Composite Structures 189 (2018) 9–26

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Flexural strengthening of RC beams with CFRP grid-reinforced ECC matrix T

a,b c b,⁎ a a,b
Xu Yang , Wan-Yang Gao , Jian-Guo Dai , Zhou-Dao Lu , Ke-Quan Yu
a
College of Civil Engineering, Tongji University, Shanghai 200092, China
b
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong 999077, China
c
School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

A R T I C L E I N F O A B S T R A C T

Keywords: This paper investigates the flexural performance of a series of RC beams externally bonded with carbon fiber-
RC beam reinforced polymer (CFRP) grid-reinforced engineered cementitious composite (ECC) matrix. A total of 15 RC
Flexural strengthening beams, including three control and twelve strengthened, were prepared and tested. The test variables included
CFRP grid the longitudinal reinforcement ratio, the strengthening configurations that consisted of different cementitious
ECC
matrices (ECC versus epoxy mortar), different installation methods (prefabricated versus cast-in-place), and
Prefabricated plate
different stiffness of CFRP grids. The test results showed that ECC is an ideal cementitious matrix for the
Debonding
strengthening applications where FRP grids are used as the external reinforcement. The flexural strengthening
configuration using the epoxy adhesive to bond prefabricated CFRP grid-reinforced ECC plate proved to be the
most efficient solution. For such configuration, the plate-end debonding can be avoided and the mid-span de-
bonding can be almost suppressed. Flexural capacity analysis was conducted and demonstrated that the plane
section assumption is valid and the full strength composite action can be nearly achieved for the strengthening
system. The average ratio of the predicted peak loads to the experimental ones of the strengthened RC beams was
1.05.

1. Introduction These weaknesses have compromised the attractiveness of the FRP

strengthening technique in some applications (e.g., applications in in-
The strengthening of reinforced concrete (RC) structures has tradi- door or moist environments or with significant temperature variations).
tionally been achieved by the enlargement of member sections using For this reason, attempts have been made to replace the epoxy in FRP
concrete/cement mortar overlays or the external bonding/bolting of by an inorganic/cementitious material, leading to two major cement-
steel plates as an additional reinforcement. Over the past three decades, based strengthening systems: (a) dry fiber sheets/textiles bonded with a
the technique of bonded (externally bonded or near-surface mounted) cementitious material and (b) FRP grids bonded with cement mortar.
fiber-reinforced polymer (FRP) reinforcement has become a highly at- The cementitious materials explored for bonding dry fiber sheets/
tractive alternative to the traditional overlaying techniques. This is due textiles for the flexural strengthening of RC beams included alumino-
to the high strength-to-weight ratio and excellent corrosion resistance silicate powder blended with a water-based activator [14,15], polymer
of FRP composites, making the bonded FRP strengthening technique a modified cement mortars and silicate fume/micro-cement added mor-
quick and efficient solution with only minimal changes to the weight tars [16–19], and geopolymer mortar [20]. In all these studies, fracture
and dimensions of the structures. of fibers was observed as an overwhelming failure mode. The average
FRP composites have their own weaknesses, which are mainly as- tensile rupture strains of fibers observed in the above cementitious
sociated with the poor fire performance (flammability, smoke genera- matrices are much lower than those found in epoxy-based matrices. It
tion and rapid degradation in mechanical properties) of the polymer has been reported that the average rupture strain of dry carbon fiber
matrix (usually an epoxy) system [1–8], degradations of the polymer sheets bonded with a cementitious matrix is only 0.6% [15]. Wiberg
matrix under UV radiations, and the low compatibility between the [21] showed through his tests that to achieve the same degree of
epoxy matrix and the concrete substrate [9–11]. Also, the thermal in- flexural strengthening, the amount of carbon fibers needed for the ce-
compatibility between the FRP composites and concrete substrates may ment-based matrix is more than double that needed for the epoxy-based
induce significant thermal stresses at the bond interface between FRP counterpart. Wu and Sun [17] reported similar results.
and concrete that may accelerate the interface debonding [12,13]. Cement mortars reinforced with textiles, which are made of either

⁎
Corresponding author.
E-mail addresses: 123yangxu@tongji.edu.cn (X. Yang), wanyanggao@sjtu.edu.cn (W.-Y. Gao), cejgdai@polyu.edu.hk (J.-G. Dai), luzhoudao@tongji.edu.cn (Z.-D. Lu).

https://doi.org/10.1016/j.compstruct.2018.01.048
Received 16 October 2017; Received in revised form 5 January 2018; Accepted 15 January 2018
Available online 20 February 2018
0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 1. Details of CFRP grid, ECC and epoxy mortar spe-

cimens.

woven, knitted or even unwoven rovings in at least two (typically or- multiple cracking behavior during the strain-hardening process that
thogonal) directions, have also been used for the flexural and shear may alleviate the stress concentration and suppress the debonding at
strengthening of RC beams and for the jacketing of columns the interface between the strengthening layer and the substrate con-
[11,21–28]. Although the bonding property of the textile fabrics in the crete [43,44]. ECC also can be prepared as a pre-cast repair system with
cementitious matrix is improved compared to the dry fiber sheets be- a multi-layer structure [45]. However, it should be noted that the use of
cause of the geometrical configuration, Triantafillou and Papanicolaou ECCs alone for the structural strengthening of existing RC members may
[11,22] concluded that the textile-reinforced mortar (TRM) strength- not be efficient due to their relatively low tensile strength. The con-
ening system is only about 50% as effective as the epoxy-based FRP tribution by ECCs to the load-carrying capacity of the whole structural
system even in applications such as jacketing RC columns where the system may be limited [46,47] unless a very thick ECC overlay, which
bond between the jacket and the concrete is not a critical concern. may not be preferred in practice, is applied.
The matrix material in a fiber composite has two important func- Considering the above background, this paper aims to investigate
tions: to transfer stresses between the fibers and the matrix, and to the flexural strengthening performance of RC beams using CFRP grid-
protect the fibers. The ineffectiveness of fibers in the above-mentioned reinforced ECC matrix. It should be noted that efforts have been made
two types of cementitious matrix systems can be attributed to two to use ECCs together with textile reinforcement for structural
factors: (a) insufficient impregnation and thus inefficient stress transfer strengthening purpose, validating that ECC is an excellent matrix and
between the fibers and the matrix; and (b) brittleness of the cementi- bonding agent of the latter [48]. Recently Zheng et al. [49] also con-
tious matrix, leading to early cracking of the matrix material. Both ducted tests on six RC beams strengthened by basalt FRP (BFRP) grid-
factors can result in poor bonding between the fibers and the ce- reinforced ECC matrix. However, the total number of their tested spe-
mentitious matrix and unintended non-uniform stresses in fibers. It has cimens (i.e., one control beam and five strengthened beams) was rela-
been reported that the pull-out strength of rovings pre-impregnated tively limited. Also, the BFRP grid used in their study had a lower
with epoxy and then embedded in a cementitious matrix is three to four elastic modulus (50–60 GPa) compared to commercially available
times that of the same rovings directly embedded in a cementitious carbon FRP (CFRP) grids (around 100 GPa), leading to a less
matrix material [29,30]. This observation suggests that the direct em- strengthening efficiency imposing limitations on the failure mode. This
bedment of dry rovings in a cementitious matrix material should be paper will further investigate the effectiveness of this innovative
avoided to minimize the premature rupture of fibers. Therefore, the use strengthening technique by examining the effects of the type of matrix
of FRP grids, in which the fibers are already impregnated with an materials (ECC versus epoxy mortar), installation method (pre-
epoxy-based resin, bonded with a cementitious matrix material to an fabricated versus cast-in-place), tensile steel reinforcement ratio and
RC member appears to be an attractive strengthening option [31–37]. the CFRP grid stiffness on the flexural behavior of RC beams.
Engineered cementitious composite (ECC), as a special class of high-
performance fiber-reinforced cementitious material, may be a pro-
mising candidate used as a matrix material of the strengthening layer
[38–41]. ECC particularly shows a strain-hardening behavior in tension
with a large tensile strain capacity up to 3–8% [42]. It also exhibits

10
X. Yang et al. Composite Structures 189 (2018) 9–26

2. Experimental program

2.1. Material properties

CFRP grids were used for the external reinforcement and ECC or
epoxy mortar served as the cementitious matrix to impregnate the CFRP
grids to form an external overlay for flexural strengthening of RC
beams. The CFRP grids are characterized by square meshes with nom-
inal dimensions of 50 mm × 50 mm. They are made up of untwisted
yarns consisting of continuous carbon fibers impregnated with ther-
moset epoxy resin. Before strengthening, the CFRP grids were cut into
appropriate sizes according to the bottom dimensions of the RC beams,
as illustrated in Fig. 1. Two kinds of CFRP grids were used in the ex-
perimental program: one was the ordinary stiffness grid and the other
was the high stiffness one, which are designated as FTG-CR5 and FTG-
CR8, respectively, according to the product datasheet provided by the
manufacturer (Nippon Steel & Sumikin Materials Co., Ltd, Japan). The
Fig. 2. Tensile stress-strain curves of ECC.
cross-sectional area of FTG-CR5 is 13.2 mm2, which is half of the cor-
responding area of FTG-CR8 (26.4 mm2). Both FTG-CR5 and FTG-CR8
are linear elastic materials with the tensile strength and elastic modulus Table 2
Summary of mechanical properties of materials.
of 1400 MPa and 100 GPa, respectively.
The mix proportion of ECC is listed in Table 1. The major material Material Tensile Elastic Maximum Compressive Curing days
ingredients of ECC are Type I ordinary Portland cement (OPC), Class F strength modules strain (%) strength (MPa)
fly ash (FA), silica sand, polyvinyl alcohol (PVA) fiber, and water-re- (MPa) (GPa)
ducing admixture. The PVA fibers have a diameter of 40 μm and a
ECC 3.8 16.9 3.04 30.2 28
length of 12 mm. They have a tensile strength of 1600 MPa, an elastic Epoxy mortar 14.3 25 0.11 72.6 7
modulus of 42 GPa, and a maximum fiber elongation of 7.0% to match Concrete – – – 52.1 28
the requirements for strain-hardening performance. The epoxy mortar CFRP grid 1400 100 – – –
(Sikadur®-41 CF Normal), used as the other type of matrix, was supplied
by the Sika Corporation in Hong Kong. The tensile properties of both Steel reinforcement Yielding strength Ultimate strength Elastic modules
ECC and epoxy mortar were measured by dogbone-shaped specimens, (MPa) (MPa) (GPa)
as recommended by the Japan Society of Civil Engineering [50] for
Grade 500B steel bar 546 660 219
standardized testing of ECC composites. The tensile tests were carried Grade 250 steel bar 407 562 219
out under a displacement control manner with a loading rate of
0.005 mm/s. Fig. 1b shows the dogbone-shaped specimens of ECC after
7-day curing, while Fig. 1c illustrates the tensile test setup with an prepared and tested under four-point bending loading, as illustrated in
epoxy mortar sample fixed on it. The compression strengths of both ECC Fig. 3. The RC beams had three different sectional areas of ribbed
and epoxy mortar were measured using cube specimens with the di- tensile steel bars: 3Φ10 mm; 3Φ12 mm; 2Φ16 mm + 1Φ12 mm. Two
mensions of 70.7 mm × 70.7 mm × 70.7 mm. Fig. 2 shows the obtained Φ10 mm ribbed steel bars were longitudinally placed as the compres-
tensile stress-strain curves of ECC, which had a tensile strength and sive steel reinforcement. The concrete cover thickness was set as 25 mm
rupture strain of 3.8 MPa and 3.04%, respectively. for both the tensile and compressive steel bars. Plain round steel bars
A ready-mix normal weight concrete supplied by a local concrete with a diameter of 8 mm and spaced at 100 mm center-to-center were
company was used to cast all the beams at the same batch to minimize used as the stirrups. As illustrated in Fig. 3b, the strengthening overlay
the experimental scatter. The compression strength and elastic modulus has a thickness of 20 mm and a bond length of 1750 mm. That is, the
of concrete were determined based on three cube specimens two ends of the overlay are 25 mm away from the supports.
(150 mm × 150 mm × 150 mm) and three standard cylinders The test parameters included: (1) the main reinforcement ratios
(ϕ150 mm × 300 mm), respectively. The tensile and compressive steel (ρD10
s = 0.79%, ρD12
s = 1.13% and ρD16s = 1.72%), (2) the strengthening
reinforcements were Grade 500B steel bars, while the stirrups were configurations consisting of different matrixes (ECC vs. epoxy mortar)
Grade 250 steel bars as recommended by the construction standard of and different installation methods (prefabricated vs. cast-in-place), and
Hong Kong [51]. For each reinforcement type, three bars were tested to (3) different cross-sectional areas of CFRP grids (FTG-CR5 vs. FTG-
determine the yield and ultimate strengths as well as the elastic mod- CR8). The total 15 beams were divided into five series, each of which
ulus. Table 2 summarizes the mechanical properties of all materials, included three specimens representing three different tensile re-
including ECC, epoxy mortar, concrete, CFRP grid and steel bars used in inforcement ratios. Table 3 shows the details of the five series of beams.
this study. Each beam is labeled using the Y-X format, where X denotes the dia-
meter of tensile steel bars corresponding to a specific reinforcement
ratio (10 for ρD10 s = 0.79%, 12 for ρD12 s = 1.13% and 16 for
2.2. Details of specimens D16
ρs = 1.72%), and Y denotes the series number. It is noteworthy that
epoxy mortar was used in Series III for comparison with the ECC matrix.
A total of 15 simply-supported RC beams with dimensions of
A preliminary trial test on RC beams strengthened with CFRP grid-
150 mm in depth, 200 mm in width and 1800 mm in span length were
reinforced ECC matrix, which was cast-in-place, indicated that the plate
end (PE) debonding failure may occur if no adequate anchorage system
Table 1
Mixing proportion of ECC (kg/m3). is provided at the end of the strengthening layer. To prevent such PE
debonding failure, U-wrap CFRP anchorages (i.e., three layers of
Water Cement Flay ash Silica sand Water-reducing admixture PVA Fiber 0.167 mm CFRP sheets) were used when the cementitious matrix was
cast-in-place ECC or epoxy mortar (Series II and III in Table 3).
330.6 396.8 925.6 396.8 6.0 26
Meanwhile, the substrate concrete was surface saturated with Sika®

11
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 3. Geometry and steel reinforcement details of test specimens (dimensions in mm).

Latex (mixed with cement and water), to enhance its adhesion with the cement matrix; Second, the well-mixed latex was painted on the bottom
cast-in-place ECC. Series IV and V in fact attempted to further explore surface. After the latex almost became dry (around 5 min in the la-
the most efficient solution for suppressing both the PE debonding and boratory environment), a 10-mm thickness of cementitious matrix (i.e.,
the intermediate crack-induced debonding (IC debonding) based on ECC or epoxy mortar) was immediately applied to the primed surface;
Series II and III test results. In these two series (i.e., Series IV and V), the then the CFRP grids were slightly pressed into the first layer of the
CFRP grid-reinforced ECC matrix was prefabricated, while the epoxy matrix, followed by the other 10-mm layer of cementitious matrix
resin (provided by Maeda Kosen Co. Ltd) was used to bond them to the which was applied on the CFRP grids. As for the prefabricated
soffits of RC beams. The epoxy had the tensile strength of about 30 MPa, strengthening system, the construction procedure was simpler: First,
which is much higher than that of ECC to ensure the failure to occur in wood formworks were prepared according to the required dimensions
the layer of ECC. Since the failure of ECC is very ductile due to the of the strengthening layer, and then the CFRP grids and the cementi-
multiple cracking and strain-hardening characteristics, no end ancho- tious matrix were placed into the formwork to form a prefabricated
rage was provided for the RC beams strengthened with prefabricated composite plate. After one day curing, the plate was demolded and
CFRP grid-reinforced ECC matrix (Series IV and V in Table 3). cured for 7 days. During the curing, the plate was sprayed with water to
The installation procedures of the cast-in-place strengthening avoid early shrinkage of the ECC. Then the prefabricated composite
system consisted of the following steps: First, the bottom surface of each plate was bonded by epoxy to the bottom surface of each RC beam,
RC beam was roughened and exposed with coarse aggregates (the which had been roughened by sandblasting and primed one day before.
roughened depth was around 5 mm) and wood formworks were fixed
onto the two sides of each beam to prevent the flowing away of the

Table 3
Details of test specimens.

Series Specimen Tensile steel Strengthening layer Use of U-wrap Concrete substrate treatment
reinforcement anchorage

I Control beams I-10 3Φ10 N/A No N/A

I-12 3Φ12
I-16 2Φ16 + 1Φ12

II Cast-in-place ECC series II-10 3Φ10 ECC + FTG-CR5 Yes Chiseling + Latex
II-12 3Φ12
II-16 2Φ16 + 1Φ12

III Cast-in-place epoxy mortar series III-10 3Φ10 Epoxy mortar + FTG- Yes Chiseling
III-12 3Φ12 CR5
III-16 2Φ16 + 1Φ12

IV Prefabricated ECC series IV-10 3Φ10 ECC + FTG-CR5 No Chiseling + Epoxy primer and bonding
IV-12 3Φ12 adhesive
IV-16 2Φ16 + 1Φ12

V Prefabricated ECC series with high V-10 3Φ10 ECC + FTG-CR8 No Chiseling + Epoxy primer and bonding
stiffness grid V-12 3Φ12 adhesive
V-16 2Φ16 + 1Φ12

Note: Chiseling denotes the concrete surface was pre-treated with a heavy chisel hammer.

12
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 4. Arrangements of strain gauges and LVDTs

(dimensions in mm).

2.3. Instrumentations and testing procedure spreader beam connected with a hydraulic actuator of 500 kN capacity.
Five linear variable differential transformers (LVDTs) were employed to
Each specimen was tested under four-point bending with a clear measure the vertical beam deflections, as illustrated in Fig. 4. Six
span of 1800 mm between two supports. The length of the constant electrical resistance strain gauges with the base length of 100 mm were
moment zone was 600 mm, which was applied by using a stiff steel used to measure the sectional concrete strains (C1 ∼ C6 in Fig. 4a),

13
X. Yang et al. Composite Structures 189 (2018) 9–26

while strain gauges with the base length of 3 mm were used to monitor
the strain responses of the longitudinal and transverse steel bars
(S1 ∼ S44 in Fig. 4b) and the CFRP grid (A1 ∼ A17 and B1 ∼ B3 in
longitudinal and transverse directions, respectively, as shown in
Fig. 4c). Data from the strain gauges, LVDTs and the load cell were
recorded using an 80-channel data acquisition system at a frequency of
1 Hz.

3. Experimental results and discussion

The flexural performance of the control beams and four series of

strengthened beams in terms of the crack pattern and failure mode,
load-deflection curves, strain profiles of steel bars and CFRP grids were
obtained and are herein discussed.

3.1. Crack patterns and failure modes

3.1.1. Control beams (Series I)

The control beams failed in a typical mode which is concrete
crushing after the yielding of the longitudinal steel reinforcement. The
vertical flexural crack first initiated within the constant moment region
at a load level of around 5–15 kN. As the load increased, new cracks
developed outside the constant moment region until the tensile steel bar
yielded. As illustrated in Fig. 5a, the crack numbers were almost the
same for these control beams with different reinforcement ratios while
a low reinforcement ratio led to more ductile behavior. Shear diagonal
cracks were not observed in this series of bending tests.

3.1.2. Beams strengthened with cast-in-place CFRP grid-reinforced ECC

matrix (Series II)
Fig. 5b and Fig. 7 illustrate the crack patterns and failure modes of
RC beams strengthened with cast-in-situ CFRP grid-reinforced ECC
matrix. The first flexural crack was initiated at a load ranging from
20 kN to 35 kN. The wide cracks formed within the substrate concrete
transformed to multiple diffused fine cracks in the cast-in-place ECC
layer. The crack numbers in this series of beams were more than those
observed in series III beams. In specimens II-10 and II-12, the failure
mode was characterized by yielding of the tensile steel followed by a
certain level of IC debonding that initiated from the mid-span (either a
flexural crack or a flexural-shear crack and then propagated towards
the U-wrap anchorage (see Fig. 7). But eventually CFRP grids were
ruptured near the U-wrap anchorage, possibly due to the stain con-
centration occurred at there. For specimen II-16, the failure occurred
due to concrete crushing followed by the rupture of CFRP grids without
any debonding. The relatively easy debonding in cases of lower re-
inforcement ratios (i.e., II-10 and II-12) might be due to the larger
deformation of the beams. To further understand the early debonding
failure of II-10 and II-12, Fig. 6 compares the strain distributions of
CFRP grids at different load levels (i.e., 0.1Fu , 0.5Fu and 0.9Fu ) for Series
II beams. It is clearly seen that near the peak load the CFRP grid strains
of II-10 and II-12 are larger than those of II-16. Here it should be noted
that the rupture means the local rupture of a single CFRP grid instead of
that of the whole beam width. Overall, it seems that the use of cast-in-
place ECC still could not fully suppress the IC debonding probably be-
cause the bond between cast-in-place ECC and concrete substrate was
not strong enough to ensure the failure to occur inside the ECC matrix
to release the local stress concentration. In addition, the U-shape an-
chorage was also needed for preventing the PE debonding failure.

3.1.3. Beams strengthened with cast-in-place CFRP grid-reinforced epoxy Fig. 5. Concrete crack patterns of specimens.
mortar matrix (Series III)
The crack patterns and failure modes are shown in Fig. 5c and
characterized by the local IC debonding of the external strengthening
Fig. 7. Initial cracks occurred in the epoxy mortar at a load level of
layer (see Fig. 5). All three specimens failed due to rupture of CFRP
30–42 kN. The cracking load was much higher than those of the pre-
grids at the ultimate states. Partial separation of the external
vious two series as the epoxy mortar had a very high tensile strength
strengthening layer was observed between the epoxy mortar and the
(14 MPa). The failure mode of these strengthened beams was

14
X. Yang et al. Composite Structures 189 (2018) 9–26

substrate concrete in specimens III-10 and III-12, whereas similar to

specimen II-16 in Series II, specimen III-16 did not exhibit any de-
bonding. For all the three specimens, the end anchorage system re-
mained intact during the whole loading process, and thus was efficient
to prevent the PE debonding of the external strengthening layer. In
addition, the cracks in the RC beams could penetrate through the
strengthening layer and the crack deformation was absorbed by slip out
of the CFRP grids from the epoxy mortar matrix. This is the most im-
portant difference between externally bonded FRP laminates and ce-
ment-bonded FRP composites as the strengthening layer. Compared to
Series II, the local debonding around the mid-span area in Series III was
reduced probably because of the relatively stronger bond achieved
between the epoxy mortar and the concrete substrate.

3.1.4. Beams strengthened with prefabricated CFRP grid-reinforced ECC

Fig. 6. Comparisons of CFRP grid strains for Series II beams. matrix (Series IV)
The crack patterns and failure modes are depicted in Fig. 5d and

Fig. 7. Failure mode of specimens.

15
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 7. The cracking loads of this series of beams increased to a load about 3% with a relatively low elastic modulus, the stiffness of each
level of 15–30 kN. Failure of the strengthened beams was induced by beam strengthened with the ECC matrix (i.e., Series II) in the pre-
the formation of wide flexural cracks within the constant moment re- cracking stage was smaller (as seen in Fig. 8a-c); however, the stiffness
gion, accompanied by rupture of CFRP grid at the vicinity of the major in the post-cracking stage decreased more gently due to the ductile
crack in cases of specimens IV-10 and IV-12. For specimen IV-16, the behavior of ECC. Additionally, the ECC could still provide a portion of
beam failed due to crushing of concrete followed by FRP rupture. The load-carrying capacity during the post-cracking stage owing to its
wide cracks in the substrate concrete diffused into multiple fine cracks strain-hardening effect, which subsequently delayed the steel yielding
in the ECC matrix. Also, no debonding failure of the external in the beam strengthened with the ECC matrix. In addition, it is shown
strengthening layer was observed during the whole loading process, that the beams strengthened through the ECC matrix always exhibited
indicating that the local stress concentration between the prefabricated better ultimate flexural deformation than their epoxy mortar counter-
overlay and the substrate concrete was successfully suppressed. In parts, leading to a more ductile failure (see Series II and III in Fig. 8a-c).
specimen IV-10, the major crack locations in the RC beam and the The test results confirmed that ECC is a promising cementitious matrix
strengthening layer had slightly staggered (Fig. 7) probably due to the which can be used together with the CFRP grid for practical strength-
occurrence of a very local slip (e.g. due to the bond imperfection) be- ening applications.
tween the strengthening layer and the substrate concrete. Overall, the
combined use of prefabricated CFRP grid-reinforced ECC matrix and 3.2.2. Effect of installation method
epoxy bonding seemed to be the most efficient solution in achieving the The different installation methods of the external strengthening
composite action of the strengthening system and avoiding the de- layers had little effect on the overall load-deflection responses before
bonding failure. the yielding of the tensile steel bars. However, series IV beams
strengthened with the prefabricated overlays had relatively higher peak
3.1.5. Beams strengthened with prefabricated high stiffness CFRP grid- loads compared to their counterparts strengthened with the cast-in-
reinforced ECC matrix (Series V) place overlays (i.e., series II beams), as shown in Fig. 8d. Meanwhile,
Series V and Series IV were basically the same unless the stiffness of series IV beams had larger yield loads than series II beams, as reported
CFRP grids in two series were different. The crack patterns and failure in Table 4. Table 4 indicates that series IV beams achieved ductile be-
modes of Series V specimens are shown in Fig. 5e and Fig. 7. The havior with the ultimate deflections comparable to series II and III
cracking loads of this series beams ranged from 30 kN to 36 kN. These beams, except for the specimen IV-12. For this beam, the achieved ul-
beams showed an increased number of cracks at their peak loads timate deflection (i.e., 42.4 mm) was slightly smaller than that of II-12
compared to the preceding four series of beams due to the higher (i.e., 48.6 mm). Overall, the excellent strength and deformation prop-
stiffness of CFRP grids used in the prefabricated strengthening layer. At erties achieved in series IV beams demonstrate that the strengthening
the final stage of the loading process, a critical shear diagonal crack was method by bonding the prefabricated CFRP grid-reinforced ECC plate to
formed in the shear span region in all three specimens. It was also the RC beam soffit through epoxy is more efficient and convenient
confirmed that the yielding of a stirrup occurred. All the specimens compared to the cast-in-place strengthening method, and in particular,
failed due to the critical diagonal crack-induced debonding (i.e., CDC it eliminates the need of using U-anchorage at the ends of the external
debonding) along the interface between the strengthening layer and the overlay. One thing to be noted is that the high flexural strengthening
substrate concrete. A further check was carried out after the tests and efficiency may trigger the CDC debonding failure at high shear load
revealed that the peak loads reached around 80–90% of the theoretical levels, as observed in series V beams. Special attention may be needed
shear resistance of the RC beam; the latter was calculated based on the to ensure that the strengthened RC beam has a sufficient shear strength
Chinese design code for concrete structures using the actually measured resistance to achieve higher flexural strength enhancement.
strength properties of steel reinforcement and concrete. Therefore, the
above CDC debonding failure should be triggered by the significant 3.2.3. Effect of CFRP grid stiffness
flexural strength enhancement by the high-stiffness CFRP grids, so that Table 4 shows that series V beams obtained higher yield and peak
the peak loads approached the shear resistance of the original RC beam. loads compared to their counterparts (i.e., series IV beams) due to the
It is obvious that the use of prefabricated CFRP grid-reinforced ECC use of high stiffness CFRP grids, despite the relatively brittle failure of
cannot suppress the CDC debonding, which should be solved through these beams observed during the tests. More specifically, the use of high
the shear strengthening. stiffness CFRP grids resulted in the yield load increases equal to
102.9%, 71.6% and 38.4% for specimens V-10, V-12 and V-16 (i.e.,
3.2. Load-deflection behavior compared to the reference beams as reported in Table 4), which were
significantly (68.3%, 52.6% and 31.4%) higher than those achieved by
The load–deflection curves of all specimens are shown in Fig. 8, their counterparts in series V (i.e., specimens IV-10, IV-12 and IV-16).
while the main test results are summarized in Table 4. In this table, the Similarly, the increases in peak loads were 116.8%, 76.6% and 43.4%
values of the cracking load (Fc ), yield load (Fy ) and yield deflection (Δy ), for specimens V-10, V-12 and V-16 respectively, 85.0%, 65.9% and
peak load (Fu ) and ultimate deflection (Δu ), and the increases in these 39.4% higher than those achieved for specimens IV-10, IV-12 and IV-
loads with the reference to the corresponding control beams are all 16, respectively. Due to the high strengthening ratio and the premature
reported in detail. The specific influences of the varied parameters are CDC debonding failure, all beams in series V had much less ultimate
discussed sequentially in the following sections. deflections compared with other series of beams.

3.2.1. Effect of matrix type 3.2.4. Effect of main reinforcement ratio

Each beam strengthened with the epoxy mortar matrix (i.e., Series Fig. 9 depicts the variations of the peak loads with the main re-
III) had a relatively larger stiffness at the initial pre-cracking stage when inforcement ratio. As expected, it is seen that the peak load increased
compared to its counterparts in Series II and IV, as indicated in Fig. 8e. almost linearly with the reinforcement ratio. Fig. 10 further illustrates
Cracking of the epoxy mortar matrix, however, occurred at a very small the effect of the main reinforcement ratio on the percent increases in
deflection as the mortar had a very limited elongation (i.e., 0.11%). the yield and peak loads of the strengthened RC beams compared to the
After wide flexural cracks formed within the external strengthening corresponding control beams. Obviously the strength increase in per-
layer, the epoxy mortar matrix almost lost its strength contribution, centage was more significant in case of a low steel reinforcement ratio,
resulting in earlier steel yielding compared to Series II and IV beams. and it declined with the increase of reinforcement ratio in an ap-
On the other hand, as the ECC matrix had a maximum elongation of proximately linear manner. It is also noteworthy that the scatter of the

16
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 8. Comparisons of load–deflection curves.

Table 4
Summary of test results.

No. Fc (kN) Increase in Fc (%) Fy (kN) Increase in Fy (%) Δy (mm) Fu (kN) Increase in Fu (%) Δu (mm) Failure mode

I-10 8 – 44.1 – 13.6 51.9 – 47.1 SY → CC

II-10 20 150.0 68.7 55.8 16.9 95.5 84.0 44.3 SY → IC → TR
III-10 30 275.0 60.3 36.7 13.5 89.3 72.1 37.0 SY → IC → TR
IV-10 15 87.5 74.2 68.3 18.1 96 85.0 45.2 SY → TR
V-10 36 350.0 89.5 102.9 13.9 112.5 116.8 31.4 SY → CDC
I-12 12 – 58.9 – 15.3 66.6 – 42.6 SY → CC
II-12 25 108.3 83.3 41.4 18.0 104.8 57.4 48.6 SY → IC → TR
III-12 40 233.3 88.6 50.4 17.5 105.5 58.4 40.5 SY → IC → TR
IV-12 20 66.7 89.9 52.6 18.7 110.5 65.9 42.4 SY → TR
V-12 36 200.0 101.1 71.6 16.7 117.6 76.6 28.7 SY → CDC
I-16 8 – 90.2 – 17.2 93.6 – 34.7 SY → CC
II-16 35 337.5 117.3 30.0 20.1 126.8 35.5 37.7 SY → CC
III-16 42 425.0 114.3 26.7 20.3 129.3 38.1 35.0 SY → IC → CC
IV-16 30 275.0 118.5 31.4 22.3 130.5 39.4 44.9 SY → CC
V-16 30 275.0 124.8 38.4 21.0 134.2 43.4 25.8 SY → CDC

Notes: 1. Fc – cracking load, Fy – yield load, Δy – yield deflection, Fu – peak load, Δu – ultimate deflection.
2. CC–concrete crushing, CDC–critical diagonal crack-induced debonding, IC–intermediate crack-induced debonding, TR–tensile rupture of CFRP grid, SY–steel yielding.

17
X. Yang et al. Composite Structures 189 (2018) 9–26

axis represents the midspan location. For the other two reinforcement
ratios (i.e., 0.79% and 1.72%) the corresponding strain distributions are
also similar and thus omitted here. In specimen I-12 (i.e., the reference
beam), the strain distributions of longitudinal steel reinforcement in
each shear span was approximately linear in spite of some zigzags due
to the cracking. In other specimens, the strain distributions at the ul-
timate state could be roughly divided into three parts: a flat zone in the
mid part and two linear zones near the two ends of the strengthening
layers. The significant strain increase in the steel reinforcement (see
specimens II-12 and III-12 in Fig. 12b and c) at the ultimate state at a
specific location was considered to be caused by the overall debonding
of the strengthened layers or the fracture failure of CFRP grids, which
led to a sudden loss of the beam stiffness and the widening of a major
flexural or flexural/shear crack. In Fig. 12c a sudden increase of strain
distributions within a certain span length in the case of the epoxy
mortar used as the matrix was also seen when the applied load was
Fig. 9. Comparison of peak load.
increased from 0.4Fu to 0.5Fu , implying the abrupt occurrence of local
interface debonding within this length. In other cases, when the ECC
increases in the yield and peak loads was larger in cases of a lower matrix was used, such a sudden shift of strain distribution was not
reinforcement ratio while became smaller with the increase of re- observed. The length of the steel yielding range could be an indicator of
inforcement ratio. This is because that the strength increase depended the debonded length because once the steel yielded, the moment in-
on the bond force transfer between the strengthening layer and the crease would be mainly contributed by the externally strengthening
substrate concrete. The bond performance, and consequently, the layer, thus increasing the strain in FRP in the strengthening layer and
strength contribution of the strengthening layer to the RC beam, usually triggering debonding [52]. It is seen in specimens IV-12 and V-12, such
has a large variation. Such variation has a more significant influence on lengths were relatively shorter, indicating the composite action be-
an RC beam with a lower original strength if the strength contribution is tween the strengthening layer and the substrate RC beam was overall
a constant. excellent before the ultimate state was reached when the prefabricated
Fig. 11 depicts the effect of the strengthening ratio of FRP grids (i.e., CFRP grid-reinforced ECC matrix was used.
represented by Efρf/Esρs, the tensile stiffness ratio of FRP grid to re- Fig. 13a to c present the strain distributions of the CFRP grids in the
inforcement) on the increases (in percentage) of yield and peak loads. strengthened layers. Here the test beams with a moderate reinforce-
As series II, III and IV beams had the same strengthening ratio, a tread ment ratio of 1.13% were still used as examples. Obviously, the strains
line is provided in the figure to represent the average values of three measured for the CFRP grids were much larger than those in the tensile
series. It can be seen from Fig. 11 that both yield load and peak load steel reinforcement due to their locations further away from the neutral
increase almost linearly with the strengthening ratio. The trend for axis. It was also shown that the strain distributions of CFRP grids in the
series V is most clear because of the larger strengthening ratio. In spite matrix ECC (specimens II-12, IV-12 and V-12) were more smooth
of the same strengthening ratio, series IV beams performed slightly compared to those in specimens III-12, for which the epoxy mortar was
better than series II and III beams, demonstrating the effectiveness of used. This phenomenon suggests that the ECC matrix helped a more
the prefabricated FRP grid-reinforced ECC system. uniform shear stress transfer at the interface between the strengthening
layer and the substrate RC beam. The high strains in CFRP grids dis-
3.3. Strain measurements tributed within a wider range in specimen III-12 indicate that the in-
terfacial debonding occurred in a wider range, which coincided well
The strain distributions in longitudinal steel reinforcement and with the above-discussed strain distributions in longitudinal steel re-
CFRP grids can provide a more insightful investigation into the me- inforcement. Another phenomenon observed was that the maximum
chanisms of interaction among CFRP grids, cementitious matrix (epoxy strain in CFRP grids almost reached their rupture strain (i.e., 1.4%)
or ECC), and the substrate RC beam, which could not be revealed from except in specimen V-12, for which the strain increase in CFRP grids
the global load-deflection behavior and failure at the ultimate state. was interrupted by the CDC shear failure.
Taking the moderate reinforcement ratio (i.e., 1.13%) as an ex- The applied load versus the strain responses of the CFRP grid and
ample, Fig. 12a to e present the longitudinal strain distributions of the the tensile steel bar at the midspan section for all tested beams are
tensile steel bars in the test beams. In these figures, the origin of the X- shown in Fig. 14a to c, which represent three different reinforcement

Fig. 10. Effect of main reinforcement ratio on the percent increases in yield and peak loads.

18
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 11. Effect of Efρf/Esρs on the percent increases in yield and peak loads.

ratios, respectively. It should be noted that the yielding strain of the the mid-span location where the major flexural cracks might not occurr.
steel bar was about 2250 με and the ultimate tensile strain of the CFRP Fig. 14a to c show that strains in all the steel bars and CFRP grids
grid was 14,000 με. However, the maximum strains achieved in steel exhibited three stages, of which the two turning points corresponded to
and CFRP grids at the ultimate state might not be exactly captured at the cracking load and the yield load. It is seen that the steel bar strains

Fig. 12. Longitudinal strain distributions of tensile steel bars at different loading levels.

19
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 13. Longitudinal strain distributions of CFRP grid at different loading levels.

in all the strengthened beams were effectively reduced compared to area of CFRP grid). The tensile force of the external strengthening layer
those in the reference beams. It is generally seen that strain develop- is applied at the centroid of CFRP grid. However, when the epoxy
ments in both the steel rebars and the CFRP grids in series III were mortar is used as the matrix, its tensile contribution is ignored because
unstable (i.e., many sudden jumps/variations with the load increase) of its linearly brittle nature.
compared to those in other test series, confirming again the advantages It should be noted that the tensile steel bars should be yielded once
of using ECC as the matrix of the strengthening layer. Unfortunately, it the tensile rupture of the FRP grid occurs, as the rupture strain of the
is not possible to quantitatively compare the efficiency of different FRP grid is much larger than the yielding strain of the steel bars.
strengthening configurations using the load versus maximum FRP/Steel Therefore, only two failure modes including the rupture of the FRP
strain relationships, mainly because the random locations of flexural grids and the crushing of the compression concrete are considered for
cracks significantly influenced the measured strains in steel rebars and determining the flexural capacity. According to Fig. 16, the depths of
CFRP grids. the compression zone and the equivalent rectangular stress block are
To further investigate the strengthening efficiency of the pre- defined as x 0 and x f , respectively. The parameter ξf is defined as the
fabricated CFRP grid-reinforced ECC matrix, Fig. 15 shows the strain ratio of x f /hf , and it can be further determined by the following
distributions over the beam depth at the mid-span location of beam IV- equation:
10. It is seen that the strain measurements were almost linearly dis- xf x0
tributed along the beam height. Furthermore, there was no sign of in- ξf = = β1·
hf hf (1)
terfacial slip occurring at the prefabricated plate-to-concrete interface,
demonstrating that satisfactory bond performance was maintained where β1 is defined as the ratio of x 0 / x f ; hf is defined as the distance
throughout the loading process. The other strengthened beams had si- from the centroid of CFRP grid to the top edge of concrete compression
milar strain distributions and therefore the discussion is omitted here. zone.
Base on the sectional strain compatibility and the relevant para-
4. Flexural capacity analysis meters defined in Fig. 16, the following equations can be derived:

εs′ x 0−ds′ x f −β1 ds′

To predict the flexural capacity of the RC beam strengthened with = =
εf hf −x 0 β1·hf −x f (2)
CFRP grid-reinforced ECC or epoxy mortar matrix, a theoretical ana-
lysis was performed based on the following assumptions: (1) plane εc x0 xf
cross-sections remain plane for the strengthened beam during the = =
εs; x 0−ds′ x f −β1·ds′ (3)
loading process; (2) the tensile strength of concrete is ignored; (3) the
maximum compression strain of concrete is taken as 3500 με according εs h −x β ·h 0−x f
to the Hong Kong design code [53]; (4) the tensile force of the external = 0 0 = 1
εc x0 xf (4)
strengthening layer is determined by combining the tensile forces
provided by both the CFRP grid and the ECC matrix. For the ECC ma- where εs′ is defined as the strain of compressive steel bar; εf is defined as
trix, the tensile force is calculated as the sum of the tensile strength of the strain of CFRP grid; εc is defined as the strain of top edge com-
ECC (i.e. fe = 3.82 MPa) multiplying the net section area (by subtracting pressive concrete; εs is defined as the strain of tensile steel bar; ds′ is
total sectional area of the strengthening layer with the cross sectional defined as the distance from the centroid of the compressive steel bar to

20
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 14. Load-strain responses of CFRP grid and tensile steel bars at the midspan section.

t
the top edge of concrete compression zone; hf = h + 20 where h is the
height of the original RC beam and t0 is the thickness of the externally
strengthening layer; h 0 is defined as the distance from the centroid of
the tensile steel bar to the top edge of concrete compression zone.
For the balanced failure (i.e., both the FRP grid rupture and the
concrete crushing occur at the same time), the compressive strain in the
concrete, εc , and the tensile strain in the FRP grid, εf , reach the ultimate
compression strain, εcu , and the rupture strain, εfu , at the same time. εcu
and εfu can be determined as follows:
εcu = 0.0035 (5)

f fu
εfu =
Ef (6)
where f fu and Ef are the tensile strength and the elastic modulus of the
FRP grids. Substituting Eqs. (5) and (6) into Eq. (1), the parameter ξf
corresponding to the balanced failure (i.e., ξbf in the following equa-
Fig. 15. Strain distributions over the beam depth in specimen IV-10. tion) can be obtained:
εcu β1
ξbf = β1· = f fu
εcu + εfu 1 + E ·ε
f cu (7)

21
X. Yang et al. Composite Structures 189 (2018) 9–26

Fig. 16. Stress and strain distributions of the strengthened RC beam.

Then, the compression zone depth, xbf , for the balanced failure can overestimated mainly because of the occurrence of CDC debonding,
be calculated using the following equation: which was due to insufficient shear resistance and could not be ap-
propriately considered in the present analysis. If series V specimens are
xbf = ξbf ·hf (8) excluded, the average ratio of the test flexural capacities to the pre-
When the FRP rupture and the yielding of the compressive steel bars dicted values of series I, II, III and IV specimens is 1.05 with the COV
occur at the same time, the balanced value of the compressive concrete (coefficient of variation) of 0.04, indicating that the accuracy of the
depth (i.e., xlfr ) can be derived from Eq. (2) as follows: model prediction is acceptable. The underestimation might be due to
the difference in the prediction of failure modes (Table 5). In the ana-
β1 (εs′·hf + εfu·ds′ )
xlfr = lysis, the failure was usually caused by concrete crushing if the rupture
εs′ + εfu (9) strain 14,000 με was assumed while in the experiment, rupture of CFRP
f y′ grids was usually observed before concrete crushing. The other possible
where εs′ = E ; f y′ is defined as the yield stress of the compressive steel factor was the shear deformation of the RC beam after the occurrence of
s
bar; Es is defined as the elastic modules of the steel bar. diagonal cracking of concrete, which introduced a dowel action on the
Correspondingly, xlfc is defined as the corresponding value of the external strengthening layer, thus reducing its tensile rupture strain in
compression concrete depth when the concrete crushing and the CFRP grids. If the shear deformation was very large, the CDC debonding
yielding of the compressive steel bar occur at the same time. It can be occurred even without the rupture of FRP grids due to the mixed-mode
derived from Eq. (3) as follows: failure of the interface between the strengthening layer and the sub-
β1·εcu·ds′ strate concrete. This mixed-mode failure mechanism may be much
xlfc = more significant in case of cement-bonded FRP strengthening system
εcu−εs′ (10)
compared to epoxy-based externally bonded FRP laminate system be-
When the original concrete beam is over-reinforced, the concrete cause the former has a much larger bending stiffness than the latter
crushing may be the failure mode without the yielding of tensile re- [54].
inforcement and thus FRP grid rupture. Under this situation, x uf is The effective tensile strain of CFRP grid at the ultimate state was
defined as the compression concrete depth corresponding to the oc- back calculated from the above-mentioned sectional analysis for each
currence of concrete crushing, and its value can be derived from Eq. (4): strengthened beam. The ratio of effective tensile strain of CFRP grid to
β1·h 0 its nominal rupture value was found to vary from 0.47 to 0.96 in the
x uf = εy present study (Table 5). It is seen in the table that this ratio generally
1+ (11)
εcu decreases with the load-carrying capacity achieved in the strengthened
where εy defined as the yield strain of the steel bar. beams regardless of the failure mode, i.e., the higher the load-carrying
It is clear that x uf is the same as the compressive concrete depth capacity, the smaller the strain efficiency. To further clarify this ten-
defined in the ordinary RC beam without the strengthening overlay dency, Fig. 17 plots the relationship between the effective tensile strain
when the balanced failure occurs. Therefore, it can be calculated as: of CFRP grids at the ultimate state and the load-carrying capacity of the
beam. The load-carrying capacity was normalized by the corresponding
x uf = ξb·h 0 (12) shear capacity of the original RC beam. It is seen that the effective
strain of CFRP grids declines linearly with the increase of flexural to
where the parameter ξb can be determined as follows:
shear strength ratio. An extreme condition is that, when the flexural
εcu β1 strength approached the shear strength of the beam (i.e., the ratio is
ξb = β1· = fy
εcu + εy 1 + E ·ε around 0.9), the CDC debonding occurred and the effective strain ratio
(13)
s cu
of CFRP girds was only around 0.5. When the flexural to shear strength
where f y defined as the yield stress of the steel bar. ratio is 0.6, the effective strain of CFRP grids approached its nominal
Based on the determinations of the values xbf, xlfr , xlfc and x uf , the value. Therefore, it is preliminarily recommended to limit the flexural
failure mode of the strengthened beam can be determined and then the to shear strength ratio below 0.6 to ensure a full utilization of FRP grid
flexural capacity can be computed. The corresponding analytical for- strength for flexural strengthening.
mulae for the flexural capacities of the test beams are provided in the
Appendix. 5. Conclusions
Table 5 compares the bending test results and analytical predictions
for all specimens. Based on the Hongkong Code [53], α1 = 0.67 and In this paper, a total of three control beams and twelve strengthened
β1 = 0.8 for the case that concrete has a cube strength fcu = 52.1 MPa beams were prepared and tested under four-point bending up to failure.
(Fig. 16). For series V specimens, all the flexural capacity was Two types of matrix materials (ECC vs. epoxy mortar), different

22
X. Yang et al. Composite Structures 189 (2018) 9–26

Table 5
Comparisons of experimental and predicted flexural capacity.

Specimen Fu,e (kN) Fu,p (kN) Fu,e CFRP grid strain εf at ultimate state Strain efficiency of CFRP grid Predicted failure mode Experimental failure sequence
Fu,p

I-10 51.9 48.0 1.08 – SY-CC SY → CC

I-12 66.6 65.2 1.02 – SY-CC SY → CC
I-16 93.6 91.9 1.02 – SY-CC SY → CC
II-10 95.5 87.6 1.09 12,780 0.91 SY-CC SY → IC → TR
II-12 104.8 104.1 1.01 10,536 0.75 SY-CC SY → IC → TR
II-16 126.8 129.5 0.98 7513 0.54 SY-CC SY → CC
III-10 89.3 81.0 1.10 13,440 0.96 SY-CC SY → IC → TR
III-12 105.5 97.6 1.08 11,099 0.79 SY-CC SY → IC → TR
III-16 129.3 123.3 1.05 7928 0.57 SY-CC SY → IC → CC
IV-10 96 87.6 1.10 12,780 0.91 SY-CC SY → TR
IV-12 110.5 104.1 1.06 10,536 0.75 SY-CC SY → TR
IV-16 130.5 129.5 1.01 7513 0.54 SY-CC SY → CC
V-10 112.5 119.8* 0.94* 10,565 0.75 – SY → CDC
V-12 117.6 135.9* 0.87* 8934 0.64 – SY → CDC
V-16 134.2 160.7* 0.84* 6629 0.47 – SY → CDC
Average 1.05
COV 0.04

Note: 1. CC–concrete crushing, CDC–critical diagonal crack-induced debonding, IC–Intermediate crack-induced debonding, TR–Tensile rupture of the CFRP grid, SY–Steel yielding.
2. SD–standard deviation, COV–coefficient of variation.
* means that the values are not included in the calculation of the average and COV.

(c) The combined use of prefabricated ECC and epoxy bonding led to
an optimal solution in terms of debonding control. Both the IC
debonding and the plate end debonding could be successfully in-
hibited with no need of additional anchorage countermeasures.
(d) The use of the higher stiffness CFRP grid led to more increase in the
flexural capacity of the strengthened beam. However, the large
strength enhancement may trigger the CDC debonding failure.
Special design checks should be provided before the flexural
strengthening to ensure that the strengthened beam has sufficient
shear strength resistance.
(e) The design rupture strain in CFRP grids at the ultimate state of the
strengthened beams may be lower than the nominal value of a
single CFRP grid that is obtained from the coupon tests. The ratio of
the former to the latter is influenced by the flexural to shear
strength ratio of the strengthened beam. If such a ratio is controlled
below 0.6, the tensile strength of CFRP grid can be fully utilized.
Fig. 17. Influence of flexural/shear strength ratio on the strain efficiency of CFRP grid.
Otherwise, the strengthening contribution by CFRP grids may be
compromised due to the dowel action exerted on the external
installation methods of the overlay strengthening systems, two different strengthening layer, which is caused by the occurrence of diagonal
levels of the CFRP grid stiffness, and three tensile steel reinforcement cracks in concrete.
ratios served as the test variables. The test results were carefully ob- (f) In spite of the existence of a certain level of local IC debonding, the
served and compared in terms of the concrete crack patterns, failure flexural capacity of RC beams strengthened with CFRP grid-re-
modes, the load-deflection responses and the strain measurements in inforced cast–in-place epoxy mortar or ECC matrix and epoxy
concrete, tensile steel bars and CFRP grids. Base on the experimental bonded prefabricated CFRP grid-strengthened ECC matrix can be
and analytical results reported in this study, the following conclusions predicted with acceptable accuracy using the sectional analyses
can be drawn: presented in the paper. The average predicted values to the ex-
perimental ones was 1.05 with the coefficient of variation of 0.04.
(a) The use of ECC or epoxy mortar alone as the matrix and bonding
adhesives of CFRP grids to strengthen RC beams needed the pro-
vision of reliable end anchorage to prevent plate end debonding Acknowledgments
failure.
(b) The use of ECC alone could not inhibit the IC debonding completely The authors are grateful for the financial support provided by the
because the weak link might still exist at the ECC-to-substrate National Key Research Program of China (Grant No: 2017YFC0703002)
concrete interface. Compared to epoxy mortar, however, use of ECC National Natural Science Foundation of China (Programs Nos.
as the matrix and bonding adhesive of CFRP grids led to more 51478406, 51408521 and 51278441) and the Research Grants Council
ductile failure due to the multiple cracking behavior in the ECC of the Hong Kong SAR (Project Code: PolyU 5145/13E) for providing
matrix, which dispersed more uniformly the shear stress transfer at the funds for their research work. The first author would also ac-
the interface between the strengthening layer and the substrate knowledge the joint PhD studentship provided by The Hong Kong
concrete. Polytechnic University through the project 4-BCAX.

Appendix: Flexural capacity analysis

Condition 1: 0 ⩽ x f < xbf

23
X. Yang et al. Composite Structures 189 (2018) 9–26

In this condition, the rupture of the FRP grid occurs before concrete crushing whereas the compression steel bars do not yield, and the force
equilibrium equation can thus be obtained as follows:
α1·fc ·x f ·b + As′ ·σs′ = As ·f y + Af ·f fu + AE ·fe (14)
where
σs′ = εs′·Es (15)
The strain of the compression steel bars (i.e., εs′) can be derived from Eq. (2):
x f −β1·ds′
εs′ = ·εfu
β1·hf −x f (16)
Substituting Eqs. (15) and (16) into Eq. (14), x f can be deduced:

A− A2 −4α1·β1·fc ·b·[hf ·(As ·f y + Af ·f fu + AE ·fe ) + As′ ·Es ·εfu·ds′]

xf =
2·b·fc ·α1 (17)
where the parameter A is defined as:
A = As ·f y + Af ·f fu + AE ·fe + α1·β1·fc ·hf ·b + As′ ·Es ·εfu (18)
The flexural capacity, Mu , can be determined as follows:
xf
Mu = α1·fc ·x f ·b·⎛ds′− ⎞ + As ·f y ·(h 0−ds′) + (Af ·f fu + AE ·fe )·(hf −ds′)
⎝ 2⎠ (19)
Condition 2: xbf < x f < xlfc
In this condition, the concrete crushing occurs before CFRP grid rupture whereas the compressive steel bars do not yield, and the force equi-
librium equations can thus be obtained as follows:
α1·fc ·x f ·b + As′ ·σs′ = As ·f y + Af ·σf + AE ·fe (20)
σf = εf ·Ef (21)

εf hf −x 0 β1·hf −x f
= =
εcu x0 xf (22)

σs′ = εs′·Es (23)

εs′ x 0−ds′ x f −β1·ds′

= =
εcu x0 xf (24)
x f can be deduced by substituting Eqs. (21)–(24) into Eq. (20):

−A + A2 + 4·α1·β1·fc ·b·εcu (ds′·As′ ·Es + hf ·Af ·Ef )

xf =
2α1·fc ·b (25)
where the parameter A is defined as:
A= As′ ·Es ·εcu−As ·f y + Af ·Ef ·εcu−AE ·fe (26)
Consequently the flexural capacity, Mu , can be obtained as follows:
xf
Mu = α1·fc ·x f ·b·⎛hf − ⎞ + As′ ·σs′·(hf −ds′)−As ·f y ·(hf −h 0)
⎝ 2⎠ (27)
Condition 3: xlfc < x f < x uf
In this condition, the concrete crushing whereas the compression steel bars yield before CFRP grid rupture, and the force equilibrium equations
can thus be obtained as follows:
α1·fc ·x f ·b + As′ ·f y′ = As ·f y + Af ·σf + AE ·fe (28)
σf = εf ·Ef (29)

εf hf −x 0 β1·hf −x f
= =
εcu x0 xf (30)
Substituting Eqs. (24) and (25) into Eq. (23), x f can be deduced as follows:

−A + A2 + 4·α1·β1·fc ·hf ·b·Af ·Ef ·εcu

xf =
2·α1·fc ·b (31)
where the parameter A is defined as:
A = As′ ·f y′−As ·f y + Af ·Ef ·εcu−AE ·fe (32)

24
X. Yang et al. Composite Structures 189 (2018) 9–26

The flexural capacity, Mu , can be determined as follows:

xf
Mu = α1·fc ·x f ·b·⎛hf − ⎞ + As′ ·f y′ ·(hf −ds′)−As ·f y ·(hf −h 0)
⎝ 2⎠ (33)
Condition 4: x uf ⩽ x f
In this condition, the concrete crushing occurs whereas the tensile steel bars do not yield before CFRP grid rupture, and the force equilibrium
equations can thus be obtained as follows:
α1·fc ·x f ·b + As′ ·f y′ = As ·σs + Af ·σf + AE ·fe (34)
σf = εf ·Ef (35)

εf hf −x 0 β1·hf −x f
= =
εcu x0 xf (36)
σs = εs·Es (37)

εs h −x β ·h 0−x f
= 0 0 = 1
εcu x0 xf (38)
Substituting Eqs. (35)–(37) into Eq. (34), x f can be deduced as follows:

−A + A2 + 4α1·β1·fc ·b·εcu·(hf ·Af ·Ef + h 0 ·As ·Es )

xf =
2α1·fc ·b (39)
where the parameter A is defined as:
A = As′ ·f y′ + As ·Es ·εcu + Af ·Ef ·εcu−AE ·fe (40)
The flexural capacity, Mu , can be calculated as follows:
xf
Mu = α1·fc ·x f ·b·⎛hf − ⎞ + As′ ·f y′ ·(hf −ds′)−As ·σs·(hf −h 0)
⎝ 2⎠ (41)

References [17] Wu HC, Sun P. Fiber reinforced cement based composite sheets for structural ret-
rofit. In: Chen JF, Teng JG, editors. Proceedings of the international symposium on
bond behaviour of FRP in structures (BBFS 2005), Hong Kong, China, 2005, p.
[1] Nanni A. A new tool for concrete and masonry repair. Concr Int 2012;34(4):43–9. 343–348.
[2] Bisby LA, Roy EC, Ward M, Stratford TJ. Fibre reinforced cementitious matrix [18] Hashemi S, Al-Mahaidi R. Investigation of bond strength and flexural behaviour of
systems for fire-safe flexural strengthening of concrete: pilot testing at ambient FRP-strengthened reinforced concrete beams using cement-based adhesives. Aust J
temperature. In: Halliwell S, Whysall C and Stratford T, editors. Proceedings of Struct Eng 2010;11(2):129–39.
advanced composites in construction (ACIC 09), Edinburgh, UK, 2009, p. 449–460. [19] Hashemi S, Al-Mahaidi R. Flexural performance of CFRP textile-retrofitted RC
[3] Bisby LA, Kodur VKR, Green MF. Fire endurance of fiber-reinforced polymer-con- beams using cement-based adhesives at high temperature. Constr Build Mater
fined concrete columns. ACI Struct J 2005;102(6):883–91. 2012;28(1):791–7.
[4] Gao WY, Dai JG, Teng JG. Simple method for predicting temperatures in insulated, [20] Dai JG, Munir S, Ding Z. Comparative study of different cement-based inorganic
FRP-strengthened RC members exposed to a standard fire. J Compos Constr pastes towards the development of FRIP strengthening technology. J Compos Constr
2015;19(6). doi: http://doi.org/10.1061/(ASCE)CC.1943-5614.0000566, 2014;18(3). doi: http://doi.org/10.1061/(ASCE)CC.1943-5614.0000420,
04015013. A4013011.
[5] Dai JG, Gao WY, Teng JG. Finite element modeling of insulated FRP-strengthened [21] Wiberg A. Strengthening of concrete beams using cementitious carbon fibre com-
RC beams exposed to fire. J Compos Constr 2015;19(2). http://dx.doi.org/10.1061/ posites. Ph.D. thesis, Royal Institute of Technology, Sweden, 2003.
(ASCE)CC.1943-5614.0000509. 04014046. [22] Triantafillou TC, Papanicolaou CG. Textile reinforced mortars (TRM) versus fiber
[6] Firmo JP, Arruda MRT, Correia JR, Rosa IC. Three-dimensional finite element reinforced polymers (FRP) as strengthening materials of concrete structures. In:
modelling of the fire behaviour of insulated RC beams strengthened with EBR and Shield C, Busel J, Walkup S, Gremel D, editors. Proceedings of the 7th international
NSM CFRP strips. Compos Struct 2018;183:124–36. symposium on fiber-reinforced (FRP) polymer reinforcement for concrete structures
[7] Yu B, Kodur VKR. Effect of high temperature on bond strength of near-surface (FRPRCS-7), SP-230, Missouri, USA, 2005, p. 99–118.
mounted FRP reinforcement. Compos Struct 2014;110:88–97. [23] Elghazy M, Refai AE, Ebead U, Nanni A. Effect of corrosion damage on the flexural
[8] Dai JG, Gao WY, Teng JG. Bond-slip model for FRP laminates externally bonded to performance of RC beams strengthened with FRCM composites. Compos Struct
concrete at elevated temperature. J Compos Constr 2013;17(2):217–28. 2017;180:994–1006.
[9] Bernat-Maso E, Escrig C, Aranha CA, Gil L. Experimental assessment of textile re- [24] Gonzalez-Libreros JH, Sabau C, Sneed LH, Pellegrino C, Sas G. Shear strengthening
inforced sprayed mortar strengthening system for brickwork wallettes. Constr Build of RC beams with FRCM: What do we know so far. In: Teng JG, Dai JG, editors.
Mater 2014;50:226–36. Proceedings of 8th international conference on fibre-reinforced (FRP) composites in
[10] Colajanni P, Domenico FD, Recupero A, Spinella N. Concrete columns confined with civil engineering, Hong Kong, China, 2016, p. 456–461.
fibre reinforced cementitious mortars: experimentation and modelling. Constr Build [25] Khan A, Masood F. Strengthening of shear deficient reinforced concrete beams with
Mater 2014;52:375–84. textile reinforced mortar (TRM). In: Teng JG, Dai JG, editors. Proceedings of 8th
[11] Triantafillou TC, Papanicolaou CG. Shear strengthening of reinforced concrete international conference on fibre-reinforced (FRP) composites in civil engineering,
members with textile reinforced mortar (TRM) jackets. Mater Struct Hong Kong, China, 2016, p. 468–472.
2006;39(1):93–103. [26] Bruckner A, Ortlepp R, Curbach M. Textile reinforced concrete for strengthening in
[12] Gao WY, Teng JG, Dai JG. Effect of temperature variation on the full-range behavior bending and shear. Mater Struct 2006;39(8):741–8.
of FRP-to-concrete bonded joints. J Compos Constr 2012;16(6):671–83. [27] Bosche A, Jesse F, Ortlepp R, Weiland S, Curbach M. Textile-reinforced concrete for
[13] Gao WY, Dai JG, Teng JG. Analysis of Mode II debonding behavior of fiber-re- flexural strengthening of RC-structures-Part 1: structural behavior and design
inforced polymer-to-substrate bonded joints subjected to combined thermal and model. In: Aldea C-M, editor. Design & applications of textile-reinforced concrete,
mechanical loading. Eng Fract Mech 2015;136:241–64. ACI Fall Convention 2007, Fajardo, Puerto Rico, SP-251, 2008, p. 19–40.
[14] Kurtz S, Balaguru P. Comparison of inorganic and organic matrices for strength- [28] Elsanadedy HM, Almusallam TH, Alsayed SH, Al-Salloum YA. Flexural strength-
ening of RC beams with carbon sheets. J Struct Eng 2001;127(1):35–42. ening of RC beams using textile reinforced mortar–Experimental and numerical
[15] Toutanji H, Zhao L, Zhang Y. Flexural behavior of reinforced concrete beams ex- study. Compos Struct 2013;97:40–55.
ternally strengthened with CFRP sheets bonded with an inorganic matrix. Eng [29] Xu S, Kruger M, Reinhardt H-W, Ozbolt J. Bond characteristics of carbon, alkali
Struct 2006;28(4):557–66. resistant glass, and aramid textiles in mortar. J Mater Civ Eng 2004;16(4):356–64.
[16] Kolsch H. Carbon fiber cement matrix (CFCM) overlay system for masonry [30] Schleser M, Walk-Lauffer B, Raupach M, Dilthey U. Application of polymers to
strengthening. J Compos Constr 1998;2(2):105–9. textile-reinforced concrete. J Mater Civ Eng 2006;18(5):670–6.

25
X. Yang et al. Composite Structures 189 (2018) 9–26

[31] Ombres L. Structural performances of reinforced concrete beams strengthened in the tensile capacity up to 8%. Constr Build Mater 2017;137:410–9.
shear with a cement based fiber composite material. Compos Struct [43] Lim YM, Li VC. Durable repair of aged infrastructures using trapping mechanism of
2015;122:316–29. engineered cementitious composites. Cem Concr Compos 1997;19(4):373–85.
[32] Taljsten B, Blanksvard T. Mineral-based bonding of carbon FRP to strengthen [44] Li VC, Horii H, Kabele P, Kanda T, Lim YM. Repair and retrofit with engineered
concrete structures. J Compos Constr 2007;11(2):120–8. cementitious composites. Eng Fract Mech 2000;65(2–3):317–34.
[33] D’Ambrisi A, Mezzi M, Feo L, Berardi VP. Analysis of masonry structures [45] Wang B, Dai JG, Xu SL. Evaluation of the fracture properties of the bond interfaces
strengthened with polymeric net reinforced cementitious matrix materials. Compos between UHTCC overlays and concrete substrates. In: Song GB, Malla RB, editors.
Struct 2014;113:264–71. Proceedings of 12th international conference on engineering, science, construction
[34] Guo R, Hino S, Hatakeyama S, Komori A, Taniguchi K. Bonding behavior and shear and operations in challenging environments, Hawaii, USA, 2010, p. 3512–3520.
strengthening effect of CFRP grid with PCM shotcrete method for RC member. In: [46] Martinola G, Meda A, Plizzari GA, Rinaldi Z. Strengthening and repair of RC beams
Teng JG, Dai JG, editors. Proceedings of 8th international conference on fibre-re- with fiber reinforced concrete. Cem Concr Compos 2010;32(9):731–9.
inforced (FRP) composites in civil engineering, Hong Kong, China, 2016, p. [47] Shin SK, Kim K, Lim YM. Strengthening effects of DFRCC layers applied to RC
473–481. flexural members. Cem Concr Compos 2011;33(2):328–33.
[35] Zheng YZ, Wang WW, Mosalam KM. Strengthened reinforced concrete beams using [48] Dai JG, Wang B, Xu SL. Textile reinforced engineered cementitious composites (TR-
fiber reinforced polymer grid and ultra high toughness cementitious composite. In: ECC) overlays for the strengthening of RC beams. In: Sim JS, editor. Proceedings of
Teng JG, Dai JG, editors. Proceedings of 8th international conference on fibre-re- the 2nd Asia-Pacific conference on FRP in structures, Seoul, Korea, 2009, p. 75–81.
inforced (FRP) composites in civil engineering, Hong Kong, China, 2016, p. [49] Zheng YZ, Wang WW, Brigham JC. Flexural behaviour of reinforced concrete beams
482–489. strengthened with a composite reinforcement layer: BFRP grid and ECC. Constr
[36] Zhang DW, Ueda T, Furuuchi H. Intermediate crack debonding of polymer cement Build Mater 2016;115:424–37.
mortar overlay-strengthened RC beam. J Mater Civ Eng 2011;23(6):857–65. [50] Japan Society of Civil Engineers. Recommendation for design and construction of
[37] Zhang DW, Ueda T, Furuuchi H. Average crack spacing of overlay-strengthened RC high performance fiber reinforced cement composites with multiple cracks, Tokyo,
beams. J Mater Civ Eng 2011;23(10):1460–72. Japan, 2008.
[38] Li VC. On engineered cementitious composites (ECC). J Adv Concr Technol [51] Standing Committee on Concrete Technology. Steel reinforcing bars for the re-
2003;1(3):215–30. inforcement of concrete. Construction Standard CS2:2012, The Government of
[39] Yu KQ, Dai JG, Lu ZD, Leung C. Mechanical properties of engineered cementitious Hong Kong Special Administrative Region, Hong Kong, 2012.
composites subjected to elevated temperatures. J Mater Civ Eng 2015;27(10). doi: [52] Dai JG, Harries KA, Yokota H. A critical steel yielding length model for predicting
http://doi.org/10.1061/(ASCE)MT.1943-5533.0001241, 04014268. intermediate crack-induced debonding in FRP-strengthened RC members. Steel
[40] Wu C, Li VC. Thermal-mechanical behaviors of CFRP-ECC hybrid under elevated Compos Struct 2008;8(6):457–73.
temperatures. Compos B Eng 2017;110:255–66. [53] Buildings Department. Code of practice for structural use of concrete. Hong Kong:
[41] Wu C, Li VC. CFRP-ECC hybrid for strengthening of the concrete structures. Compos The Government of Hong Kong Special Administrative Region; 2013.
Struct 2017;178:372–82. [54] Yu H, Bai YL, Dai JG, Gao WY. Finite element modeling for debonding of FRP-to-
[42] Yu KQ, Wang YC, Yu JT, Xu SL. A strain-hardening cementitious composites with concrete interfaces subjected to mix-mode loading. Polymers 2017;9(9). No. 438.

26