new formulations are different from the
Efrp tfrp τmax
r
existing ones because different failure modes are taken into con- P = bfrp β (4)
sideration, and their advantages will be discussed. In particular, 3
comparison with the experimental results in Refs. [3–6] and the where β = 1 for L ≥ Le and β = sin(π L/2Le ) for L < Le , with the
available bond strength models [17,24–27] validates the new for- effective bond length given by
mulations. A finite element analysis incorporating CZM elements s
also will be performed to estimate some important parameters Efrp tfrp
such as the fracture toughness. Le = √
4τmax

2. Single shear test and failure modes and the maximum bond shear stress depends on the tensile
strength of concrete as
The sketch of single shear test is shown in Fig. 1, where the FRP s
plate is bonded to the concrete prism and is pulled in one direction 2 − bfrp /bc
τmax = 1.8kb ft , with kb = 1.5 . (5)
until the structure is broken. The geometric parameters of the test 1 + bfrp /100
include the FRP bond length Lfrp , the width ratio of the bonded FRP
plate and concrete bfrp /bc and the FRP thickness tfrp . Täljsten [16] considered a three-layered model with a crack in
Experiments conducted by Yao [5] show four typical failure the adhesive layer and calculated the work done when the size of
modes of FRP-concrete debonding: (a) cover separation; (b) plate crack changes from a to a + da. In this way, he obtained
end interfacial debonding; (c) intermediate (flexural or flexural- r
2Efrp tfrp Gc Efrp tfrp
shear) crack (IC) induced interfacial debonding; (d) critical diago- P = bfrp , with α = (6)
nal crack (CDC) induced interfacial debonding. The first two failure 1+α Ec tc
modes often happen in the single shear test. When the FRP plate where Ec tc is the stiffness of the concrete prism, and Gc is defined
and concrete prism have a similar width and FRP plate is relatively as the fracture energy of the adhesive.
short, the concrete prism failure may happen. Wu et al. [17] used nonlinear interfacial constitutive laws to
Except for the concrete prism failure, which is not considered analyze a three-layered model and deduced a similar formula
in this paper, the failure modes of single shear test can be divided s
into two kinds according to where the debonding happens, either 2Efrp tfrp Gc bfrp Efrp tfrp
at the interface or inside of the concrete. Yao [5] tested 72 FRP- P = bfrp , with β = . (7)
1+β bc Ec tc
concrete specimens by the single shear test. Fifty-six out of the
72 specimens failed due to debonding in concrete adjacent to the It may be seen that when Ec tc  Efrp tfrp or bc Ec tc  bfrp Efrp tfrp ,
adhesive-concrete interface; eight specimens failed in debonding Eqs. (6) and (7) bear the same form as Eq. (1). However, either
at the interface. These two main failure modes will be considered Täljsten [16] or Wu et al. [17] didn’t show how to calculate the
in this paper. interface fracture energy Gc .
Y. Wu et al. / Engineering Structures 32 (2010) 897–905 899

Lu et al. [26] used new bond-slip models which are based on


their earlier work of meso-scale finite element predictions [28],
and suggested another similar model

P = βw βl bfrp
p
2Gc Efrp tfrp (8)
where the two geometric parameters βw and βl are given by
1.0 (Lfrp ≥ Le )
s 
2.25 − bfrp /bc
βw = , βl = ,

L L
1.25 + bfrp /bc 2− (Lfrp < Le )
Le Le

Their definition of the effective bond length Le follows Yuan


et al. [29]
λ1 + tan(λ2 a) Fig. 2. A two-layer structure with a semi-infinite crack at the interface.
 
1
Le = a + ln
2λ1 λ1 + tan(λ2 a) 4. Fracture mechanics models for different failure modes
with
τmax τmax
r r
The new models are simply established based on the work
λ1 = , λ2 = , of Suo and Hutchsion [11,12], where a sandwich structure with
s0 Efrp tfrp (sf − s0 )Efrp tfrp
s ! a semi-infinite crack is considered within the regime of linear
1 sf − s0 fracture mechanics. The first model is for the interface bonding and
a= arcsin 0.99 , τmax = 1.50βw ft ,
λ2 sf the second for the concrete cover separation, as to be shown below.

2Gc
s0 = 0.0195βw ft , sf = . 4.1. Interface debonding
τmax
The interfacial fracture energy in their model is defined as According to Suo and Hutchinson [12], the energy release rate
(energy release per unit of new crack area) of a semi-infinite crack
Gc = 0.308βw ft .
2
p
(9) at the interface between two elastic layers as shown in Fig. 2 is:
After summarizing nearly all the previous bond strength models !
and analyzing a large amount of test data in the literature, Chen and 1 p2 m2 pm
G= + + 2√ sin γ (12)
Teng [27] proposed a simple expression for calculating the ultimate 2Ēfrp Atfrp 3
Itfrp 2
AItfrp
bond strength as follows
where m and p are the applied moment and force per unit width
P = 0.427βw βl Le bfrp
p
fc0 (10) as shown in Fig. 2, Ēfrp is the equivalent elastic modulus of FRP
where fc0 (in MPa) is the cylinder concrete compressive strength, (Ēfrp = Efrp for plane stress problem and Ēfrp = Efrp /(1 − νfrp
2
) for
the two geometric factors βw and βl and the effective bond length plane strain problem, here Efrp and νfrp are the elastic modulus and
Le (in mm) are given by Poisson’s ratio of FRP, respectively), A = 1/[1 + Σ (4η+ 6η2 + 3η3 )]
with η = tfrp /tc being the thickness ratio and Σ = Ēfrp /Ēc being
s s
2 − bfrp /bc Efrp tfrp
βw = , , the Young’s modulus ratio, I = 1/[12(1 + η )] and sin γ =
P 3
Le =
1 + bfrp /bc √
p
fc0
η (1 + η) AI.
P 2
6
1.0  (Lfrp ≥ Le ) We have η → 0 because the thickness of the FRP plate is much

βl = π Lfrp smaller than that of the concrete substrate. Then accounting for

(11)
sin (Lfrp < Le ) the limited value of Σ , we derive A ≈ 1. Furthermore, since we are
2 Le
considering single shear (or pure pull-out) test, we have m = 0.
where the elastic modulus of FRP plate Efrp is given in MPa, and the Thus, we obtain from Eq. (12)
thickness of FRP plate tfrp is in mm.
All the models based on fracture mechanics require the p2
knowledge of interface or adhesive fracture toughness. Efforts
G= . (13)
2Ēfrp tfrp
have been made to relate the interface fracture toughness to the
tensile strength of concrete [24,26]; but the proposed relations are When the energy release rate increases to Gci , the fracture
empirical and the forms by different authors are quite diverse, see toughness of the interface, the crack becomes unstable and
Eqs. (2) and (9) for example. The two semi-empirical models in propagates forward. At this critical value of the energy release rate,
Eqs. (4) and (10) don’t involve the interface fracture toughness. the shear bond strength p per unit width and hence the total load
They are somehow similar to the semi-analytical models which are capacity P can be calculated as:
based on fracture mechanics but with empirical relations between q q
interface fracture toughness and tensile/compressive strength of p= 2Ēfrp tfrp Gci , P = bfrp 2Ēfrp tfrp Gci . (14)
concrete. The semi-empirical model (10) has been shown to be
very accurate and is very convenient to be used for the design As in the literature, we may introduce a parameter to account
purpose [27]. However, either the semi-empirical models or the for the width effect of the FRP plate. However, since the interface
semi-analytical models are inconsistent in dimension, and lack a energy release rate in Eq. (13) is assumed to be independent of the
strict physical sense in its nature. Furthermore, since the FRP plate material and geometric properties of the concrete substrate, it is
is bonded onto the concrete by adhesive, the property of which reasonable to use Eq. (14) directly. For the single shear test, the
may play an important role in the bond capacity. This factor has plane stress state holds so that we have Ēfrp = Efrp in Eq. (14).
not been considered in the semi-empirical models, nor in the semi- Comparing our formulation in Eq. (14)2 with other fracture
analytical models. Moreover, it is necessary to develop different mechanics modelspas reviewed in Section 3, we find that they all
fracture mechanics models to predict the different failure modes contain the same Efrp tfrp Gc . However, the definition of Gc is dif-
observed in the single shear test. ferent. Here we define it as the fracture toughness of the adhesive
900 Y. Wu et al. / Engineering Structures 32 (2010) 897–905

τc

Shear stress
GII

δt1 δt δt2 δtc


Shear displacement
Fig. 3. A three-layer structure with a semi-infinite crack at one interface.
Fig. 4. Traction-separation curve of the mode-II CZM. (GII is the mode-II
(interface), which is appropriate for the failure mode of interface traction separation energy, at shear displacement δt ; δtc is the critical tangential
displacement for mode-II fracture; GIIc is the mode-II fracture toughness, equal to
debonding. Furthermore, we will put forward a numerical method
the total area under the shear traction-separation law.)
to determine the value of Gci by using the cohesive element as to be
shown in the next section. Note that the fracture toughness usually
cannot be obtained directly from experiment, and an appropriate The bond strength per unit width and the total load capacity
simulation tool should be expected and also helpful. can be calculated from the condition that the energy release rate
It is pointed out that the model shown in Fig. 2 is semi-infinite, equals the fracture toughness of the concrete, Gcc :
with an infinite FRP bond length Lfrp . Hence, Eq. (14) is valid when s 
λ

Lfrp is larger than the effective bond length, which is the common
p = βw 2 1+ Ēfrp tfrp Gcc ,
situation in the engineering practice. Σ
Moreover, Eq. (14) is obtained based on the assumption that the s 
crack exists before loading. However, practically in the beginning λ

of loading process in the single shear test, there should be no crack. P = βw bfrp 2 1+ Ēfrp tfrp Gcc (17)
Σ
Thus, it can be used to calculate the bond strength, rather than to
predict the whole load–displacement curve. where, in order to take account of the width effect of FRP plate,
the factor βw in Eq. (11) [27] is employed, and λ = λtfrp /tfrp =
4.2. Concrete cover separation td /tfrp is also called the relative crack depth of the concrete cover.
According to Yao [6], the thickness of debonded concrete varies
approximately between 1 mm–5 mm.
If the bonding is carefully prepared, the toughness of the adhe-
sive is usually very high and debonding will occur in the concrete
near the interface. The corresponding fracture model is shown in 5. Numerical simulation with cohesive zone element
Fig. 3, where a crack parallel to the interface is located in the con-
crete. Yang and Thouless [23] presented a mode-dependent embe-
The energy release rate of the interface crack in a three-layer dded-process-zone model (also referred to CZM) and used it to
structure as shown in the Fig. 3 has also been given by Suo and simulate the mixed-mode fracture of deforming adhesive joints.
Hutchinson [11] This paper uses their CZM to simulate the single shear test, for
" # which the use of reliable mode-II parameters is proved to be
1 p2 m2 pm essential in obtaining accurate predictions of fracture, because the
G= + 3
+ 2√ sin γ (15) mode-I component of fracture is almost neglectable. The traction-
Utfrp Vtfrp 2
2Ēc UV tfrp
separation law of mode II, as shown in Fig. 4, can be characterized
by the fracture toughness GIIc , the peak stress τc , and two shape
where Ēc is the equivalent elastic modulus of concrete for plane
parameters δt1 /δtc and δt2 /δtc . Numerical experiments show that
problem, U, V and sin γ are dimensionless parameters which de-
GIIc and τc dominate the analysis, while the two shape parameters
pend on λ, λ0 and Σ , with λ being the thickness ratio of debonded
are less important. Thus, in this paper we simply take δt1 /δtc =
concrete cover and FRP plate, λ0 being the thickness ratio of con-
0.1 and δt2 /δtc = 0.2. By varying the values of GIIc and τc , it
crete block and FRP plate (see Fig. 3) and Σ being the Young’s mod-
is possible to obtain numerical results that are reasonably close
ulus ratio of FRP and concrete. Applying this theoretical model to
to the experiments, and in this way the two parameters can be
the single shear test (m = 0), the two items associated with m in
quantified. Parametric study shows that τc influences the stiffness
the right-hand side of Eq. (15) vanish. Furthermore, since the FRP
of the structure, while GIIc affects the failure load of the structure.
plate thickness tfrp is about 10−1 –1 mm while that of the concrete
Yao [5] used the near-end supported single shear pull exper-
block tc is about 102 mm, λ0 equals approximately 102 –103 . Thus, iment to test 72 FRP-concrete specimens, and presented load–

λ2 +2Σ λ+Σ
2 displacement curves of series I, II and VI specimens, which are
1 1 1 λ+Σ
+ λ0 − λ 1 selected here for performing numerical simulations and quantify-
= + +3 ≈ ing the cohesive parameters to be used in Eqs. (14) and (17). In each
U λ+Σ λ0 − λ (λ0 − λ)3 λ+Σ
series, all parameters are identical except the bond length of FRP.
by neglecting the last two terms. In this case, we may obtain a The specimens with a bond length larger than the effective bond
simple expression for U as U = λ + Σ . We then obtain from length can be regarded identical because the bond strength doesn’t
Eq. (15) that increase anymore. Thus, only three analytical models are needed
to simulate all the three series of experiments. The failure mode of
1 p2 series VI is interface debonding (DB-I), and that of series I and II is
G= . (16)
2Ēc Utfrp concrete cover debonding (DB-C). The determination of the value
Y. Wu et al. / Engineering Structures 32 (2010) 897–905 901

analytical solution is for the interface debonding model, but they


used the DB-C experimental results of specimen II-5 in Ref. [30] to
quantify the interface parameters. Very interestingly, their values
of τc = 7.2 MPa and Gc = 0.58 N/mm are very close to our results
as presented above.
Similarly, to obtain the fracture energy of the concrete Gcc in
Eq. (17), another FE model (the diagram is not shown) is adopted. In
this model, which is used to simulate the concrete cover debonding
failure, the cohesive elements are located in the concrete prism,
paralleling to the FRP plate and with a distance of 3.5 mm from the
FRP plate, which is equal to the mean thickness of the debonded
Fig. 5. FE model (interface debonding). concrete cover observed in experiments. When τc = 2.4 MPa,
GIIc = 0.17 N/mm, the predictions of series I and II specimens
agree well with the experimental results, as shown in Fig. 6(b, c).
of Gci will be based on the comparison between simulation and Thus, the fracture energy of concrete Gcc is taken to be 0.17 N/mm,
experiment of DB-I specimens, and the value of Gcc will be deter- which is reasonably close to that obtained by Elices et al. [21]
mined from the results of DB-C specimens. Thus, we use different (Gc = 126 ± 14 N/m).
simulation models to obtain Gci and Gcc , which will then be used in
Eqs. (14) and (17) for the two different failure modes, respectively. 6. Comparisons
Fig. 5 shows the plane elasticity model for interface debonding
used in the finite element analysis. The cohesive elements (user The two formulations in Eqs. (14) and (17) are derived more
compiled in Fortran, supported by the UEL function in the finite strictly based on linear fracture mechanics, and here their vali-
element package ABAQUS) are set in the adhesive layer between dation will be checked by comparison with the experimental re-
FRP and concrete. In order to simplify the model, the FRP plate sults [3–6]. The details of these available tests are given in Table A.1
and concrete are assumed to be elastic, and four-node plane stress in Appendix. It is noted that Eqs. (14) and (17) are for two different
elements are used. The geometrical and material parameters of the failure modes. Hence, the experimental specimens should be clas-
specimen can be found in Ref. [30]. sified accordingly: failure mode of series VI specimens in Ref. [5]
Through numerical study, we find that the simulated load– and I-3–III-4 in Ref. [6] (18 specimens) is debonding at interface,
displacement curves have the following two common features: and failure mode of the other 104 specimens are debonding in the
(1) load increases linearly with displacement in the beginning, and concrete. Now we shall use Gci = 0.52 N/mm for the fracture
τc controls the initial slope of the curves; and (2) after the load toughness of the interface and Gcc = 0.17 N/mm for the frac-
reaches the ultimate value, curves go horizontally, and Gc influ- ture toughness of the concrete, as obtained in Section 5. It should
ences the ultimate load. Thus, in our analysis, τc and Gc are deter- be noted that the fracture toughness, whose magnitude clearly de-
mined when the slope and ultimate load of the simulated curves pends on the material properties of FRP, concrete or adhesive, may
are identical to those of the experimental curves. be different for different specimens in Refs. [3–6]. However, since
When τc = 7.0 MPa and GIIc = 0.52 N/mm, the predicted load- there is the lack of load–displacement curves as in Ref. [5], we use
deformation curve for series VI specimens is in good agreement only identical values as given above for all specimens, which may
with the experimental results, as shown in Fig. 6(a). However, after cause certain deviation. For comparison, the formulae given in Eqs.
the initiation of FRP debonding, the experimental curves may still (3), (4), (7), (8) and (10) are simultaneously used to calculate the
ascend. Such a kind of increase in load may be due to the aggregate critical loads of all specimens listed in Table A.1 in Appendix. The
interlocking and friction in the debonded zone. In our numerical values of interfacial fracture toughness in Eq. (7) are also taken to
simulation, these affects have been neglected since they are out of be 0.52 N/mm.
the scope of this work. The calculated bond strengths are shown against the exper-
Yuan et al. [29] presented an analytical solution of the prob- imental results in Fig. 7. The statistical results are listed in the
lem of FRP-concrete bonded joints. They also showed how the Table 1. From Fig. 7(a) and (b), it is seen that, Neubauer and
experimental load–displacement (the displacement is the total dis- Rostásy’s model and Monti et al.’s model overestimate the bond
placement of FRP plate relative to the concrete) response of the strength. Wu et al.’s model is very similar to Eq. (14), and good pre-
joints can be used to quantify the interfacial fracture energy. Their dictions are achieved for DB-I specimens. However, its prediction

6 6 7

5 5 6

5
Load (kN)

4
Load (kN)

4
Load (kN)

4
3 3
Experiment of VI2 3
Experiment of VI3 Experiment of II2
Experiment of I2
2 Experiment of VI4 2 Experiment of II3
Experiment of VI5
Experiment of I3 2 Experiment of II5
Experiment of I6
Experiment of VI6 FEA (DB-C) of II
1 Experiment of VI7 1 Experiment of I9
1
FEA (DB-C) of I
FEA(DB-I) of VI
0 0 0
0 0.5 1 1.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Displacement (mm) Displacement (mm) Displacement (mm)
(a) Interface debonding. (b) Concrete cover debonding of I. (c) Concrete cover debonding of II.

Fig. 6. Numerical predictions and experiments of load–displacement curves.


902 Y. Wu et al. / Engineering Structures 32 (2010) 897–905

Table 1
Statistic results of the calculating errors of each failure modes.
Models Eq. (3) Eq. (4) Eq. (7) Eq. (8) Eq. (10) Eq. (14) Eq. (17)

Average Ptest /Ppred 0.8237 0.7828 1.1257 0.8513 1.0301 1.0632 1.0075
SD 0.0964 0.0993 0.1460 0.1221 0.1202 0.1196 0.1416
COV 0.1170 0.1269 0.1297 0.1434 0.1167 0.1125 0.1405

(a) Neubauer and Rostásy’s model, Eq. (3). (b) Monti et al.’s model, Eq. (4). (c) Wu et al.’s model, Eq. (7).

(d) Lu et al.’s model, Eq. (8). (e) Chen and Teng’s model, Eq. (10). (f) Present models, Eqs. (14) and (17).

Fig. 7. Experimental versus predicted bond strength using various models.

becomes worse for DB-C specimens, because it is established for FRP-concrete structures, based on the linear fracture mechanics of
an interface crack model, not for concrete cover separation. The layered structures. These formulations are similar to some those
same problem exists in the model suggested by Lu et al. [26], who obtained by others available in literature, but the definition of frac-
used Yuan et al.’s interface crack model [29] and calculated the ture energy is different, which is obtained in this paper through
interfacial fracture toughness by Eq. (9). In √this case, the interfa- numerical simulation using CZM. A comparison with experimen-
cial fracture toughness is proportional to ft , which may lead to tal results validates our formulae. In contrast to Chen and Teng’s
an overestimated interfacial fracture toughness when the concrete formula, the presented formulae have a clearer physical back-
strength is high. In fact, for the DB-I, the interfacial fracture tough- ground and a more mathematically strict sense. In particular, there
ness may mainly depend on the material property of adhesive. This is completely no problem with the dimensions of parameters in the
difficulty has been avoided by our models which distinguish the formula. Furthermore, when compared with the existing formu-
two failure modes, and the fracture toughness is separately deter- lae based on fracture mechanics, the present formulae distinguish
mined according to experimental load–displacement curves. From clearly the two different failure modes observed in experiments.
Fig. 7(e) and (f), it can be seen that our models have almost the
same accuracy as Chen and Teng’s semi-empirical model, which is Acknowledgements
however improper in dimension, and perform better than all the
other models. The predicted bond strengths are very close to the This work was supported by the National Natural Science
experimental results. This can also be seen from Table 1. Foundation of China (Nos. 10725210) and the National Basic
Research Program of China (No. 2009CB623200).
7. Conclusions
Appendix
This paper presents two formulations of bond strength for
two different failure modes observed in the single shear test of See Table A.1.
Y. Wu et al. / Engineering Structures 32 (2010) 897–905 903

Table A.1
Specimen details.
Reference Specimen FRP plate Concrete prisma
Thickness Width Bond length Elastic Tensile strength Width Cylinder Test load
tfrp (mm) bfrp (mm) Lfrp (mm) modulus ffrp (MPa) bc (mm) compressive Ptest (kN)
Efrp (GPa) strength fc0 (MPa)

Chajes et al. [3] C1 1.016 25.4 76.2 108.478 1655 228.6 36.1 8.462
C2 1.016 25.4 76.2 108.478 1655 228.6 47.1 9.931
C3 1.016 25.4 76.2 108.478 1655 228.6 47.1 10.638
C4 1.016 25.4 76.2 108.478 1655 228.6 47.1 10.638
C5 1.016 25.4 76.2 108.478 1655 228.6 43.6 10.531
C7 1.016 25.4 76.2 108.478 1655 228.6 43.6 9.610
C8 1.016 25.4 76.2 108.478 1655 228.6 43.6 10.518
C9 1.016 25.4 76.2 108.478 1655 228.6 43.6 11.199
C10 1.016 25.4 76.2 108.478 1655 228.6 24.0 9.869
C11 1.016 25.4 76.2 108.478 1655 228.6 28.9 9.343
C12 1.016 25.4 76.2 108.478 1655 228.6 43.7 11.204
C13 1.016 25.4 76.2 108.478 1655 228.6 36.4 8.094
C14 1.016 25.4 76.2 108.478 1655 228.6 36.4 12.811
C15 1.016 25.4 76.2 108.478 1655 228.6 36.4 11.917
C16 1.016 25.4 76.2 108.478 1655 228.6 36.4 11.570
Takeo et al. [4] 1-11 0.167 40 100 230 3481 100 28.88 8.75
1-12 0.167 40 100 230 3481 100 26.66 8.85
1-21 0.167 40 200 230 3481 100 28.88 9.30
1-22 0.167 40 200 230 3481 100 26.66 8.50
1-31 0.167 40 300 230 3481 100 28.88 9.30
1-32 0.167 40 300 230 3481 100 26.66 8.30
1-41 0.167 40 500 230 3481 100 28.88 8.05
1-42 0.167 40 500 230 3481 100 28.88 8.05
1-51 0.167 40 500 230 3481 100 26.47 8.45
1-52 0.167 40 500 230 3481 100 26.47 7.30
2-11 0.167 40 100 230 3481 100 24.99 8.75
2-12 0.167 40 100 230 3481 100 24.99 8.85
2-13 0.167 40 100 230 3481 100 26.17 7.75
2-14 0.167 40 100 230 3481 100 26.17 7.65
2-15 0.167 40 100 230 3481 100 24.40 9.00
2-21 0.167 40 100 230 3481 100 24.99 12.00
2-22 0.167 40 100 230 3481 100 24.99 10.80
2-31 0.167 40 100 230 3481 100 24.99 12.65
2-32 0.167 40 100 230 3481 100 24.99 14.35
2-41 0.167 40 100 230 3481 100 24.40 11.55
2-42 0.167 40 100 230 3481 100 24.40 11.00
2-51 0.167 40 100 230 3481 100 26.17 9.85
2-52 0.167 40 100 230 3481 100 26.17 9.50
2-61 0.167 40 100 230 3481 100 26.17 8.80
2-62 0.167 40 100 230 3481 100 26.17 9.25
2-71 0.167 40 100 230 3481 100 26.17 7.65
2-72 0.167 40 100 230 3481 100 26.17 6.80
2-81 0.167 40 100 230 3481 100 49.97 7.75
2-82 0.167 40 100 230 3481 100 49.97 8.05
2-91 0.167 40 100 230 3481 100 24.40 6.75
2-92 0.167 40 100 230 3481 100 24.40 6.80
2-101 0.167 40 100 230 3481 100 24.99 7.70
2-102 0.167 40 100 230 3481 100 26.17 6.95
Yao [5] I-1 0.165 25 75 256 4114 150 23.0 4.75
I-2 0.165 25 85 256 4114 150 23.0 5.69
I-3 0.165 25 95 256 4114 150 23.0 5.76
I-4 0.165 25 95 256 4114 150 23.0 5.76
I-5 0.165 25 95 256 4114 150 23.0 6.17
I-6 0.165 25 115 256 4114 150 23.0 5.96
I-7 0.165 25 145 256 4114 150 23.0 5.95
I-8 0.165 25 190 256 4114 150 23.0 6.68
I-9 0.165 25 190 256 4114 150 23.0 6.35
I-10 0.165 25 95 256 4114 150 23.0 6.17
I-11 0.165 25 75 256 4114 150 23.0 5.72
I-12 0.165 25 85 256 4114 150 23.0 6.00
I-13 0.165 25 95 256 4114 150 23.0 6.14
I-14 0.165 25 115 256 4114 150 23.0 6.10
I-15 0.165 25 145 256 4114 150 23.0 6.27
I-16 0.165 25 190 256 4114 150 23.0 7.03
II-1 0.165 25 95 256 4114 150 22.9 5.20
II-2 0.165 25 95 256 4114 150 22.9 6.75
II-3 0.165 25 95 256 4114 150 22.9 5.51
II-4 0.165 25 190 256 4114 150 22.9 7.02
II-5 0.165 25 190 256 4114 150 22.9 7.07
II-6 0.165 25 190 256 4114 150 22.9 6.98
(continued on next page)
904 Y. Wu et al. / Engineering Structures 32 (2010) 897–905

Table A.1 (continued)

Reference Specimen FRP plate Concrete prisma


Thickness Width Bond length Elastic Tensile strength Width Cylinder Test load
tfrp (mm) bfrp (mm) Lfrp (mm) modulus ffrp (MPa) bc (mm) compressive Ptest (kN)
Efrp (GPa) strength fc0 (MPa)

III-1 0.165 25 100 256 4114 150 27.1 5.94


III-2 0.165 50 100 256 4114 150 27.1 11.66
III-3 0.165 75 100 256 4114 150 27.1 14.63
III-4 0.165 100 100 256 4114 150 27.1 19.07
III-7 1.27 25 100 225 351 150 27.1 4.78
III-8 1.27 51 100 22.5 351 150 27.1 8.02
IV-1 0.165 25 95 256 4114 150 18.9 5.86
IV-2 0.165 25 95 256 4114 150 18.9 5.90
IV-3 0.165 25 95 256 4114 150 19.8 5.43
IV-4 0.165 25 95 256 4114 150 19.8 5.76
IV-5 0.165 25 95 256 4114 150 18.9 5.00
IV-6 0.165 25 95 256 4114 150 19.8 7.08
IV-7 0.165 25 95 256 4114 150 18.9 5.50
IV-8 0.165 25 95 256 4114 150 19.8 5.93
IV-9 0.165 25 95 256 4114 150 18.9 5.38
IV-10 0.165 25 95 256 4114 150 19.8 6.60
IV-11 0.165 25 95 256 4114 150 18.9 5.51
IV-12 0.165 25 95 256 4114 150 19.8 5.67
IV-13 0.165 25 95 256 4114 150 18.9 6.31
IV-14 0.165 25 95 256 4114 150 19.8 6.19
V-1 0.165 15 95 256 4114 150 21.1 3.81
V-2 0.165 15 95 256 4114 150 21.1 4.41
V-3 0.165 25 95 256 4114 150 21.1 6.26
V-4 0.165 50 95 256 4114 150 21.1 12.22
V-5 0.165 75 95 256 4114 150 21.1 14.29
V-6 0.165 100 95 256 4114 150 21.1 15.58
VI-1 0.165 25 95 256 4114 150 21.9 6.01
VI-2 0.165 25 95 256 4114 150 21.9 5.85
VI-3 0.165 25 145 256 4114 150 21.9 5.76
VI-4 0.165 25 145 256 4114 150 21.9 5.73
VI-5 0.165 25 190 256 4114 150 21.9 5.56
VI-6 0.165 25 190 256 4114 150 21.9 5.58
VI-7 0.165 25 240 256 4114 150 21.9 5.91
VI-8 0.165 25 240 256 4114 150 21.9 5.05
VII-1 0.165 25 95 256 4114 150 24.9 6.80
VII-2 0.165 25 95 256 4114 150 24.9 6.62
VII-3 0.165 25 145 256 4114 150 24.9 7.33
VII-4 0.165 25 145 256 4114 150 24.9 6.49
VII-5 0.165 25 190 256 4114 150 24.9 7.07
VII-6 0.165 25 190 256 4114 150 24.9 7.44
VII-7 0.165 25 240 256 4114 150 24.9 7.16
VII-8 0.165 25 240 256 4114 150 24.9 6.24
Toutanji et al. [6] I-3 0.825 50 100 110 660 200 17.0 11.64
I-4 0.990 50 100 110 660 200 17.0 12.86
II-1 0.495 50 100 110 660 200 46.2 12.55
II-2 0.660 50 100 110 660 200 46.2 14.25
II-3 0.825 50 100 110 660 200 46.2 17.72
II-4 0.990 50 100 110 660 200 46.2 18.86
III-1 0.495 50 100 110 660 200 61.5 13.24
III-2 0.660 50 100 110 660 200 61.5 15.17
III-3 0.825 50 100 110 660 200 61.5 18.86
III-4 0.990 50 100 110 660 200 61.5 19.03
Cube compressive strength of concrete is given by fcu = fc0 /0.79, the tensile strength of concrete is calculated by ft = 0.53 fc0 MPa, and the Young’s modulus of the
a
p
p
concrete is calculated by Ec = 4730 f0
c MPa according to ACI 318-89 [31].

References [7] Roberts TM, Haji-Kazemi H. Theoretical study of the behaviour of reinforced
concrete beams strengthened by externally bonded steel plates. Proc Inst Civ
Eng 1989;87(1-Part 2):39–55.
[1] Teng JG, Chen JF, Smith ST, Lam L. FRP-strengthened RC structures. UK: John [8] Malek AM, Saadatmanesh H, Ehsani MR. Prediction of failure load of RC beams
Wiley & Sons; 2002. strengthened with FRP plates due to stress concentration at the plate end. ACI
[2] Meier U, Deuring M, Meier H, Schwegler G. CFRP bonded sheets. In: Fiber- Struct J 1998;95(1):142–52.
Reinforced-Plastic (FRP) reinforcement for concrete structures: Properties and [9] Smith ST, Teng JG. Interfacial stresses in plated beams. Eng Struct 2001;23(7):
applications. Netherlands: Elsevier Science Publishers B.V.; 1993. 857–71.
[3] Chajes MJ, Finch WW, Januszka TF, Thonson TA. Bond and force transfer of [10] Robinovitch O, Frostig Y. Delamination failure of RC beams strengthened with
composite material plates bonded to concrete. ACI Struct J 1996;93(2):209–17. FRP strip—A closed-form high-order and fracture mechanics approach. J Eng
[4] Takeo K, Matstushiba H, Makizumi T, Nagashima G. Bond characteristics of Mech 2001;127(5):852–61.
[11] Suo ZG, Hutchinson JW. Steady-state cracking in brittle substrates beneath
CFRP sheets in the CFRP bonding technique. Proc Japan Concr Inst 1997;19(2):
adherent films. Internat J Solids Structures 1989;25(8):1337–53.
1599–604.
[12] Suo ZG, Hutchinson JW. Interface crack between two elastic layers. Internat J
[5] Yao J. Debonding failure in reinforced concrete structures strengthened with Fract 1990;43(1):1–18.
externally bonded FRP sheets/plates. Hong Kong: Doctorial Dissertation of [13] Wang JL, Qiao P. Interface crack between two shear deformable elastic layers.
Hong Kong Polytechnic University; 2004. J Mech Phys Solids 2004;52(4):891–905.
[6] Toutanji H, Saxena P, Zhao LY, Ooi T. Prediction of interfacial bond failure of [14] Au C, Büyüköztürk O. Debonding of FRP plated concrete: A tri-layer fracture
FRP-concrete surface. J Compos Constr 2007;11(4):427–36. treatment. Eng Fract Mech 2006;73(3):348–65.
Y. Wu et al. / Engineering Structures 32 (2010) 897–905 905

[15] Holzenkämpfer O. Ingenieurmodelle des vverbundes geklebter bewehbrung [24] Neubauer U, Rostásy FS. Design aspects of concrete structures strengthened
für betonbauteile. Disscertation, TU Braunschweig; 1994 [in German]. with externally bonded CFRP plates. In: Proceedings of 7th international
[16] Täljsten B. Strengthening of concrete prisms using the plate bonding conference on structural faults and repairs, vol. 2. 1997. p. 109–18.
technique. Internat J Fract 1996;82(3):253–66. [25] Monti G, Renzelli M, Luciani P. FRP adhesion in uncracked concrete zones.
[17] Wu ZS, Yuan H, Niu HD. Stress transfer and fracture propagation in different In: Tan KH, editor. Proceedings of the sixth international symposium on FRP
kinds of adhesive joints. J Eng Mech 2002;128(5):562–73. reinforcement for concrete structures, Vol. 2. Singapore: World Scientific;
[18] Chen JF, Yuan H, Teng JG. Debonding failure along a softening FRP-to-concrete 2003. p. 183–92.
interface between two adjacent cracks in concrete members. Eng Struct 2007;
[26] Lu XZ, Teng JG, Ye LP, Jiang JJ. Bond-slip models for FRP sheets/plates bonded
29(2):259–70.
to concrete. Eng Struct 2005;27(6):920–37.
[19] Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys Solids 1960;
[27] Chen JF, Teng JG. Anchorage strength for FRP and steel plates bonded to
8(2):100–4.
[20] Blackman BRK, Hadavinia H, Kinloch AJ, Williams JG. The use of cohesive zone concrete. J Struct Eng 2001;127(5):784–91.
model to study the fracture of fiber composites and adhesively-bonded joints. [28] Lu XZ, Ye LP, Teng JG, Jiang JJ. Meso-scale finite element model for FRP
Internat J Fract 2003;119(1):25–46. plates/sheets bonded to concrete. Eng Struct 2005;27(4):564–75.
[21] Elices M, Guineae GV, Gómez J, Planas J. The cohesive zone model: Advantages, [29] Yuan H, Teng JG, Seracino R, Wu ZS, Yao J. Full range of FRP-concrete bonded
limitations and challenges. Eng Fract Mech 2002;69(2):137–63. joints. Eng Struct 2004;26(5):553–65.
[22] Wang JL. Cohesive zone model of FRP-concrete interface debonding under [30] Yao J, Teng JG, Chen JF. Experimental study on FRP-to-concrete bonded joints.
mixed-mode loading. Int J Solid Struct 2007;44(20):6551–68. Composites 2005;36(2):99–113.
[23] Yang QD, Thouless MD. Mixed-mode fracture analysis of plastically- [31] ACI Committee. Building code requirements for reinforced concrete. ACI 318-
deformation adhesive joints. Internat J Fract 2001;110(2):175–87. 89. Detroit: American Concrete Institute; 1989.

You might also like