You are on page 1of 12

Prediction Models for Debonding Failure Loads of Carbon

Fiber Reinforced Polymer Retrofitted Reinforced


Concrete Beams
Huy Binh Pham1 and Riadh Al-Mahaidi, M.ASCE2
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This study focuses on debonding failure in reinforced concrete beams with carbon fiber reinforced polymer composite bonded
on the soffit using the wet lay-up method. An experimental study, which involved 26 tests, was carried out. The experiments showed two
failure modes: Intermediate span debond and end debond. The first failure is the result of the high bond stress near the tip of a
flexure-shear crack, whereas the second type of failure is due to the high shear stress developed in the weakest concrete layer at the tension
reinforcement level. The experiments have shown that U-straps can be effective in preventing intermediate span and end debond. Based
on experimental observations, two simple and practical theoretical models were developed and verified with the experimental data,
together with a large database of other existing tests.
DOI: 10.1061/共ASCE兲1090-0268共2006兲10:1共48兲
CE Database subject headings: Concrete beams; Fiber reinforced polymers; Bonding; Predictions; Retrofitting; Models; Failures.

Introduction concentrations. Although there has been a significant amount of


research on this topic, a simple reliable design method is still not
Laminate bonding is a strengthening solution, where laminates available.
are externally bonded on reinforced concrete members and take In this study, the writers attempt to address this issue. The
tension in the same way as internal steel reinforcements. This study includes three main parts. In the first part, previous experi-
method has proven to be effective in increasing members’ stiff- ments and existing theories are summarized and assessed. Testing
ness and ultimate load capacity. The first type of laminate bond- of RC beams is then carried out to study the failure mechanisms
and the effect of several important parameters. Two methods to
ing was steel plate bonding, which has been used since the 1960s.
predict the failure loads of the two main debonding modes,
However, the steel system has some drawbacks, such as the pos-
namely end debond and intermediate span debond, are proposed
sible corrosion of the plates and the difficulties that arise from
and verified with the data from the present experiments together
their high self-weight.
with a database of previous tests.
Within the last decade, a second type of laminate bonding has
been developed, in which the laminates used are fiber reinforced
polymer 共FRP兲 composites. This system possesses many advan-
tages over the steel system, such as high strength of laminates, Review of Previous Experiments and Existing
easy installation, excellent durability, and minimum maintenance
Theories
requirement. Although it is a relatively new technique, FRP bond-
There have been a number of tests carried out to study the behav-
ing has been used to strengthen numerous bridges and buildings
ior of RC beams retrofitted for flexural strengthening. The failure
worldwide.
modes can be grouped into two main categories: Sectional fail-
One of the main applications of FRP strengthening is to im-
ures and debond failures. Sectional failures include compression
prove the flexural capacity of reinforced concrete 共RC兲 beams by
failure of the concrete or tensile rupture of the laminate. The
bonding FRP composites on the soffit. However, the tension in the
failures are similar to those of normal RC beams. Debond failures
composites often results in peeling or debonding of the strength- are more frequently observed. They can be classified into two
ening materials. This phenomenon is complicated since it is main categories based on the initial starting point. End debond,
affected by various factors, such as beam cracking and stress also known as plate end debond, is the failure that originates near
the plate end and propagates in the concrete along the tension
1
Research Student, Dept. of Civil Engineering, Monash Univ., steel reinforcement. Intermediate span debond is the failure that
Victoria 3800, Australia.
2
originates either from a wide flexural crack 共also known as inter-
Head of Structures Group, Dept. of Civil Engineering, Monash Univ., mediate crack兲 or a flexure-shear crack. The failure then propa-
Victoria 3800, Australia. gates to the composite ends along the adhesive-concrete interface.
Note. Discussion open until July 1, 2006. Separate discussions must
There have also been a number of models developed to predict
be submitted for individual papers. To extend the closing date by one
month, a written request must be filed with the ASCE Managing Editor. the ultimate capacities of retrofitted beams. Pham and Al-Mahaidi
The manuscript for this paper was submitted for review and possible 共2004兲 have verified the beam theory 共also known as the
publication on December 3, 2004; approved on May 26, 2005. This paper Bernoulli compatibility truss model or bending theory兲 against 35
is part of the Journal of Composites for Construction, Vol. 10, No. 1, beams and shown that the theory can predict the sectional capaci-
February 1, 2006. ©ASCE, ISSN 1090-0268/2006/1-48–59/$25.00. ties accurately. Predicting debond failure loads is more difficult

48 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


due to the complexity of the failure mechanisms. There have been Table 1. Experimental Studies on Flexural Strengthening by Fiber Rein-
a number of models proposed. They are briefly summarized in the forced Polymer Retrofitting
next two paragraphs. The summary is by no means an exhaustive No. Reference
review of all existing models. More detailed studies have been
1 Ritchie et al. 共1991兲
reported elsewhere 共Smith and Teng 2002; Pham and Al-Mahaidi
2004兲. 2 Saadatmanesh and Ehsani 共1991兲
Intermediate span debond can be divided into two submodes: 3 Triantafillou and Plevris 共1992兲
Flexural crack-induced debond or flexure-shear crack-induced 4 Chajes et al. 共1994兲
debond. To predict the failure load of the first mode, two recent 5 Sharif et al. 共1994兲
models by Teng et al. 共2001兲 and Maruyama and Ueda 共2001兲 can 6 Quantrill et al. 共1996兲
be used. Teng et al. 共2001兲 calculated the maximum force that the 7 Arduini and Nanni 共1997兲
bonded FRP composite can take based on the capacity of a similar 8 He et al. 共1997兲
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

shear-lap test. Maruyama and Ueda 共2001兲 reported a different 9 Garden et al. 共1997兲
design approach by limiting the FRP tensile stress at the location 10 Mukhopadhyaya et al. 共1998兲
of the crack. To predict the failure load of flexure-shear crack 11 Garden et al. 共1998兲
induced debond, the expression by Blaschko 共1997兲 might be 12 Ahmed and Van Gemert 共1999兲
used, in which the acting shear force was limited to the modified 13 Beber et al. 共1999兲
concrete shear capacity without shear reinforcement. 14 David et al. 共1999兲
End debond has received much more attention and there are
15 Hau 共1999兲
several models developed to predict the failure load of this mode
16 Tumialan et al. 共1999兲
for a retrofitted beam. Three main methodologies adopted are:
17 Ross et al. 共1999兲
Interfacial stress based, concrete tooth based, and beam shear
capacity based. In the first method, the interfacial stress state near 18 Bonacci and Maalej 共2000兲
the composite end is considered as the determining factor. The 19 Gao et al. 共2001兲
most recently verified formulas were developed by El-Mihilmy 20 Fanning and Kelly 共2001兲
and Tedesco 共2001兲. The second method considers the cantilever 21 Zarnic and Bosiljkov 共2001兲
action of a concrete tooth between two adjacent cracks under the 22 Rahimi and Hutchinson 共2001兲
application of horizontal interfacial shear stresses. A practical de- 23 Nguyen et al. 共2001兲
sign method was proposed by Chaallal et al. 共1998兲. In the third 24 Kishi et al. 共2001兲
method, end debond is related to the concrete shear capacity of 25 Pornpongsaroj and Pimanmas 共2003兲
the beam. These models tend to be simpler. Two examples are the 26 Takahashi and Sato 共2003兲
models proposed by Oehlers 共1992兲 and Jansze 共1997兲. 27 Valcuende and Benlloch 共2003兲
To assess the above mentioned models, a database of 181 ex- 28 Leong and Maalej 共2003兲
isting RC beam tests was built. The tests were selected from a 29 Smith and Teng 共2003兲
number of experimental studies as listed in Table 1. The selected 30 Khomwan et al. 共2004兲
beams were simply supported, rectangular, and retrofitted with
31 Pham and Al-Mahaidi 共2003兲
FRP laminates on their soffit. They failed either due to sectional
32 This study 共Pham and Al-Mahaidi兲
or debond failures. The database covers a wide range of beam
configurations. The beam spans vary from 0.87 to 6.0 m. The
beam section aspect ratios 共width over depth兲 range from width, was reported to be 0.176 mm. MBrace Primer was used to
0.45 to 1.67. The concrete strengths reported range from improve the bonding of the composite to the substrate. It is a
25 to 80 MPa. two-part epoxy product with low viscosity and 100% solids con-
The assessment results are shown in Figs. 1 and 2. The statis- tent. It has the ability to penetrate the substrates and to bond to a
tical parameters of the ratios of the experimental measurement to saturated surface dry concrete surface. The resin used was
the predicted value of the ultimate load are summarized in Table MBrace Saturant, which is a two-part epoxy with 100% solids
2. It can be seen that existing models generally produce conser- content used to both impregnate the fibers to form a composite
vative and relatively scattered results. The minimum value of the and bond it to the primed surface. The main properties of these
average of the ratio between the predicted and the actual failure strengthening materials are listed in Table 3 together with the
load Vexp / Vcal is 1.39 and the lowest coefficient of variation is properties of the steel reinforcement.
21%. It is also not possible to conclude which model and which In experimental program No. 1, a total of 18 RC beams were
mechanism describe the failure best. constructed. Two were control beams and sixteen were retrofitted
with CFRP fabrics using a wet lay-up method. The typical beam
cross-sectional dimensions are shown in Fig. 3. The variables for
Experiments experimental program No. 1 are listed in Table 4. The beams were
divided into two main groups. The E group had ten beams retro-
fitted with a relatively thick layer of CFRP 共six layers or more兲.
Beam Details and Experimental Setup
The S group had six beams retrofitted with two layers of CFRP.
Two experimental programs were carried out. They involved RC Two identical beams were manufactured for each configuration
beams retrofitted with carbon FRP 共CFRP兲 fabrics bonded using and denoted as “a” and “b.” For example, the E1 configuration
wet lay-up method. The CFRP fibers used were MBrace CF 130 has two beams: E1a and E1b. A steel clamp was used on all
fibers, which are also known as S&P C-sheet 240. They are sup- beams on one side to force end debond 共if it occurred兲 on the
plied in unidirectional tow sheets of 300 mm width. The nominal other side. All beams in experiment program No. 1 were tested
thickness, based on the total thickness of fibers only in a unit under four-point bending 关Fig. 4共a兲兴.

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 49

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Comparison between the predictions from existing theoretical models and experimental results for intermediate span debond

In experimental program No. 2, eight tests on beams of similar placed on the two sides of the strap over the unbonded area close
dimensions were carried out under three-point bending to the beam soffit. To avoid slipping of the wedge over the lon-
关Fig. 4共b兲兴. The variables in this experimental program are sum- gitudinal composite, grooves were introduced to the wedge bond
marized in Table 5. Four tests were done on the undamaged side surface. The grooves were designed for one-directional slippage
of four beams tested in the first experiment. These four selected only. To further prevent slippage, epoxy resin was also injected
beams were observed to have minimum damage on the clamped between the grooves and the wedge. Slipping was however al-
end since the loading was stopped immediately after end debond lowed between the wedge and the bottom side of the U-strap. This
occurred on the other end. Four additional tests were also per- method of prestressing was designed specifically for CFRP fab-
formed on two newly manufactured beams of similar dimensions rics bonded using a wet lay-up method since the fibers cannot be
as E1. stressed prior to the formation of the composite to avoid breaking
Out of these eight tests, one was carried out on the beam of the individual strands. Another prestressing method is also
without anchorage. In the other seven tests, the beams were an- available for CFRP thermoplastic tape and has been reported else-
chored with nonprestressed or prestressed CFRP U-straps placed where 共Lees et al. 2002兲.
either at the CFRP end only or at a spacing of 180 mm in the The process of applying the CFRP fabrics to concrete involved
shear span. In the nonprestressed anchorage system, two plies of two main steps: Surface preparation and bonding. The concrete
CFRP fabrics of 50 mm width were wrapped and bonded around surface was prepared using a high-pressure water jet to remove a
the sides and the soffit of the concrete beam near the end of the thin layer of the paste to expose the coarse aggregates. The water
longitudinal CFRP 关Fig. 5共a兲兴. In the prestressed system, a gap jet operated at 4,000 psi or 28 MPa. To ensure maximum bond,
was introduced between the strap and concrete soffit. Prestressing MBrace Primer was applied on the surface thoroughly with a
was introduced into the sides of the CFRP strap by inserting a brush. Bonding operation included resin undercoating, carbon
wedge into the gap 关Figs. 5共b and c兲兴. More description of this fiber sheet application, and resin overcoating.
new prestressed system is followed.
The strap had at a slope of 1:20 to the vertical, which was also Experimental Results
the wedge surface slope. This slope was chosen so that a pre-
stressing strain of 500⫻ 10−6 could be introduced to the strap The failure loads and failure modes for all beams tested in experi-
sides. The prestressing strain was monitored by two strain gauges mental program No.1 are listed in Table 6. The load deflection

Fig. 2. Comparison between the predictions from existing theoretical models and experimental results for end debond.

50 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


Table 2. Assessment Results of Some Existing Models
Mean of Coefficient of
Failure type Model Vexp / Vcal variation 共%兲
Intermediate Maruyama and Ueda 1.49 21
span debond 共2001兲
Teng et al. 共2001兲 1.66 23
Blaschko et al. 共1997兲 1.39 29
End debond El-Mihilmy and Tedesco 3.92 45
共2001兲
Chaallal et al. 共1998兲 8.72 119
Oehlers 共1992兲
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

1.81 23
Jansze 共1997兲 1.41 25

Table 3. Material Mechanical Properties


Material E 共MPa兲 f sy 共MPa兲 f t 共MPa兲 Fig. 4. Loading setup for four-point bending 共a兲 and three-point
Steel 共Y12兲 205,000 551 — bending 共b兲
Steel 共Y10兲 204,000 334 —
Steel 共Y6兲 238,000 423 —
CFRP fabrics 209,000 — 3,900*
curves of four example beams in this program 共Beams C1a, E1a,
Adhesive ⬎3,500* — ⬎120*
*
S2a, and S1a兲 are plotted in Fig. 6. Beam C1a failed by typical
Note: given by the manufacturer. steel yielding followed by the secondary compression failure of
concrete. Beam E1a failed by end debond 关Fig. 7共a兲兴. Flexural
vertical cracks were observed first in the pure bending region of
the beam. As the load increased, shear cracks became visible in
the shear span. At 39 kN shear load level, a shear crack originat-
ing from the CFRP composite end was observed. The shear cracks
widened progressively as the load increased. The portion of the
end shear crack in the concrete cover layer became more inclined
and finally joined the adjacent shear crack. At 59 kN, this crack
opened further and about 100 mm of the composite was separated
from the concrete. The crack propagated further into the shear
span until the load reached a peak of 71 kN.
Beam S2a failed by intermediate span debond 关Fig. 7共b兲兴. The
tensile steel in this beam yielded at around 65 kN. Flexural and
flexure-shear cracks were clearly visible after that. A wide flex-
ural shear crack was observed under the load point and delami-
nation of the laminate from concrete was initiated at its tip as the
shear load reached 78 kN. This was followed by gradual delami-
Fig. 3. Beam E1 cross section
nation of the composite from the beam along the bond surface
toward the beam end. The load continued to increase to a peak of
80.4 kN, when the concrete cover near the tip of the flexure-shear

Table 4. Variables in Experimental Program No. 1


Concrete Concrete
Beam strength As cover Af L0
designation 共MPa兲 ns ⫻ dias diasv − s 共mm兲 nf ⫻ tf 共mm兲 U-strap
C1 53.7 3 ⫻ 12 10–125 24 N/A N/A No
E1 53.7 3 ⫻ 12 10–125 24 6 ⫻ 0.176 150 No
E2 53.7 3 ⫻ 12 10–125 24 6 ⫻ 0.176 350 No
E3 53.7 2 ⫻ 12 10–125 24 6 ⫻ 0.176 150 No
E4 53.7 3 ⫻ 12 10–125 44 6 ⫻ 0.176 150 No
E5 53.7 3 ⫻ 12 10–125 24 9 ⫻ 0.176 150 No
S1 47.7 3 ⫻ 12 06–125 24 2 ⫻ 0.176 150 No
S2 47.7 3 ⫻ 12 06–90 24 2 ⫻ 0.176 150 No
S3 47.7 2 ⫻ 12 06–125 24 2 ⫻ 0.176 150 No
Note: ns ⫻ dias = number of tension steel bars x bar diameter 共mm兲; diasv − s = stirrup diameters 共mm兲 − spacing 共mm兲; n f ⫻ t f = number of plies x ply
thickness 共mm兲; and L0 = distance from end of fiber reinforced polymer to nearest support 共mm兲.

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 51

J. Compos. Constr. 2006.10:48-59.


Table 5. Variables in Experimental Program No. 2
Concrete Concrete
Beam strength As cover Af L0
designation 共MPa兲 ns ⫻ dias diasv − s 共mm兲 nf ⫻ tf 共mm兲 U-strap
E3b2 53.7 2 ⫻ 12 10–125 24 6 − 0.176 150 None
E3a2 53.7 2 ⫻ 12 10–125 24 6 − 0.176 150 1-N
E1b2 53.7 3 ⫻ 12 10–125 24 6 − 0.176 150 1-P2
E5a2 53.7 3 ⫻ 12 10–125 24 9 − 0.176 150 3-P
A1a 53.7 3 ⫻ 12 10–125 24 6 − 0.176 150 1-N
A1b 53.7 3 ⫻ 12 10–125 24 6 − 0.176 150 3-N
A2a 53.7 3 ⫻ 12 10–125 24 6 − 0.176 150 1-P
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

A2b 53.7 3 ⫻ 12 10–125 24 6 − 0.176 150 3-P


Note: 1-N = one end nonprestressed U-strap; 3-N = three nonprestressed U-straps at 180 mm spacing; 1-P = one end prestressed U-strap; 3-P = three
prestressed U-straps at 180 mm spacing; and 1-P2 = end strap was moved closer to the beam middle by 60 mm due to unsuccessful prestressing of the first
end strap.

Fig. 5. Two anchorage systems used

Table 6. Experimental Program No. 1 Results


Vexp f f.max ␶cv.d Vcal
Designation Failure modes 共kN兲 共microstrain兲 共MPa兲 共kN兲
Ca Concrete crushing 54.4 — 50.7
Cb Concrete crushing 53.9 — 51.7
E1a End debond 70.7 3,036 0.87 75.6
E1b End debond 74.6 3,414 0.98 76.7
E2a End debond 51.4 1,740 0.78 55.7
E2b End debond 53.4 1,988 0.90 56.2
E3a End debond 66.0 3,502 1.00 66.2
E3b End debond 65.2 3,996 1.15 66.0
E4a End debond 79.0 3,386 0.97 80.0
E4b End debond 61.2 2,286 0.66 79.5
E5a End debond 63.3 2,329 1.00 66.4
E5b End debond 63.2 2,350 1.01 67.6
S1a Intermediate span and end debond 73.8 9,445 0.90 70.4
S1b Intermediate span and end debond 74.5 9,452 0.90 71.4
S2a Intermediate span debond 80.4 9,014 0.86 72.2
S2b Intermediate span debond 74.5 7,564 0.72 70.7
S3a Intermediate span debond 60.3 8,707 0.83 55.4
S3b Intermediate span and end debond 60.2 8,292 0.79 55.5

52 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Load-deflection curves of Beams C1a, E1a, S1a, and S2a

crack broke from the beam leading to complete separation of the


composite from that location to the steel clamp.
Beam S1a failed by a combination of intermediate span de-
bond and end debond. At ultimate, concrete fracture occurred
simultaneously from the unclamped end of the composite and
from the tip of a wide flexure-shear crack near the middle of the
shear span. Stress transferred and concrete cracking propagated
toward the middle of the beam along the tension reinforcement
level.
The variations in CFRP longitudinal strain and interfacial bond
stress at different load levels for three example beams E1a, S2a,
and S1a are plotted in Figs. 8–10. The strain was found to drop
from the maximum value under the load point to a zero at the end
of the plate for all beams. As expected, the maximum strain level
in S beams was much higher than that in E beams. The average Fig. 8. Carbon fiber reinforced polymer strain distributions 共a兲 and
interfacial shear stress distributions 共b兲 in Beam E1a
bond stress between the two strain gauge locations was calculated
by dividing the force difference by the bond area and is given by
the following equation: clamped side of Beam S2a, where the end debond was prevented,
E f 共␧ f,i+1 − ␧ f,i兲t f this beam failed by intermediate span debond.
␶= 共1兲 The failure loads and failure modes for all beams tested in
⌬L
experimental program No. 2 are listed in Table 7. The loading
where E f and t f = CFRP elastic modulus and thickness, respec- curves are plotted in Figs. 11 and 12. The end U-strap proved to
tively. ␧ f,i+1 and ␧ f,i = CFRP strains; and ⌬L = distance between the be effective in limiting end debond, but not intermediate span
strain gauges. The average bond stress distributions of Beam E1a debond. Since the strap increased concrete shear resistance at ten-
show a clear peak near the FRP end indicating end debond mode. sion reinforcement level even after cracks were formed there, the
With the exception of the distributions at the peak load, the aver- failure mode shifted to intermediate span debond at a higher load.
age shear stress distributions of Beams S1a and S2a seem to have This was observed in the beams with one end strap, i.e., Beams
two peaks. This is the indication of the mixed mode of end and E3a2, E1a2, A1a, and A2a 关Figs. 13共a and b兲兴. The increases in
intermediate span debond. Since the failure happened to be on the the ultimate capacity compared to the corresponding beams with-

Fig. 7. Failure modes observed in experimental program No. 1

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 53

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Carbon fiber reinforced polymer strain distributions 共a兲 and
interfacial shear stress distributions 共b兲 in Beam S2a
Fig. 9. Carbon fiber reinforced polymer strain distributions 共a兲 and
interfacial shear stress distributions 共b兲 in Beam S1a

Proposed Models

out end straps are 15, 32, 34, and 44%, respectively. Two prediction models are presented in this section. The models
The experiments also demonstrated that the intermediate span rely heavily on the capacity of the beam theory to analyze a
debond strength of a retrofitted beam can, however, be improved retrofitted section. Therefore, a verification study of the theory is
by placing U-straps in the shear span at a certain spacing. These presented first below.
straps crossed and limited the opening of flexure-shear cracks and
also increased the bond strength between the longitudinal CFRP
Verification of the Beam Theory
and concrete by confinement under the strap. As a result of that,
the ultimate capacity of the beam increased significantly. Beams The beam theory has been previously proven to be able to predict
E5a2, A1b, and A2b failed by debonding or rupture of the legs of the ultimate load for beams failing by concrete crushing or FRP
the U-straps near the load point followed by intermediate span rupture 共Pham and Al-Mahaidi 2004兲. The aim of the current
debonding 关Figs. 13共c and d兲兴. The increase in the ultimate capac- study is to verify the application of the beam theory to compute
ity was up to 79% compared to the corresponding beams without reinforcement strain levels at any applied moment, which is es-
straps. sential for the debond models proposed later.
For the beams with nonprestressed straps, sliding of the longi- To analyze a retrofitted section subjected to an applied mo-
tudinal plate underneath the U-strap before failure was observed ment, the basic assumption that the strain varies linearly along the
visually. The sliding action caused bending of the strap legs near section from the top to the bottom fiber was assumed. The con-
the beam soffit, and led to local debonding and rupture of these stitutive properties of the component materials are plotted in
legs before the longitudinal CFRP was completely separated from Fig. 14. The nonlinear stress-strain curve by Hognestad 共1951兲
the beam. For the beams with prestressed straps, no evidence of was used to describe concrete behavior in compression, where the
sliding of the main composite near its end was found after failure compressive strain, ␧0, at peak was given as 2f c / Ec.
due to the increase in concrete shear capacity, which was the Comparisons of CFRP strain distributions for the tested beams
result of the compressive stress caused by the prestressing force. are plotted in Fig. 15. The plots show that the beam theory out-
The legs of the U-strap did not rupture, but delaminated from the lined above can predict the peak CFRP strain level with good
concrete surface. A slight increase of around 5% in the ultimate accuracy. However, the predicted FRP strains in the shear span
capacity was recorded compared to the beams with nonpre- deviate significantly from the measured values at high loads. This
stressed straps. is due to the presence of shear and flexure-shear cracks. Deviation

54 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


Table 7. Experimental Program No. 2 Results
Vexp. Increase Vcal
Designation Failure modes Anchoragea 共kN兲 共compared to兲 共kN兲
E3b2 End debond None 60.0 N/A 66.0
E3a2 Intermediate span debond and rupture of end strap 1-N 75.9 15% 共E3a兲 70.5
E1b2 Intermediate span debond and debond of end strap 1-P2 98.1 32% 共E1b兲 83.6
E5a2 Debond of straps and intermediate span debond 3-P 113.1 79% 共E5a兲 —
A1a Intermediate span debond and rupture of end strap 1-N 94.5 34% 共E1a兲 83.6
A1b Debond and rupture of straps and intermediate span 3-N 108.7 54% 共E1a兲 —
debond
A2a Intermediate span debond and debond of end strap 1-P 101.7 44% 共E1a兲 83.6
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

A2b Debond of straps and intermediate span debond 3-P 114.5 62% 共E1a兲 —
a
See the notes for Table 4.

occurs most clearly when the steel reinforcement undergoes yield- plane 关Fig. 17共a兲 Surface A兴 and the bond area between concrete
ing. For those cases, a more gradual decrease of the FRP strain is and steel rebars 关Fig. 17共a兲 Surface B兴. The shear stress is also not
observed 关Fig. 15共b兲兴. uniform over the failure surface. However, Surface B is generally
small as compared to Surface A, and the bond strength between
the steel rebars and the concrete cover 共unconfined concrete兲 is
Derivation of Models to Predict Debonding Failure
also possibly small compared to concrete shear strength. There-
Loads
fore, for simplification, the average shear stress is assumed to be
Two debond models are presented here based on the failure distributed over the failure of Surface A only. End debond can be
mechanisms depicted in Fig. 16. assumed to occur when the following inequality is satisfied

End Debond Model ␶ave 艌 f cv.d 共2兲


End debond is due to the tensile force F f,0 on the FRP end. The where ␶ave = average shear stress on the assumed debond surface
tensile force leads to a high shear stress level at the weakest layer 共Surface A兲; and f cv.d⫽shear strength of the surface. To calculate
near the tension reinforcement level, inducing a longitudinal ␶ave, the force distribution in the composite plate can be assumed
debonding crack there. Therefore, it can be envisaged that end to be linear over the shear span based on the observation from the
debond occurs when the shear stress in the weakest layer reaches experiments described previously 关Fig. 17共b兲兴. Therefore, the av-
a limiting value. This concept was first introduced by Mukho- erage shear stress can be estimated to be
padhyaya and Swamy 共2001兲. The researchers stated that the in-
terface shear stress value between the composite and concrete ␶ave = F f.max/L f 共3兲
could be used as a simple and practical design criterion to predict
The shear strength at the tensile reinforcement layer, f cv.d is
the failure at the adhesive-concrete interface. However, it was
difficult to determine. Limited experimental data are available on
pointed out in the review section of this paper that the critical
the shear strength of plain concrete when the normal stress is
failure surface is not at the adhesive-concrete interface but at the
negligible 共which is the case here兲. In addition, the strength is also
layer near tension reinforcement. Nevertheless, the interface shear
affected by a number of factors including the stress state in the
stress is closely related to the stress along the weakest plane since
longitudinal steel bars, the location of the vertical steel bars, and
the concrete cover is generally small compared to the CFRP
the location of flexural and flexure-shear cracks. However, it can
bonded length.
be assumed that the shear strength is proportional to the square
Strictly speaking, the tensile force in the composite is trans-
root of concrete compressive strength 共Standards Australia 2001兲.
ferred through shear stress on the concrete area at the weakest

Fig. 11. Load-deflection curves of retested beams Fig. 12. Load-deflection curves of A beams

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 55

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Failure modes observed in experimental program No. 2

Calibration using the strain measurements for ten beams tested by et al. 共2001兲. They calculated the maximum pulling force in the
the writers showed that the limiting average stress at which end FRP in a shear-lap test using Chen and Teng’s formula 共Chen and
debond occurs, f cv.d, was 1.22 MPa, which was 0.17冑 f c. This is Teng 2001兲.
P = ␣␤ f ␤L冑 f cb f Le
demonstrated in Fig. 18共a兲.
共4兲
Intermediate Span Debond Model where

冑冑 冑
Intermediate span debond is the result of the high tensile force F f,i
in the CFRP at the tip of a main flexural crack or flexure-shear Eftf 2 − b f /bc
Le = ; ␤f = ;
crack. Peeling occurs in the concrete substrate next to the bond fc 1 + b f /bc

再 冎
surface and propagates away from the initial location. When peel-
ing occurs at the tip of a flexure-shear crack, cracking near the 1 if L 艌 Le
tension reinforcement level also occurs when the tension is trans- ␤L = 共5兲
sin共␲L/2Le兲 if L ⬍ Le
ferred to the adjacent concrete tooth formed between the two
flexure-shear cracks. For both cases, the maximum tensile stress and bc = concrete width. b f , t f , and E f = CFRP width, thickness,
that the composite can take depends on the shear strength of the and elastic modulus, respectively. L = bond length; and
concrete substrate. This strength can be determined using the re- f c = concrete compressive strength. ␣ = calibration factor to ac-
sults of shear-lap testing. This observation was first made by Teng count for any difference between the behavior of a beam failed by

Fig. 14. Stress-strain relationships for fiber reinforced polymer 共a兲, steel 共b兲, and concrete 共c兲

56 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Comparison of fiber reinforced polymer strain distributions in E1a 共a兲 and S1a 共b兲 between experiments and beam theory predictions

intermediate span debond and a shear-lap specimen failed by in- Conclusions


terfacial debond. Despite many similarities, the situation in a ret-
rofitted beam is much more complex due to the presence of sev- This paper presented a study of the behavior of beams retrofitted
eral factors, such as bending deformation, shear deformation, and with CFRP fabrics bonded to their soffits. In the experiments, two
the presence of tensile steel reinforcement. Teng et al. 共2001兲 brittle failure modes were observed. They were debonding at the
calibrated ␣ with eight beams and nine slabs and suggested a FRP end and debonding at a flexure-shear crack. In the first
value of 0.4. Similar calibration study was carried out using the debond mode, the propagation plane was near the tension rein-
writers’ experimental results reported in the review section. The forcement level; whereas in the second mode, it was near the
study indicates that ␣ can be taken as 1.04. This is demonstrated bond surface. High shear stress levels that developed either at the
in Fig. 18共b兲. end of CFRP or at a tip of a flexure-shear crack were the cause of
these failure modes.
Model Verification Based on the findings from both the experimental programs,
two design methods are proposed to predict end and intermediate
The two methods described above were first used to calculate the span debond. They have been validated and proved to be reliable.
failure loads for the beams reported in the experimental section. One effective way of limiting debonding is to provide CFRP
The calculation results are shown in Tables 6 and 7 showing good U-straps. These straps can limit end debond by improving con-
agreement with the experiments. Further verification was carried
out using the database described in the review section with the
results shown in Fig. 19. Good agreement is also observed for
most of the beams. The average of the ratio of the experimental
results to the predicted failure loads is 1.01 and 1.05 for interme-
diate span and end debond, respectively. The coefficients of varia-
tion are 13 and 19%, respectively. Compared with the verification
results of other models mentioned in the review section, the two
methods proposed here produce more accurate and less scattered
predictions.

Fig. 16. Failure mechanisms Fig. 17. Average shear stress concept

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 57

J. Compos. Constr. 2006.10:48-59.


Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 18. Calibration for debond models

Fig. 19. Comparison between the predictions by the proposed models and experimental results

crete shear resistance at tension steel level. When placed at cer- f cv.d ⫽ average concrete shear strength at the tensile
tain spacing 共180 mm in the experiments reported here兲 in the reinforcement level;
shear span, intermediate span debond can also be limited as the f fu ⫽ tensile strength of a FRP composite;
opening of flexure-shear cracks is restricted and the bond between f sy ⫽ yield strength of tension reinforcement steel;
the longitudinal CFRP and concrete is improved. Prestressed f t ⫽ direct tensile strength;
U-straps tend to perform slightly better than nonprestressed ones. L f ⫽ bond length;
t f ⫽ FRP elastic thickness;
Vcal ⫽ calculated maximum shear force that a beam can
Acknowledgments support;
Vexp ⫽ actual maximum shear force that a beam can support;
The research reported in this paper is part of a study financed by ␣ ⫽ calibration factor;
a Monash Univ. Engineering Grant. All CFRP fabrics and adhe- ⌬L ⫽ distance between strain gauges;
sives were partly sponsored by MBT 共Australia兲 Pty Ltd. ␧ f,i ⫽ FRP strain at location i; and
␶ave ⫽ average shear stress at the tensile reinforcement level.

Notation
References
The following symbols are used in this paper:
bc ⫽ concrete beam width; Ahmed, O., and Van Gemert, D. 共1999兲. “Effect of longitudinal carbon
b f ⫽ FRP width; fiber reinforced plastic laminates on shear capacity of reinforced con-
E f ⫽ FRP elastic modulus; crete beams.” Proc., 4th Int. Symp. on Fiber Reinforced Polymer Re-
F f,i ⫽ tensile force in FRP; inforcement for Reinforced Concrete Structures, 933–943.
F f,max⫽ tensile force in FRP under the load point; Arduini, M., and Nanni, A. 共1997兲. “Behavior of precracked RC beams
f c ⫽ concrete compressive strength; strengthened with carbon FRP sheets.” J. Compos. Constr., 1共2兲,
f ct ⫽ concrete tensile strength; 63–70.

58 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006

J. Compos. Constr. 2006.10:48-59.


Beber, A. J., Filho, A. C., and Campagnolo, J. L. 共1999兲. “Flexural new design criterion for plate debonding.” J. Compos. Constr., 5共1兲,
strengthening of R/C beams with CFRP sheets.” Structural faults and 35–43.
repair ’99, Engineering Technics Press, Edinburgh, Scotland. Mukhopadhyaya, P., Swamy, R. N., and Lynsdale, C. J. 共1998兲. “Opti-
Blaschko, M. 共1997兲. “Strengthening with CFRP.” Munchner Massivbau mizing structural response of beams strengthened with GFRP plates.”
Seminar, TU Muchen 共in German兲. J. Compos. Constr., 2共2兲, 87–95.
Bonacci, J. F., and Maalej, M. 共2000兲. “Externally bonded fiber- Nguyen, D. M., Chan, T. K., and Cheong, H. K. 共2001兲. “Brittle failure
reinforced polymer for rehabilitation of corrosion damaged concrete and bond development length of CFRP-concrete beams.” J. Compos.
beams.” ACI Struct. J., 97共5兲, 703–711. Constr., 5共1兲, 12–17.
Chaallal, O., Nollet, M. J., and Perraton, D. 共1998兲. “Strengthening of RC Oehlers, D. J. 共1992兲. “Reinforced concrete beams with plates glued to
beams by externally bonded fiber-reinforced-plastic plates: Design their soffits.” J. Struct. Eng., 118共8兲, 2023–2038.
guidelines for shear and flexure.” Can. J. Civ. Eng., 692–764. Pham, H. B., and Al-Mahaidi, R. 共2003兲. “An investigation into debond-
Chajes, M. J., Thomson, T. A. J., Januszka, T. F., and Finch, W. W. J. ing of FRP in flexural strengthening.” Proc., 21st ARRB & 11th
共1994兲. “Flexural strengthening of concrete beams using externally
Downloaded from ascelibrary.org by IMPERIAL COLLEGE LONDON on 06/09/14. Copyright ASCE. For personal use only; all rights reserved.

REAAA Conf.
bonded composite materials.” Constr. Build. Mater., 8共3兲, 191–201. Pham, H. B., and Al-Mahaidi, R. 共2004兲. “Assessment of available pre-
Chen, J. F., and Teng, J. G. 共2001兲. “Anchorage strength models for FRP diction models for the strength of FRP retrofitted RC beams.” Com-
and steel plates bonded to concrete.” J. Struct. Eng., 127共7兲, pos. Struct., 66共1–4兲, 601–610.
784–791. Pornpongsaroj, P., and Pimanmas, A. 共2003兲. “Effect of end wrapping on
David, E., Djelal, C., Ragneau, E., and Bodin, F. B. 共1999兲. “Use of FRP peeling behavior of FRP-strengthened beams.” Proc., FRPRCS-6—
to strengthen and repair RC beams: Experimental study and numerical Fibre-Reinforced Polymer Reinforcement for Concrete Structures,
simulations.” Proc., 8th Int. Conf. on Advanced Composites for Con- 277–286.
crete Repair. Quantrill, R. J., Hollaway, L. C., and Thorne, A. M. 共1996兲. “Experimen-
El-Mihilmy, M. T., and Tedesco, J. W. 共2001兲. “Prediction of anchorage tal and analytical investigation of FRP strengthened beam response:
failure for reinforced concrete beams strengthened with fiber- Part I.” Mag. Concrete Res., 48共177兲, 331–342.
reinforced polymer plates.” ACI Struct. J., 98共3兲, 301–314. Rahimi, H., and Hutchinson, A. 共2001兲. “Concrete beams strengthened
Fanning, P. J., and Kelly, O. 共2001兲. “Ultimate response of RC beams with externally bonded FRP plates.” J. Compos. Constr., 5共1兲, 44–56.
strengthened with CFRP plates.” J. Compos. Constr., 5共2兲, 122–127. Ritchie, P. A., Thomas, D. A., Lu, L.-W., and Connelly, G. M. 共1991兲.
Gao, B., Leung, W.-H., Cheung, C.-M., Kim, J.-K., and Leung, C. K. Y. “External reinforcement of concrete beams using fiber reinforced
共2001兲. “Effects of adhesive properties on strengthening of concrete plastics.” ACI Struct. J., 88共4兲, 490–500.
beams with composite strips.” FRP composites in civil engineering, Ross, C. A., Jerome, D. M., Tedesco, J. W., and Hughes, M. L. 共1999兲.
Elsevier, Hong Kong, 423–432. “Strengthening of reinforced concrete beams with externally bonded
Garden, H. N., Hollaway, L. C., and Thorne, A. M. 共1997兲. “A prelimi- composite laminates.” ACI Struct. J., 96共2兲, 212–220.
nary evaluation of carbon fibre reinforced polymer plates for strength- Saadatmanesh, H., and Ehsani, M. R. 共1991兲. “RC beams strengthened
ening reinforced concrete members.” Proc. Inst. Civ. Eng., Struct. with GFRP plates. I. Experimental study.” J. Struct. Eng., 117共11兲,
Build., 127–142. 3417–3433.
Garden, H. N., Quantrill, R. J., Hollaway, L. C., Thorne, A. M., and Sharif, A., Al-Sulaimani, G. J., Basunbul, I. A., Baluch, M. H., and Gha-
Parke, G. A. R. 共1998兲. “An experimental study of the anchorage leb, B. N. 共1994兲. “Strengthening of initially loaded reinforced con-
length of carbon fibre composite plates used to strengthen reinforced crete beams using FRP plates.” ACI Struct. J., 91共2兲, 160–168.
concrete beams.” Constr. Build. Mater., 12, 203–219. Smith, G., and Teng, J. G. 共2003兲. “Shear-bending interaction in debond-
Hau, K. M. 共1999兲. Experiments on concrete beams strengthened by ing failures of FRP-plated RC beams.” Adv. Struct. Eng., 6共3兲,
bonding fibre reinforced plastic sheets, The Hong Kong Polytechnic 183–199.
Univ., Hong Kong, China. Smith, S. T., and Teng, J. G. 共2002兲. “FRP-strengthening RC beams. I:
He, J. H., Pilakoutas, K., and Waldron, P. 共1997兲. “Strengthening of re- Review of debonding strength models.” Eng. Struct., 24, 385–395.
inforced concrete beams with CFRP plates.” Non-Metallic (FRP) Re- Standards Australia. 共2001兲, “Concrete structures.” AS3600–2001 Austra-
inforcement for Concrete Struct. Proc., 3rd Symp., 343–350. lian Standard, Sydney, Australia.
Hognestad, E. 共1951兲. “A study of combined bending and axial load in Takahashi, Y., and Sato, Y. 共2003兲. “Flexural behavior of RC beams ex-
reinforced concrete members.” Eng. Expt. Station Bull., No. 399, Univ ternally reinforced with carbon fiber sheet.” Proc., FRPRCS-6—
of Illinois. Fibre-Reinforced Polymer Reinforcement for Concrete Structures,
Jansze, W. 共1997兲. Strengthening of RC members in bending by externally 237–246.
bonded steel plates, Delft Univ. of Technology, Delft. Teng, J. G., Smith, G., Yao, J., and Chen, J. F. 共2001兲. “Strength model
Khomwan, N., Foster, S. J., and Smith, S. T. 共2004兲. “Debonding failure for intermediate flexural crack induced debonding in RC beams and
of CFRP strengthened concrete beams.” FRP Composites in Civil slabs.” FRP composites in civil engineering, Elsevier, Hong Kong,
Engineering—Proc., CICE 2004, 505–513. 579–587.
Kishi, N., Mikami, H., Matsuoka, K. G., and Kurihashi, Y. 共2001兲. “Fail- Triantafillou, T. C., and Plevris, N. 共1992兲. “Strengthening of RC beams
ure behavior of flexural strengthened RC beams with AFRP sheet.” with epoxy-bonded fibre-composite materials.” Mater. Struct., 25,
Proc., FRPRCS-5—Fibre-Reinforced Plastics for Reinforced Concrete 201–211.
Structures, 85–95. Tumialan, G., Serra, P., Nanni, A., and Belarbi, A. 共1999兲. “Concrete
Lees, J. M., Winistörfer, A., and Meier, U. 共2002兲. “External prestressed cover delamination in reinforced concrete beams strengthened with
carbon fiber-reinforced polymer straps for shear enhancement of con- carbon fiber reinforced polymer sheets.” Proc. 4th Int. Symp. on Fiber
crete.” J. Compos. Constr., 6共4兲, 249–256. Reinforced Polymer Reinforcement for Reinforced Concrete Struc-
Leong, K. S., and Maalej, M. 共2003兲. “Effect of beam size on interfacial tures, 725–735.
shear stresses and failure mode of FRP-bonded beams.” Proc. Valcuende, M., and Benlloch, J. 共2003兲. “Ductility of reinforced concrete
FRPRCS-6—Fibre-Reinforced Polymer Reinforcement for Concrete beams strengthened with CFRP strips and fabric.” Proc. FRPRCS-6—
Structures, 257–266. Fibre-Reinforced Polymer Reinforcement for Concrete Structures,
Maruyama, K., and Ueda, T. 共2001兲. “JSCE design recommendations for 337–346.
upgrading of RC members by FRP sheet.” Proc. FRPRCS-5—Fibre- Zarnic, R., and Bosiljkov, V. 共2001兲. “Behaviour of beams strengthened
Reinforced Plastics for Reinforced Concrete Structures, 441–446. with FRP and steel plates.” Proc., 2001 Structural Congress and
Mukhopadhyaya, P., and Swamy, R. N. 共2001兲. “Interface shear stress: A Exposition.

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JANUARY/FEBRUARY 2006 / 59

J. Compos. Constr. 2006.10:48-59.

You might also like