You are on page 1of 20

Width Effect in FRP–Concrete Debonding Mechanism:

A New Formula
Mattia Santandrea1; Imohamed Ali Omar Imohamed2; and Christian Carloni, Ph.D.3
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This paper presents the results of an experimental work aimed at determining the effect of the width of a steel fiber-reinforced
polymer (FRP) composite on the load-carrying capacity of a composite–concrete interface. Single-lap shear tests were performed in
which the main parameter was the width of the composite strip. Steel FRP strips were applied to three different faces of each concrete
prism and it was observed that the load-carrying capacity is strongly related to the face to which the composite is applied. In addition, a
few tests were carried out at a displacement rate, used to control the test, equal to ten times the rate employed for the majority of the spec-
imens. A great influence of the rate on the load-carrying capacity was observed, although the number of tests was limited. The authors argue
that most of the width effect formulas available in the literature use data from different sources without taking into account the face of ap-
plication and the rate. A new width effect formula is proposed, which takes into account the most recent articles on this subject.
DOI: 10.1061/(ASCE)CC.1943-5614.0001001. © 2020 American Society of Civil Engineers.
Author keywords: Steel FRP; Concrete; Debonding; Single-lap shear test; Width effect; DIC.

Introduction and Pecce 2007; Hawileh et al. 2014; Napoli and Realfonzo
2015; Mitolidis et al. 2012; Bencardino and Condello 2015) and
In the field of structural rehabilitation of existing buildings, com- slabs (Napoli et al. 2016b) as a flexural reinforcement, and to
posite materials stood out as one of the most employed strengthen- plain concrete columns (Sneed et al. 2017; El-Hacha and Mashrik
ing techniques in previous decades. In particular, the large use of 2012; Napoli and Realfonzo 2016) as an external confinement. The
fiber-reinforced polymer (FRP) composites to increase the capacity results show that steel FRP can increase the capacity of the struc-
of an existing structural member is mainly due to several advan- tural member and its ductility.
tages associated with FRPs when compared with traditional Carloni et al. (2017a) showed that there is no difference between
strengthening systems. Some of the key features of FRP systems the debonding phenomenon of steel FRP and other FRPs. They
are the high strength-to-weight ratio, the ease of installation, and proved that the facture mechanics approach used to study the de-
the possibility of designing the arrangement of the fibers to fit bonding of FRP strips from a concrete substrate can be applied
the design needs of a specific application. Recently, a new type to steel FRP composites. For this reason, the title of this work refers
of FRP, which employs high-strength steel fibers embedded in a in general to FRP composites and the term FRP is also used for
thermosetting epoxy resin, has been developed for strengthening steel FRP in this paper. The debonding appears to be cohesive,
applications. Steel FRP composites, also known as steel-reinforced that is, a thin layer of mortar-rich concrete remains attached to
polymer (SRP) composites, are a promising strengthening tech- the steel FRP strip as it separates from the substrate. As long as
nique that owes its success to the low cost of the steel fibers with the fracture process occurs in the substrate, the substrate itself
respect to most of the common fibers employed in traditional does not feel the differences between the different fibers employed
FRP systems (carbon, glass, or aramid) and the possibility of bend- in the FRP. Thus, the fracturing process is related to the quality of
ing the fibers without the need of chamfering the edges of columns the substrate and the stiffness of the composite. It might also de-
(Sneed et al. 2017) and beams (Sneed et al. 2016). Steel FRP com- pend on the type of epoxy as it penetrates into the concrete pores.
posites have been successfully applied to reinforced concrete Within the framework of fracture mechanics, adhesion of the
beams (Wobbe et al. 2004; Kim et al. 2005; Barton et al. 2005; composite to concrete is described by a fictitious zero-thickness in-
Casadei et al. 2005; Prota et al. 2006; Pecce et al. 2006; Ceroni terface subjected to Mode-II loading. The Mode-I loading condition is
typically neglected although its presence cannot be avoided in labora-
1
Dept. of Civil, Chemical, Environmental, and Materials Engineering,
tory testing and applications (Carrara et al. 2011; Ghorbani et al. 2017;
Univ. of Bologna, Viale Risorgimento 2, 40136 Bologna, Italy. Email: Carrara and Ferretti 2013; Rabinovitch 2008; Rabinovitch 2012). As
mattia.santandrea3@unibo.it separation between the two materials occurs in a thin layer of concrete;
2 the cohesive behavior of the interface is described by a shear stress (τ)
Dept. of Civil, Chemical, Environmental, and Materials Engineering,
Univ. of Bologna, Viale Risorgimento 2, 40136 Bologna, Italy. Email: versus slip (s) relationship that features a softening branch (Rostasy
imohamedali.imohamed@unibo.it and Neubauer 1997; Ali-Ahmad et al. 2006; Mazzotti et al. 2008;
3
Dept. of Civil Engineering, Case Western Reserve University, Napoli et al. 2016a; Grande et al. 2015; Ceroni et al. 2016).
10900 Euclid Avenue, Cleveland OH 44106 (corresponding author). In this work, an extensive experimental campaign is carried out
ORCID: https://orcid.org/0000-0003-1663-7535. Email: christian.carloni@
to investigate the effect of the width of the FRP strip on the debond-
case.edu
Note. This manuscript was submitted on December 17, 2018; approved
ing mechanism of steel FRP–concrete joints. Tests are performed
on August 14, 2019; published online on April 22, 2020. Discussion period using a single-lap shear test setup. The main parameter investigated
open until September 22, 2020; separate discussions must be submitted for is the width of the composite strip. In addition, tests are performed
individual papers. This paper is part of the Journal of Composites for Con- by varying the face to which the FRP strip is applied. Three tests are
struction, © ASCE, ISSN 1090-0268. performed at a different loading rate in order to show the influence

© ASCE 04020024-1 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


of the rate on the experimental results, as already pointed out by with another quantity used in this paper and to be consistent with
Carloni et al. (2017c) for fabric-reinforced cementitious matrix current standards. The width factor kb is
(FRCM) composites bonded to a concrete substrate. Digital 
image correlation (DIC) is used to investigate the displacement 2 − (bf /b)
and strain fields on the surface of the composite strip. kb = 1.125 (2)
1 + (bf /400)

Background on the Width Effect where b = the width of the concrete prism (Fig. 1). It should be
noted that kb was first proposed by Holzenkämpfer (1994) based
Early studies on the width effect in FRP–concrete joints date back on the size effect formula by Baž ant (Baž ant and Planas 1997;
to the end of the1990s. In this section, only the main contributions Hoover and Baž ant 2013). Eq. (1) was derived by assuming that
are reported. The symbols used in this section are consistent with the fracture energy of the interface was determined as
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

the original papers. Among the first researchers, who observed


GF = kb2 CF fh (3)
that the load-carrying (or bond) capacity of the FRP strip bonded
to a concrete substrate varied with the ratio of the width bf of the The coefficient CF was calibrated against 70 bond tests by the
composite to the width b of the face of concrete prism to which authors. Eq. (3) implies that the fracture energy is not a true mate-
the composite strip is applied (see Fig. 1), Neubauer and Rostasy rial property but depends on the width factor.
(Neubauer and Rostasy 1997; Rostasy and Neubauer 1997) pro- Brosens and Van Gemert (1999) adjusted Eq. (2) by changing
posed the following formula for the maximum bond force Tu,max the expression of kb
(i.e., the bond capacity, which in the next section is identified by 
the symbol Pcrit for experimental values): 2 − (bf /b)
 kb = 1.5 (4)
Tu, max = 0.64kb bf EFRP tFRP fh (1) 1 + (bf /100)

where bf, EFRP, and tFRP = width, Young’s modulus, and thickness Later, Brosens (2001) modified the expression of the fracture
of the composite; and fh = surface tensile strength of concrete [EN energy reported in Brosens and Van Gemert (1999)
1542 (CEN 1999)]. It should be noted that the original symbol used
for the surface tensile strength has been changed to avoid confusion GF = kc kb2 CF fh (5)

(a) (b)

(c) (d)

Fig. 1. (a) Single-lap shear test setup; (b) specimen DS_300_50_HD_D_B_2; (c) casting procedure and definition of the faces of the prism; and
(d) LVDT c and d.

© ASCE 04020024-2 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


where kc depends on the concrete surface condition and varies be- widths, and obtained the following formula for the width factor:
tween 0.65 and 1.00. The width factor is ′R −S
 w = 1 + f c [8(EFRP tFRP )
κ LIN + 0.001](1 − bf /b)0.5 /(1 + 0.01bTf )
2 − (bf /b) (13)
kb = k (6)
1 + (bf /b0 ) where R, S, and T = 0.385, 0.438, and 1.7, respectively.
The width factor proposed by Lin et al. (2017) shall be used in
where k is calibrated against the experiments (as well as CF) and
Eq. (12) proposed by Wu and Jiang (2013) in order to evaluate the
b0 = tref/(k – 1); tref (first introduced in Brosens and Van Gemert
load-carrying capacity of composite strips bonded to a concrete
1999) = the thickness of the portion of concrete near the surface
surface.
of concrete where the composite is applied and can be assumed
Additional numerical work has been carried out to investigate
equal to 2.5–3 times the aggregate size.
the effect of the width (Xu et al. 2015; Benvenuti et al. 2016;
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

Chen and Teng (2001) collected a large database of experimental


Neto et al. 2016; Salomoni et al. 2011), which points to that ob-
results to calibrate the width factor. The bond capacity Pu proposed
served by Subramaniam et al. (2007, 2011), that is, the distribution
by the authors is (when the bonded length is sufficiently long)
of the longitudinal strain component is not constant across the

 width of the composite, which is the source of the width effect.
Pu = 0.427βp bf EFRP tFRP f ′c (7)

where f’c = the cylindrical compressive strength of concrete and the Experimental Program
width factor βp is
 This section reports the procedure used to perform single-lap shear
2 − (bf /b) tests, information about the geometry of the specimens, and the me-
βp = (8)
1 + (bf /b) chanical properties of the materials employed in this experimental
campaign.
It should be noted that, as is highlighted subsequently in this
paper, when data are collected from different studies, certain pa-
rameters, such as the rate of testing and the face of the concrete Materials
prism to which the composite is applied, are typically not reported,
All concrete specimens employed in this experimental campaign
although they strongly affect the results. Teng et al. (2003) slightly
were cast from the same batch of concrete, including specimens
modified the expression of the bond capacity without changing the
used for material characterization. Concrete was normal weight
width factor.
portland cement mixed with limestone and sandstone aggregates,
Additional data were considered by Lu et al. (2005), who pro-
with a water–cement ratio equal to 0.43. The maximum diameter
posed a slightly different expression of the width factor βw
of the aggregates (Dmax) was equal to 15 mm. Compressive tests
 were performed on 150 mm cubes, while splitting tests were per-
2.25 − (bf /b)
βw = (9) formed on 150 mm (diameter) × 300 mm (length) cylinders in
1.25 + (bf /b) order to determine indirectly the tensile strength of concrete. Com-
pressive and tensile tests were performed at different ages accord-
It should be noted that Lu et al. (2005) applied the width factor ing to EN 12390-3 (CEN 2001) and EN 12390-6 (CEN 2009),
to the fracture energy GF. respectively, in order to determine the compressive and tensile
Subramaniam et al. (2007, 2011) conducted two experimental strength as a function of time. The cubic compressive (Rcm) and
campaigns to investigate the width effect and their results somehow tensile splitting ( fctm) strengths at 28 days, obtained as the average
contradicted the expressions of the width factor proposed in the lit- of three tests, were 24.9 and 2.3 MPa, respectively. Their coeffi-
erature. The results presented by the authors were all obtained from cient of variation (CoV) was 0.09 and 0.06, respectively. The
tests conducted at the same testing rate, but the face of the prism to behavior of the compressive and tensile strength versus time is re-
which the composite was applied was not reported. ported elsewhere (Carloni et al. 2017a, b), for the sake of brevity.
Wu and Jiang (2013) performed an analytical study based on a Single-lap shear tests were performed approximately 300 days
database of 628 shear tests. A closed-form solution was used to de- after the concrete prisms were cast. Tests were performed after con-
termine the bond parameters by matching the analytical solution crete was sufficiently aged to assume no variation of its properties
with test results. The coefficients of the bond capacity model during the days of testing since the tests were performed over a
were derived by regression analyses. Wu and Jiang (2013) pro- span of one month. In addition, three-point bending tests of notched
posed a new formulation of the width factor as a function of both beams were performed approximately 300 days after concrete spec-
the width ratio and the concrete strength imens were cast to evaluate the fracture energy of concrete, using
κ WU a fracture mechanics setup reported in Carloni et al. (2017a, b).
w = λ + (1 − λ) · bf /b (10)
Fracture tests were performed on three 600 mm (length) × 150 mm
(width) × 150 mm (depth) notched concrete specimens and on three
λ = 1 + 0.222 · ( f ′c )0.304 (11) 300 mm (length) × 75 mm (width) × 75 mm (depth) notched concrete
specimens. The average value of the fracture energy was equal to
Wu and Jiang (2013) proposed the following formula to evalu-
104 N/m (CoV 0.05) and 109 N/m (CoV 0.12), respectively. The frac-
ate the bond capacity:
ture energy was evaluated based on the concept of work of fracture
Ptheor,WU = Q(EFRP tFRP )−0.5 ( f ′c )J κWU (12) (Hillerborg 1985; Hoover and Baž ant 2013; Elices et al. 1992).
w
Further details about the fracture mechanics tests are provided in
where Q and J = 0.703 and 0.108, respectively. Carloni et al. (2017a, b).
Lin et al. (2017) performed a numerical analysis on composite The composite material consisted of steel fibers (cords) embed-
strips bonded to a concrete substrate, considering several bonded ded in a thermosetting resin. Steel fibers were in the form of a

© ASCE 04020024-3 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Table 1. Properties of the steel fibers (textile) provided by manufacturer with silica sand to obtain a prescribed roughness of 2 mm. The steel
Property High density
FRP composite strips were applied using a wet lay-up procedure, fol-
lowing the manufacturer’s recommendations. After they were applied,
Number of cords/mm 0.472 the composite strips were left to cure for seven days at room temper-
Tensile strength, f fA,u (MPa) f fHD,u = 3,000 ature and humidity. Steel FRP strips were applied approximately be-
Young’s modulus, Ef (GPa) 190
tween 290 and 320 days after the concrete prisms were cast.
Ultimate deformation, ɛf,u (%) 2
Equivalent thickness, t *f ,A (mm) t *f ,HD = 0.254 The main parameter investigated in this study is the width bf of
the composite strip. Six different bonded widths of the steel FRP
Source: Data from Kerakoll (2018). strips were used to strengthen the concrete specimens, that is, nom-
Note: f fA,u is the tensile strength of the fibers. t *f ,A is the equivalent thickness inally 15, 30, 40, 50, 75, and 90 mm. Out of 46 specimens, 40 spec-
of the fibers. Superscript and subscript A represents the steel fiber density
imens had a nominal bonded length equal to 300 mm, while six
(HD = high density). The technical datasheet of the manufacturer reports a
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

value of the tensile strength >3,000. specimens had a nominal bonded length equal to 200 mm. The
bonded length was chosen in order to be longer than the effective
bond length, that is, the length needed to activate completely the
Table 2. Mechanical properties of the epoxy provided by manufacturer stress-transfer mechanism during the debonding phenomenon.
The effective bond length for FRP strips with HD steel fiber
Property Value
sheet was determined in Carloni et al. (2017a) and is within the
Tensile strength (MPa) 14 range 100–150 mm. The nominal thickness of the steel FRP strip
Flexural young’s modulus (GPa) 2.50 was equal to 4 mm. Steel fibers were arranged across the width
Young’s modulus under compression (GPa) 5.30 of the steel FRP strip in order to obtain a distance between the ex-
Source: Data from Kerakoll (2018). ternal fibers and the edges of the matrix equal to half of the fiber
spacing. Since the actual dimensions of the steel FRP strips slightly
unidirectional sheet made of ultrahigh strength galvanized steel differed from the nominal ones, the actual bonded width, bf,actual,
cords fixed to a fiberglass micromesh to facilitate installation (fur- and the actual thickness, tSRP,actual, are reported in Table 3 for
ther details can be found in Santandrea et al. 2016; Sneed et al. each specimen. The value of bf,actual and tSRP,actual for each speci-
2017). Each cord is obtained by joining five filaments. Three fila- men were obtained as the average of three measurements evenly
ments out of five are straight and the remaining two filaments are spaced in the unbonded region of the steel FRP strip and taken
wrapped with a high-torque angle. The cross-sectional area of the after single-lap shear tests were performed. The bonded region
cord, Acord, is equal to 0.538 mm2. High-density (HD) steel fiber started 70 mm from the top edge of the concrete prism, in order
sheets with a fiber spacing approximately equal to 2.12 mm and to provide an initial notch. Fibers were embedded within the matrix
an equivalent thickness, t*f, HD, equal to 0.254 mm, were employed for the entire length of the strip, including the portion outside the
bonded area. In this work, the portion of the steel FRP strip gripped
(subscript HD represents the steel fiber sheet density). The equiva-
by the wedges of the testing machine is referred to as the clamped
lent thickness is obtained by dividing the total cross-sectional area
end. The beginning of the bonded region near the clamped end is
of the fiber sheet by the width of the steel FRP strip. Mechanical
referred to as the loaded end, while the opposite edge of the bonded
properties of the fibers reported by the manufacturer (Kerakoll
area is referred to as the free end (Fig. 1).
2018) are provided in Table 1.
The side to which the steel FRP strip was applied was also varied in
The epoxy matrix is a two-component epoxy thixotropic gel
order to fully understand the width effect and interpret the results pub-
system. The mechanical properties of the epoxy reported by the
lished in the literature. The faces of the concrete prisms were classified
manufacturer (Kerakoll 2018) are provided in Table 2.
in two different types. The side (S) face was one of the two rectangular
Three steel FRP strips were tested in tension according to
formed faces adjacent to the screeded (or casting) face, while the bot-
ASTM D3039 (ASTM 2008). The steel FRP strips were selected
tom (B) face was the formed face opposite to the screeded face
from the strips used to strengthen the concrete specimens, after
[Fig. 1(c)]. Steel FRP strips were applied to both S and B faces.
single-lap shear tests were performed. Since a thin layer of concrete
All specimens were tested using a direct single-lap shear test
was attached to the steel FRP strips at failure, only the unbonded
setup. Two linear variable differential transformers (LVDTs)
region of the composite strips was tested in tension, with nominal were mounted on the concrete surface close to the beginning of
dimensions equal to 370 mm (length) × 50 mm (width) × 4 mm the bonded area. The LVDTs [LVDT a and LVDT b in Fig. 1(a)]
(thickness). The average value of the tensile strength, ff,u, resulted reacted off a thin aluminum Ω-shaped plate that was glued directly
equal to 3,060 MPa (CoV = 0.02). Young’s modulus computed as to the steel FRP strip at the beginning (loaded end) of the bonded
the slope of the stress–strain response between two fixed percentages area. The average value of the LVDTs is referred to as global slip,
of the tensile strength of the fibers, that is, 20% and 40%, was deter- g, in this paper, and was used to control the test at a constant rate.
mined from the strain obtained with DIC. The average Young’s mod- Global slip is a term used in many publications. An alternative and
ulus of the composite strips (EHDf ,SRP ), referred to the area of the fibers, probably more correct term is loaded-end slip because it is the slip
were equal to 258 GPa (CoV = 0.06). Further details about the tensile between the faces of the crack at the beginning of the bonded area
tests of the steel FRP strips are provided in Carloni et al. (2017a). where the load is applied. Out of 46 specimens, 43 were tested at a
rate equal to 0.00084 mm/s (which is considered as the standard
rate in this study), while three specimens were tested at a rate
Methods
equal to 0.0084 mm/s. Two additional LVDTs, named LVDT c
Forty-six specimens were tested using a single-lap shear test setup in and LVDT d, were placed in the back of the prism and were
order to evaluate the stress-transfer mechanism in steel FRP–concrete mounted horizontally in order to measure the out-of-plane displace-
joints. All concrete prisms had the same nominal dimensions equal to ments, wc and wd, respectively [Figs. 1(a–d)]. LVDT c and LVDT d
150 mm (width) × 150 mm (depth) × 600 mm (length). In this paper, reacted off the face of the concrete prism opposite to the one where
the width of the concrete prism is named b, as shown in Fig. 1. the steel FRP strip was applied and were fixed to the bottom plate of
Prior to applying the composite material, all prisms were sandblasted the test setup using two magnets [Fig. 1(d)]. A prestressing load

© ASCE 04020024-4 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Table 3. Characteristics of the single-lap shear test specimens
Steel FRP strip
Fibers
Specimen name Concrete face Actual bonded width bf,actual (mm) Actual thickness tSRP,actual (mm) Number of cords DIC
DS_300_15_HD_D_S_1 S 18.0 4.4 8 Yes
DS_300_15_HD_D_S_2 S 17.5 4.6 8 Yes
DS_300_15_HD_D_S_3 S 18.0 4.4 8 Yes
DS_300_15_HD_D_S_4 S 18.0 5.4 8 Yes
DS_300_15_HD_D_S_5 S 18.0 4.3 8 Yes
DS_300_15_HD_D_B_1 B 18.0 4.2 8 Yes
DS_300_15_HD_D_B_2 B 18.0 4.3 8 Yes
DS_300_30_HD_D_S_1 S 30.0 4.5 14 Yes
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

DS_300_30_HD_D_S_2 S 29.0 4.4 14 Yes


DS_300_30_HD_D_S_3 S 32.0 4.5 14 Yes
DS_300_30_HD_D_S_4 S 32.0 4.5 14 Yes
DS_300_30_HD_D_S_5 S 31.0 4.5 14 Yes
DS_300_30_HD_ND_B_1 B 32.0 4.3 14 No
DS_300_30_HD_D_B_2 B 31.5 4.9 14 Yes
DS_300_30_HD_D_B_3 B 31.5 4.6 14 Yes
DS_300_40_HD_D_S_1 S 43.0 4.8 19 Yes
DS_200_40_HD_D_S_2 S 40.0 4.2 19 Yes
DS_200_40_HD_D_S_3 S 40.0 4.1 19 Yes
DS_200_40_HD_D_S_4 S 40.0 4.3 19 Yes
DS_300_40_HD_ND_B_1 B 42.0 4.8 19 No
DS_300_40_HD_D_B_2 B 42.0 4.6 19 Yes
DS_300_40_HD_D_B_3 B 41.5 4.6 19 Yes
DS_300_50_HD_D_S_1 S 51.0 4.3 24 Yes
DS_300_50_HD_D_S_2 S 52.0 4.6 24 Yes
DS_300_50_HD_D_S_3 S 50.0 4.2 24 Yes
DS_300_50_HD_D_S_4 S 52.0 4.2 24 Yes
DS_300_50_HD_D_S_5 S 52.0 4.7 24 Yes
DS_200_50_HD_ND_S_6 S 52.5 4.5 24 No
DS_200_50_HD_D_S_7 S 53.5 4.5 24 Yes
DS_300_50_HD_ND_S_10R_1 S 53.0 3.9 24 No
DS_300_50_HD_D_S_10R_2 S 53.0 4.3 24 Yes
DS_300_50_HD_D_S_10R_3 S 53.0 4.2 24 Yes
DS_300_50_HD_ND_B_1 B 52.0 4.3 24 No
DS_300_50_HD_D_B_2 B 52.0 4.6 24 Yes
DS_300_50_HD_D_B_3 B 52.5 4.5 24 Yes
DS_200_50_HD_D_B_4 B 52.5 4.2 24 Yes
DS_300_75_HD_D_S_1 S 75.0 4.2 36 Yes
DS_300_75_HD_D_S_2 S 76.5 4.0 36 Yes
DS_300_75_HD_D_S_3 S 74.5 4.4 36 Yes
DS_300_75_HD_D_S_4 S 75.0 4.0 36 Yes
DS_300_75_HD_ND_B_1 B 74.5 4.7 36 No
DS_300_75_HD_D_B_2 B 76.0 4.4 36 Yes
DS_300_75_HD_D_B_3 B 76.0 4.3 36 Yes
DS_300_90_HD_D_B_1 B 94.0 4.0 44 Yes
DS_300_90_HD_ND_B_2 B 93.0 4.2 44 No
DS_300_90_HD_D_B_3 B 92.0 4.4 44 Yes
Note: Specimens for which DIC was employed are marked “Yes” in the DIC column.

was applied to the concrete prism before single-lap shear tests were mm; A represents the steel fiber sheet density (HD = high density); B
performed. Three strain gages were mounted, at midheight, 120° indicates the use of DIC in the test (D = DIC, ND = no DIC); C de-
apart along the circumference of each steel bar used to restrain notes the face of the prism to which the FRP strip was applied (B =
the concrete prism. The average measurement of the three strain bottom, S = side); E (if present) indicates that the loading rate was
gages was used to determine the tensile stress on each bar at the be- different from the standard rate (0.00084 mm/s) used for the majority
ginning of the test and for the entire duration of the test. By check- of the specimens (10R = ten times the standard rate); and Z = speci-
ing the average of the strain measured by the strain gages, the same men number (Table 3). The same specimen notation used in Carloni
prestressing level (1 MPa) was applied to each specimen at the be- et al. (2017a) was adopted in this work, as some specimens published
ginning of the test. For 34 specimens, DIC was used to measure the there were used or recalled in the present study.
displacement field and, consequently, determine the strain field on
the steel FRP strip and adjacent concrete surface. Further details
about the test setup can be found in Carloni et al. (2017a) and Experimental Results
Santandrea et al. (2016).
Specimens were named following the notation DS_X_Y_A_B_C_ This section reports the results of the 46 single-lap shear tests.
E_Z, where X = bonded length (l ) in mm; Y = bonded width (bf) in The applied load (P) divided by the actual width of the composite

© ASCE 04020024-5 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


(bf,actual) versus global slip (g) response of representative specimens section “Failure Modes”). For these specimens, the crack started
is presented and discussed. Failure modes are also discussed. At to develop in the concrete substrate. As the residual bonded length
least three specimens were tested for each width, except for speci- was approximately 100 mm, crack propagation shifted from the
mens with a 15 mm-wide strip applied to the B face. The number of substrate to the matrix–fiber interface. The initial crack propagation
specimens tested for each width is limited but sufficient to investi- in the concrete substrate was associated with the first part of the
gate the trend. A larger number of specimens for each width could load response, similarly to other specimens, and was characterized
be beneficial. However, a longer time to cast all specimens and a by a drop that corresponded to the onset of the crack propagation in
larger number of tests to be completed within a short period the concrete substrate [Fig. 2(b)]. After the drop, the load remained
might cause a variation of the properties of concrete. constant within a limited range of the global slip, and then contin-
ued to increase until failure. The second portion of the load
response, characterized by an increasing branch, corresponded to
Load Responses
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

the propagation of the crack at the matrix–fiber interface.


The load per unit width-global slip responses of representative The value of Pcrit for specimens DS_300_15_HD_D_B_1 and
S-face and B-face specimens are reported in Figs. 2(a and b), re- DS_300_15_HD_D_B_2 is evaluated within the range of the
spectively. Specimens with different widths of the steel FRP strip global slip where the load is approximately constant. The average
are also reported in Fig. 2. All load responses show the same behav- of Pcrit for specimens characterized by the same width of the
ior, characterized by an initial linear portion, followed by a nonlin- steel FRP strip and face to which the composite strip is applied is
termed P  Y,C,E . Superscripts Y, C, and E (if present) are defined
ear branch until the peak load, P*, is reached. After the peak load is crit
reached, a drop in the load response marks the onset of the debond- in the “Methods” section. Here, w  d are the average of the
 c and w
ing mechanism and the interfacial crack starts to propagate along horizontal displacements measured from LVDT c and LVDT d in
the concrete surface. An idealized load response is reported in the global slip range [g1,g2]. The test results are reported in Table 4
Carloni et al. (2017a). As debonding progresses after the load for each specimen presented herein.
drop, the load levels off until the end of the test when the steel
FRP strip completely detaches from the concrete substrate. The pla-
Failure Modes
teau load (or experimental bond capacity), Pcrit, is evaluated as the
average load within the global slip range [g1,g2]. The global slip All specimens tested using the single-lap shear test setup showed
range [g1,g2] is the range where the interfacial cohesive crack prop- the same failure mode, characterized by the detachment of the
agates in a self-similar way along the bonded region and, for most FRP strip from the concrete substrate. A thin layer of concrete,
specimens, is determined through the strain analysis (Carloni et al. rich in cementitious paste, remained attached to the FRP strip as
2017a) based on DIC results. For specimens without DIC, the can be observed in Fig. 3 for specimen DS_300_50_HD_D_B_2.
global slip range [g1,g2] is directly evaluated on the load response. The thickness of the layer of concrete attached to the composite
For these specimens, g1 is determined as the point in the load re- strip was between 1 and 2 mm. Specimens DS_300_15_HD_D_B_1
sponse where a substantial drop is observed after the peak load is and DS_300_15_HD_D_B_2 showed a mixed failure mode,
reached, while g2 is determined as the point in the load response characterized by detachment of the steel FRP strip from the con-
that marks the onset of the last ascending branch in the load re- crete substrate in the first 100 mm of bonded length and by crack
sponse, which is due to the presence of Mode I (Carloni et al. propagation at the matrix–fiber interface for the remaining part
2017a; Carrara et al. 2011; Ghorbani et al. 2017). For all specimens of the bonded area (Fig. 4). The propagation of the crack at
strengthened with a 15 mm-width FRP strip, the global slip range the matrix–fiber interface can be due to the presence of
[g1,g2] was determined directly on the load-global slip response, Mode-I. Since, as observed in Carloni et al. (2017a), the pres-
because of the limited width of the steel FRP strip (see section ence of Mode I determined a greater load in the last portion of
“Longitudinal Strain Across the Composite Width”). Specimens the load-response, it is possible that for specimens with a narrow
DS_300_15_HD_D_B_1 and DS_300_15_HD_D_B_2 showed a strip, the fracture propagated at the fiber–matrix interface be-
different failure mode with respect to all the other specimens (see cause it required a lower load than the one needed to continue

(a) (b)

Fig. 2. Load response of representative specimens: (a) S-face specimens; and (b) B-face specimens.

© ASCE 04020024-6 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Table 4. Results of single-lap direct shear tests
Specimen Name g1 (mm) g2 (mm) P* (kN) Pcrit (kN)  Y,C,E (kN)
P  c (mm)
w  d (mm)
w
crit

DS_300_15_HD_D_S_1 0.60 1.23 7.04 5.37  15,S 4.82


P 0.05 0.03
crit
DS_300_15_HD_D_S_2 0.46 0.72 5.26 4.40 0.00 0.01
DS_300_15_HD_D_S_3 0.65 1.21 6.23 5.37 0.07 0.02
DS_300_15_HD_D_S_4 0.38 0.65 5.12 4.45 0.00 0.00
DS_300_15_HD_D_S_5 0.62 1.19 5.16 4.51 0.00 0.00
DS_300_15_HD_D_B_1 0.42 0.52 7.47 5.78  15,B 5.76
P 0.05 0.06
crit
DS_300_15_HD_D_B_2 0.47 0.52 7.28 5.73 0.03 0.07
DS_300_30_HD_D_S_1 0.41 0.92 8.44 7.50  30,S 7.06
P 0.00 0.02
crit
DS_300_30_HD_D_S_2 0.42 0.84 8.05 7.20 0.00 0.00
DS_300_30_HD_D_S_3 0.62 0.91 9.11 7.50 0.04 0.00
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

DS_300_30_HD_D_S_4 0.40 1.02 7.64 6.71 0.02 0.00


DS_300_30_HD_D_S_5 0.41 0.91 8.50 6.39 0.01 0.04
DS_300_30_HD_ND_B_1 0.27 0.75 10.60 8.76  30,B 8.83
P 0.14 0.20
crit
DS_300_30_HD_D_B_2 0.70 1.01 9.57 8.58 0.08 0.05
DS_300_30_HD_D_B_3 0.78 1.16 11.44 9.16 0.19 0.12
DS_300_40_HD_D_S_1 0.66 0.98 11.09 9.01  40,S 8.89
P 0.10 0.15
crit
DS_200_40_HD_D_S_2 0.21 0.46 11.32 9.01 0.09 0.06
DS_200_40_HD_D_S_3 0.21 0.42 10.04 8.50 0.07 0.06
DS_200_40_HD_D_S_4 0.25 0.46 9.82 9.05 0.14 0.12
DS_300_40_HD_ND_B_1 0.42 0.94 14.17 11.63  40,B 11.57
P 0.16 0.17
crit
DS_300_40_HD_D_B_2 0.50 0.92 12.60 10.99 0.16 0.20
DS_300_40_HD_D_B_3 0.68 1.38 13.81 12.09 0.38 0.42
DS_300_50_HD_D_S_1 0.39 0.78 14.45 12.25  50,S 11.98
P 0.13 0.25
crit
DS_300_50_HD_D_S_2 0.64 1.02 13.82 12.27 0.13 0.07
DS_300_50_HD_D_S_3 0.48 0.80 13.15 11.92 0.05 0.11
DS_300_50_HD_D_S_4 0.49 0.81 12.33 10.44 0.14 0.18
DS_300_50_HD_D_S_5 0.48 0.77 13.01 11.47 0.37 0.33
DS_200_50_HD_ND_S_6 0.36 0.60 15.70 13.07 0.44 0.58
DS_200_50_HD_D_S_7 0.33 0.80 14.10 12.45 0.78 0.68
DS_300_50_HD_ND_S_10R_1 0.60 1.10 15.70 13.87  50,S,10R 14.57
P 0.52 0.40
crit
DS_300_50_HD_D_S_10R_2 0.65 1.20 19.34 15.48 0.68 0.70
DS_300_50_HD_D_S_10R_3 0.44 1.20 17.50 14.37 0.75 0.70
DS_300_50_HD_ND_B_1 0.44 0.98 14.76 12.78  50,B 13.02
P 0.53 0.56
crit
DS_300_50_HD_D_B_2 0.47 0.96 14.22 12.64 0.40 0.37
DS_300_50_HD_D_B_3 0.43 0.85 13.89 12.28 0.17 0.30
DS_200_50_HD_D_B_4 0.39 0.81 15.71 14.39 0.40 0.54
εyy
DS_300_75_HD_D_S_1 0.35 0.92 19.71 17.23 Pcrit 16.94 0.87 0.88
DS_300_75_HD_D_S_2 0.58 0.96 17.34 15.10 0.52 0.50
DS_300_75_HD_D_S_3 1.10 2.38 22.06 18.69 1.07 1.15
DS_300_75_HD_D_S_4 0.74 0.93 19.88 16.76 0.87 0.87
DS_300_75_HD_ND_B_1 0.70 1.48 21.79 20.25  75,B 20.57
P 0.69 1.21
crit
DS_300_75_HD_D_B_2 0.97 1.45 22.56 21.68 1.63 1.54
DS_300_75_HD_D_B_3 0.55 1.05 22.08 19.79 1.47 1.34
DS_300_90_HD_D_B_1 0.56 1.21 28.44 25.77  90,B 24.83
P 2.08 2.18
crit
DS_300_90_HD_ND_B_2 0.42 0.90 25.37 23.37 1.45 1.43
DS_300_90_HD_D_B_3 0.47 1.10 27.88 25.37 1.68 1.63

to fracture the concrete substrate through the detachment of a Longitudinal Strain Across the Composite Width
large bulb of concrete. The presence of Mode-I was observed
for single-lap shear tests on steel FRP composites in Carloni DIC was employed to determine the displacement and consequently,
et al. (2017a) and for other FRP composites in Carrara et al. obtain the strain fields on the composite surface and on the portion of
(2011). It can be observed that for B-face specimens, the amount concrete surface close to the composite edges. The displacement field
of large aggregates that remained attached to the steel FRP strips was determined using a 5-pixel step size, which implies a spacing be-
was greater than the amount for S-face specimens, for which the tween two consecutive points equal to 1.67 mm. DIC analysis was
concrete that remained attached to the composite is rich in small performed using three different subsets (i.e., 21, 31, and 41 pixels),
aggregates. in order to investigate the influence of the subset dimension on the
Fig. 5 compares the distribution of the aggregates in the layer displacement field. Results obtained using different subsets were
of concrete attached to the steel FRP strips after tests were per- similar. The 21-pixel subset, which corresponds to 7 mm square
formed for specimen DS_300_75_HD_D_S_2 and specimen areas, was chosen to evaluate the strain profile in the remainder of
DS_300_75_HD_ND_B_1. From this comparison, it appears that the paper (Carloni and Subramaniam 2010). The strain field evalu-
during casting a large amount of aggregates concentrated near ated from the DIC analysis and reported in this experimental work
the B-face of the prism with respect to the S-face. The influence refers to the Cartesian system in Fig. 1(a). The variation of the lon-
of the face, to which the composite strip is applied, on the load re- gitudinal strain component, ɛyy, across the width of the specimen is
sponses is discussed in the “Width Effect” section. shown in Fig. 6 for all the widths investigated in this study. The strain

© ASCE 04020024-7 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


plots of Figs. 6(a–f) refer to the representative B-face specimens
represented in Fig. 2b. Here, ɛyy is evaluated at different locations
along the steel FRP strip, corresponding to different values of the
y coordinate. For all specimens, the values of y, for which the longi-
tudinal strain has been plotted, are comprised within the stress trans-
fer zone (STZ), that is, the portion of the steel FRP strip where shear
stresses are transferred to the concrete substrate (Carloni et al.
2017a). If the width of the composite strip is between 30 and
90 mm, the trend of ɛyy across the width of the composite is charac-
(a)
terized by a central region in which ɛyy is nominally constant. Close
to the edges of the composite strip, ɛyy starts to decrease until it be-
comes null on the concrete surface at a distance from the composite
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

edge approximately equal to 10 mm. The central region of the steel


FRP strip, characterized by a constant value of ɛyy, becomes larger as
(b)
the width of the composite strip increases. The lateral regions, char-
acterized by a gradient of ɛyy, have approximately the same width
equal to 20 mm on both sides of the composite strip, regardless of
the width of the strip. The 15 mm-wide strips do not exhibit a central
(c) region within the composite width with nominally constant strains
[Fig. 6(a)], which could be associated with the different failure
Fig. 3. Failure mode for specimen DS_300_50_HD_D_B_2: (a) con- mode observed previously. Fig. 7 shows the variation of the shear
crete substrate; and (b) fracture surface on the steel FRP strip; and strain, ɛyx, across the width of the composite strip for specimen
(c) side view of the steel FRP strip. DS_300_15_HD_D_B_1 and specimen DS_300_50_HD_D_B_2.
The shear strain, for specimen DS_300_15_HD_D_B_1, continu-
ously changes across the width of the steel FRP strip, determining
a fracture mixed Mode II and Mode III condition (Subramaniam
et al. 2007). Conversely, for specimen DS_300_50_HD_D_B_2,
the central region of the composite strip does not experience any
shear strain, which implies that a Mode II loading condition exists.
It should be noted that a mixed Mode II and Mode III fracture con-
dition occurs in the edge regions of the composite strip.
It can be assumed that as the width of the composite strip in-
creases, the central portion characterized by a pure Mode II condi-
tion tends to become larger. This behavior is confirmed for S-face
specimens. It can be observed that in the central region of the com-
posite strip, for specimens with a width of the FRP strip larger or
(a) equal to 30 mm, ɛyy exhibits some fluctuation due to the nonuni-
form distribution of the aggregates across the width of the specimen
(Ali-Ahmad et al. 2006).

(b) Strain Analysis in Direct-Shear Tests and Interfacial


Cohesive Material Law (CML)
Fig. 4. Failure mode for specimen DS_300_15_HD_D_B_1: (a) con-
crete substrate; and (b) fracture surface on the steel FRP strip.
The DIC analysis, introduced in the “Longitudinal Strain Across
the Composite Width” section, was also used to obtain the profile
of the longitudinal strain component ɛyy along the bonded length
of the composite strip. Because of the nonconstant trend of ɛyy
across the width of the composite (observed in the previous sec-
tion), for each value of the y coordinate along the strip, ɛyy was
averaged across a 15 mm-wide strip centered with respect to the
width of the composite. Averaging the values of ɛyy allowed for
(a) taking in to account the nonhomogeneous distribution of the aggre-
gates. The 15-mm strip was chosen based on the maximum aggre-
gate size (Dmax) of concrete (Ali-Ahmad et al. 2006). Different
widths (10 and 20 mm) of the strip were used to average the strain
and no significant difference was observed. Three different func-
tions were used to fit the experimental profile of ɛyy along the
steel FRP strip, that is, an exponential function (Dai et al. 2005),
a sinusoidal function [Eq. (3) in Carloni et al. 2017a], and a poly-
(b) nomial function [Eq. (7) in Carloni et al. 2017a]. Additional details
on the choice of the functions are reported in Carloni et al. (2017a).
Fig. 5. Fracture surface: (a) specimen DS_300_75_HD_D_S_2; and
Since the fitting procedure is fully described in Carloni et al.
(b) specimen DS_300_75_HD_ND_B_1.
(2017a), it will not be revisited herein for the sake of brevity.

© ASCE 04020024-8 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c) (d)

(e) (f)

Fig. 6. Plot of ɛyy across the width of the steel FRP strip at different locations along the bonded area for specimen (a) DS_300_15_HD_D_B_1;
(b) DS_300_30_HD_D_B_2; (c) DS_300_40_HD_D_B_3; (d) DS_300_50_HD_D_B_2; (e) DS_300_75_HD_D_B_2; and
(f) DS_300_90_HD_D_B_3.

The stress-transfer zone (STZ) is the portion of the FRP strip where maximum value (ɛmax). Fig. 8(a) shows the load response of speci-
the stress transfer occurs, and is usually characterized by an men DS_300_90_HD_D_B_3. Fig. 8(b) shows the best fit of the
“S-shaped” profile of ɛyy. The fully debonded zone (FDZ) is the strain profile using the exponential function (Dai et al. 2005) for
portion of the FRP strip that has already detached from the concrete different DIC images that correspond to different points of the load-
substrate and where ɛyy is nominally constant and equal to the global slip response [Fig. 8(a)]. It can be observed that the STZ

© ASCE 04020024-9 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

(b)
(a)

Fig. 7. Shear strains component ɛxy across the width of the steel FRP strip at different locations along the bonded area for specimen
(a) DS_300_15_HD_D_B_1; and (b) DS_300_50_HD_D_B_2.

(a) (b)

(c) (d)

Fig. 8. Specimen DS_300_90_HD_D_B_3: (a) load response; (b) experimental ɛyy profile for point C, and fitting strain profiles, using the exponential
function (Dai et al. 2005), for points A, B, C, D, and E; (c) experimental ɛyy profile for point C and fitting functions; and (d) cohesive material law
τzy(s) obtained using the exponential function (Dai et al. 2005), the sinusoidal function [Eq. (3) in Carloni et al. 2017a], and the polynomial function
[Eq. (7) in Carloni et al. 2017a] to fit the ɛyy profile of point C.

© ASCE 04020024-10 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


moves gradually from the loaded end to the free end of the compos- The effective bond length, that is, the length of the composite
ite strip. In Fig. 8(a), the global slip range [g1,g2] (see “Load strip needed to completely establish the shear stress transfer, can
Responses” section) is marked with dashed lines. Fig. 8(c) shows be evaluated as the length of the STZ. The procedure to determine
the best fit of the strain profile at point C of the load-global slip re- Leff associated with each fitting strain function is described in
sponse [Fig. 8(a)] using the three functions mentioned previously. Carloni et al. (2017a).
Once an analytical function that approximates the experimental The average values of the fracture energy, G  F , and the effective
trend of ɛyy is available, the interfacial shear stress and slip at any 
bond length, Leff , were determined for each specimen by averaging
location along the composite strip can be determined (Carloni the values of the parameters obtained by the DIC analysis of at least
et al. 2017a) 10 images within the range [g1,g2]. Table 5 provides for each speci-
men the values of G  F and L
eff , obtained using the three different
dεyy
τzy (y) = Ef t *f ,A (14) functions (exponential, sinusoidal, and polynomial) to approximate
dy the experimental ɛyy profile.
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

y
s(y) = εyy dy (15)
0 Width Effect
Combining Eqs. (14) and (15) allows for the determination of The failure modes, presented in the “Failure Modes” section, indi-
the cohesive material law (CML) τzy(s). The CMLs evaluated at cate that debonding of the steel FRP strip occurs as interfacial crack
point C of the load response of specimen DS_300_90_HD_D_B_3 propagation in a thin layer of concrete, which is the typical failure
[see Fig. 8(a)], using three different functions to fit ɛyy, are shown in mode of traditional FRP systems. The analogy between FRP and
Fig. 8(d). steel FRP, discussed in Carloni et al. (2017a), allows employing
The Mode-II fracture energy, that is, the energy needed to fully the formula proposed by Täljsten (1996) for FRP–concrete joints
break the unit area of a cohesive crack, is equal to the area under in order to evaluate the theoretical load-carrying (or bond) capacity
the CML Ptheor of the steel FRP–concrete interface
∞

GF = τzy (s)ds (16) Ptheor = bf 2Ef t*f ,A GF (17)
0

Table 5. Fracture mechanics parameters


eff (mm) (CoV)
L  F (N/m) (CoV)
G

Sinusoidal Eq. (3) Polynomial Eq. (7) Sinusoidal Eq. (3) Polynomial Eq. (7)
Exponential in Carloni et al. in Carloni et al. Exponential in Carloni et al. in Carloni et al.
Specimen name (Dai et al. 2005) (2017a) (2017a) (Dai et al. 2005) (2017a) (2017a)
DS_300_30_HD_D_S_1 114.5 (0.13) 151.2 (0.09) 100.3 (0.11) 450 (0.08) 480 (0.09) 440 (0.12)
DS_300_30_HD_D_S_2 101.0 (0.27) 148.6 (0.12) 96.7 (0.29) 380 (0.16) 390 (0.16) 370 (0.16)
DS_300_30_HD_D_S_3 86.0 (0.20) 133.4 (0.04) 69.8 (0.17) 490 (0.15) 470 (0.16) 450 (0.14)
DS_300_30_HD_D_S_4 76.0 (0.19) 101.6 (0.24) 71.3 (0.18) 310 (0.14) 330 (0.14) 310 (0.14)
DS_300_30_HD_D_S_5 85.0 (0.16) 149.6 (0.11) 79.3 (0.17) 270 (0.18) 290 (0.16) 270 (0.17)
DS_300_30_HD_D_B_2 113.0 (0.12) 126.3 (0.05) 97.5 (0.15) 670 (0.16) 670 (0.17) 640 (0.17)
DS_300_30_HD_D_B_3 116.0 (0.13) 146.5 (0.05) 103.5 (0.28) 720 (0.11) 760 (0.11) 710 (0.10)
DS_300_40_HD_D_S_1 125.3 (0.16) 145.3 (0.12) 105.4 (0.25) 350 (0.10) 350 (0.11) 330 (0.12)
DS_200_40_HD_D_S_2 93.9 (0.15) 97.8 (0.15) 79.2 (0.16) 320 (0.27) 340 (0.19) 320 (0.19)
DS_200_40_HD_D_S_3 89.7 (0.17) 98.4 (0.11) 79.4 (0.16) 320 (0.09) 310 (0.10) 310 (0.05)
DS_200_40_HD_D_S_4 124.5 (0.15) 110.9 (0.14) 101.0 (0.19) 380 (0.10) 370 (0.10) 350 (0.12)
DS_300_40_HD_D_B_2 84.7 (0.04) 133.4 (0.15) 77.0 (0.10) 530 (0.21) 580 (0.14) 510 (0.18)
DS_300_40_HD_D_B_3 110.1 (0.27) 126.9 (0.21) 98.7 (0.28) 750 (0.10) 770 (0.08) 760 (0.09)
DS_300_50_HD_D_S_1 147.5 (0.16) 147.7 (0.12) 119.5 (0.21) 470 (0.11) 460 (0.08) 440 (0.07)
DS_300_50_HD_D_S_2 125.4 (0.14) 141.5 (0.12) 109.6 (0.12) 420 (0.11) 420 (0.08) 400 (0.07)
DS_300_50_HD_D_S_3 165.1 (0.11) 173.7 (0.10) 133.0 (0.26) 430 (0.17) 450 (0.23) 440 (0.20)
DS_300_50_HD_D_S_4 144.9 (0.09) 173.9 (0.06) 128.7 (0.19) 300 (0.14) 340 (0.20) 320 (0.21)
DS_300_50_HD_D_S_5 111.6 (0.10) 133.9 (0.09) 99.6 (0.18) 360 (0.12) 370 (0.10) 350 (0.09)
DS_200_50_HD_D_S_7 88.0 (0.10) 101.2 (0.12) 78.0 (0.12) 400 (0.11) 410 (0.11) 400 (0.10)
DS_300_50_HD_D_B_2 120.0 (0.18) 130.7 (0.20) 106.1 (0.23) 420 (0.11) 440 (0.12) 430 (0.11)
DS_300_50_HD_D_B_3 138.4 (0.16) 136.3 (0.15) 124.7 (0.23) 420 (0.12) 450 (0.17) 440 (0.14)
DS_200_50_HD_D_B_4 121.5 (0.16) 119.6 (0.14) 85.1 (0.19) 730 (0.16) 730 (0.11) 650 (0.15)
DS_300_75_HD_D_S_1 122.6 (0.14) 153.1 (0.06) 118.9 (0.24) 380 (0.16) 370 (0.22) 350 (0.18)
DS_300_75_HD_D_S_2 121.3 (0.35) 129.6 (0.29) 94.8 (0.32) 290 (0.15) 270 (0.11) 280 (0.06)
DS_300_75_HD_D_S_3 110.2 (0.18) 151.8 (0.08) 90.5 (0.21) 470 (0.11) 460 (0.13) 460 (0.10)
DS_300_75_HD_D_S_4 153.4 (0.15) 153.8 (0.12) 114.5 (0.24) 370 (0.08) 350 (0.12) 350 (0.12)
DS_300_75_HD_D_B_2 135.0 (0.09) 140.2 (0.07) 105.9 (0.10) 880 (0.10) 790 (0.05) 760 (0.06)
DS_300_75_HD_D_B_3 108.0 (0.22) 147.5 (0.08) 92.7 (0.16) 570 (0.09) 560 (0.11) 530 (0.10)
DS_300_90_HD_D_B_1 133.0 (0.12) 150.1 (0.11) 118.4 (0.19) 520 (0.07) 600 (0.10) 580 (0.12)
DS_300_90_HD_D_B_3 111.2 (0.15) 127.0 (0.11) 96.2 (0.17) 500 (0.06) 520 (0.11) 490 (0.11)
DS_300_50_HD_D_S_10R_2 162.6 (0.29) 146.7 (0.32) 128.2 (0.39) 850 (0.15) 860 (0.13) 810 (0.11)
DS_300_50_HD_D_S_10R_3 115.6 (0.29) 122.6 (0.19) 100.9 (0.25) 590 (0.09) 590 (0.08) 570 (0.08)

© ASCE 04020024-11 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Table 6. Comparison between the theoretical and experimental bond capacity
 theor /Pcrit (%)
P

Exponential (Dai et al. Sinusoidal Eq. (3) in Carloni et al. Polynomial Eq. (7) in Carloni et al.
Specimen name Pcrit (kN) 2005) (2017a) (2017a)
DS_300_30_HD_D_S_1 7.50 83.7 86.1 82.4
DS_300_30_HD_D_S_2 7.20 76.9 78.4 75.9
DS_300_30_HD_D_S_3 7.50 92.4 90.8 88.5
DS_300_30_HD_D_S_4 6.71 82.6 84.7 82.0
DS_300_30_HD_D_S_5 6.39 78.2 81.1 77.7
DS_300_30_HD_D_B_2 8.58 93.6 93.5 91.3
DS_300_30_HD_D_B_3 9.16 90.7 93.4 89.8
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

DS_300_40_HD_D_S_1 9.01 87.2 88.1 85.6


DS_200_40_HD_D_S_2 9.01 78.6 80.1 78.2
DS_200_40_HD_D_S_3 8.50 82.1 82.0 82.0
DS_200_40_HD_D_S_4 9.05 84.6 83.0 81.6
DS_300_40_HD_D_B_2 10.99 86.2 90.1 84.4
DS_300_40_HD_D_B_3 12.09 92.6 93.8 92.8
DS_300_50_HD_D_S_1 12.25 88.3 88.0 85.4
DS_300_50_HD_D_S_2 12.27 85.3 85.3 83.6
DS_300_50_HD_D_S_3 11.92 85.2 87.8 86.2
DS_300_50_HD_D_S_4 10.44 84.6 90.6 86.9
DS_300_50_HD_D_S_5 11.47 84.0 86.2 83.3
DS_200_50_HD_D_S_7 12.45 84.2 85.6 83.9
DS_300_50_HD_D_B_2 12.64 82.5 84.6 83.3
DS_300_50_HD_D_B_3 12.28 85.2 89.1 88.4
DS_200_50_HD_D_B_4 14.39 97.0 96.7 91.1
DS_300_75_HD_D_S_1 17.23 83.1 82.5 79.7
DS_300_75_HD_D_S_2 15.10 85.5 81.9 82.9
DS_300_75_HD_D_S_3 18.69 85.2 83.6 83.8
DS_300_75_HD_D_S_4 16.76 84.3 81.8 81.7
DS_300_75_HD_D_B_2 21.68 101.9 96.8 94.9
DS_300_75_HD_D_B_3 19.79 90.2 89.5 87.1
DS_300_90_HD_D_B_1 25.77 81.5 87.4 86.2
DS_300_90_HD_D_B_3 25.37 80.0 81.0 78.5
DS_300_50_HD_D_S_10R_2 15.48 98.0 98.7 95.7
DS_300_50_HD_D_S_10R_3 14.37 87.8 87.7 86.5

Since Ptheor requires that the fracture energy is known, the aver-
age fracture energy of each test G  F , obtained from the direct proce-
dure (see previous section) associated with the fitting of the ɛyy
profile with three different strain functions, was used. Thus, three
values of the theoretical load-carrying capacity, named P  theor and
associated with the three fitting functions, are computed for each
test. Table 6 reports, for each specimen analyzed with DIC, the per-
centage ratio P  theor /Pcrit . For almost all the specimens, the percent-

age ratio Ptheor /Pcrit is comprised within the range 80%–95%. The
value of P  theor is always lower than Pcrit. As discussed in Carloni (a) (b)
et al. (2017a), the value of P  theor depends on the values of Young’s
Fig. 9. Fracture surface, longitudinal strain component in the FRP, and
modulus and thickness of the fibers. If the modulus of the compos-
 theor would be al- interfacial shear stress distribution: (a) pure Mode-II condition; and
ite and its actual thickness are used, the value of P
(b) experimental evidence.
ways greater than the experimental Pcrit. Several factors should be
considered in discussing the difference between P  theor and Pcrit.
1. Eq. (17) is derived under the assumption of constant ɛyy across
the entire width of the composite [Fig. 9(a)]. In the “Longitudi- is a 3D phenomenon that affects an area that is larger than the
nal Strain across the Composite Width” section, it has been ob- horizontal projection of the bonded region of the FRP strip
served that ɛyy is constant only in the central region of the width [Fig. 9(b)]. The fracture surface is not plane and therefore,
of the FRP strip, while close to the edges there is a high gradient Eq. (17) should be modified. The actual shape of the fracture
of ɛyy with the additional presence of shear strain. surface depends on the porosity of concrete and therefore on
2. In addition to a constant strain cross the width of the composite, the depth of penetration of epoxy into the concrete substrate.
as shown in Fig. 9(a), Eq. (17) implies that the interfacial frac- This aspect is not investigated in this work. It would require re-
ture surface has the same width of the composite and is nomi- peating the same experimental campaign for different types of
nally plane. concrete. It is expected that the load-carrying capacity will de-
3. Furthermore, when the width of the FRP strip, bf, is smaller than pend on the type of concrete, that is, on the thickness of the
the width of the concrete prism, b, the debonding phenomenon thin layer of substrate involved. However, it is impossible to

© ASCE 04020024-12 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


yyε
Table 7. Evaluation of Pcrit for specimen DS_300_90_HD_D_B_3
ε
Point Parameter yD − Dmax
2 yD − 38 Dmax yD − Dmax
4 yD − Dmax
8 yD yD + Dmax
8 yD + Dmax
4 yD + 38 Dmax yD + Dmax
2
yy
Pcrit (kN)
Point A y (mm) 282.9 281.1 279.3 277.5 275.7 273.9 272.1 270.4 268.6 25.25
εyy
Pcrit (kN) 27.20 26.60 25.79 24.93 24.16 23.61 23.19 22.89 22.86
Point B y (mm) 256.1 254.3 252.5 250.7 248.9 247.1 245.4 243.6 241.8
εyy
Pcrit (kN) 30.60 30.71 30.60 30.07 29.21 28.00 26.55 25.08 23.85
Point C y (mm) 240.0 238.2 236.4 234.6 232.9 231.1 229.3 227.5 225.7
εyy
Pcrit (kN) 27.31 27.53 27.43 27.06 26.55 26.03 25.52 24.99 24.25
Point D y (mm) 197.1 195.4 193.6 191.8 190.0 188.2 186.4 184.6 182.9
εyy
Pcrit (kN) 22.93 21.41 20.26 19.59 19.47 19.98 20.96 22.01 22.91
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

Point E y (mm) 159.6 157.9 156.1 154.3 152.5 150.7 148.9 147.1 145.4
εyy
Pcrit (kN) 23.36 22.79 23.15 24.03 25.71 27.16 28.47 29.54 30.20

predict whether the trend of the width effect could be affected by for bf/b equal to 0.2, 0.27, 0.33, and 0.5, respectively. It can be ob-
the thickness of the layer of concrete detached. Further work is served that the fracture energy is almost constant, regardless of the
needed in this area. ratio bf/b. Similar results were obtained using the fitting function
It can be observed that if the strain on the surface is assumed to be proposed by Dai et al. (2005) or the sinusoidal function proposed
representative of the strain throughout the thickness of the composite, by the authors. Those results are not plotted for the sake of brevity.
then the strain across the width at the beginning of the STZ (at the The fracture energy GFPcrit , for S-face specimens, is equal to 550,
loaded end) can be used to compute the load-carrying capacity: 495, 541, and 530 N/m for bf/b equal to 0.2, 0.27, 0.33, and 0.5,
b f respectively. The average value of G  F is lower than GPcrit . The rea-
F
εyy  F was
Pcrit = t *f ,HD E HD
f ,SRP εyy |y=yD dx (18) son of this difference can be explained considering that G
0 evaluated considering a 15 mm-wide strip within the central region
where yD = the coordinate corresponding to the end of the STZ eval- of the steel FRP strip, while GFPcrit is determined directly from the
uated, for example, through the polynomial function [Eq. (7) in load-carrying capacity Pcrit. Here, G  F is a pure Mode-II fracture en-
Carloni et al. 2017a]. Eq. (18) does not take into account what was ob- ergy and does not take into account the mixed Mode-II–Mode-III
served in point 3). In Eq. (18), Young’s modulus of the composite re- condition close to the edges of the strip. On the other hand, GFPcrit
ferred to the fibers, EHD f ,SRP , evaluated in the “Materials” section, has
can be considered as a fictitious fracture energy influenced by the
been employed. As discussed in Carloni et al. (2017a), the actual mod- presence of different fracture modes across the width of the strip.
ulus of the composite should be always used for the evaluation of the Considering B-face specimens analyzed with the polynomial strain
load-carrying capacity of FRP–concrete joints. The actual Young’s function, the average value of G  F was equal to 673, 631, 504, 647,
modulus of the composite varies for each strip since it depends on and 532 N/m for bf/b equal to 0.2, 0.27, 0.33, 0.5, and 0.6, respec-
the percentage volume of fibers and matrix in the cross section of tively. Similar results, not reported in this work for the sake of brevity,
the strip. To overcome this problem, the Young’s modulus of the fi- were obtained using the other fitting functions. The fracture energy,
bers, Ef, and the equivalent thickness of the fibers, t*f ,A are generally GFPcrit , for B-face specimens, is equal to 822, 794, 652, 773, and 783
used in the literature to evaluate the load-carrying capacity of FRP N/m for bf/b equal to 0.2, 0.27, 0.33, 0.5, and 0.6, respectively. As ob-
strips bonded to a concrete substrate, even if this procedure entails a served for S-face specimens, GFPcrit was always also greater than G F
slight underestimation of the load-carrying capacity. also for B-face specimens (if Young’s modulus of the fibers Ef is
εyy
For specimen DS_300_90_HD_D_B_3, the value of Pcrit is used). Results are more scattered than the ones obtained for the
evaluated for different values of the global slip, that is, points A, S-face specimens. It is worth noting that the overall number of
B, C, D, and E in the load response of Fig. 8(a). For each point, B-face specimens tested with DIC (11 specimens) is lower than the
εyy
the value
 of DPmax
crit is evaluated for different values of y within the number of S-face specimens tested with DIC (19 specimens), and
εyy
range yD − 2 ; yD + Dmax 2 . Finally, averaging the value of Pcrit for each width no more than three B-face specimens were tested
obtained for the five different points, the mean value of the load- with DIC. The scatter in the results for B-face specimens can be attrib-
εyy
carrying capacity, Pcrit , is calculated (Table 7). It can be observed uted partially to the low number of specimens tested with DIC.
εyy
that for specimen DS_300_90_HD_D_B_3 the value of Pcrit is Comparing the average values of G  F obtained for each bf/b ratio
equal to 25.25 kN, which is very similar to the experimental for B-face specimens with the equivalent values determined for
value of Pcrit (25.37 kN). S-face specimens, it can be noted that B-face specimens always
The fracture energy and effective bond length, evaluated in the present a fracture energy that is greater than S-face specimens.
previous section, are plotted in Fig. 10 as a function of the ratio bf/b. Considering the polynomial strain function, the ratio of the average
Figs. 10(a and b) show the average value of G  F together with the  F for B-face specimens to the average value of G  F for
value of G
standard deviation obtained using the polynomial function (Carloni S-face specimens is equal to 1.84, 1.90, 1.30, and 1.81 for bf/b
et al. 2017a), for S-face and B-face specimens, respectively. In the equal to 0.2, 0.27, 0.33, and 0.5, respectively. Part of these results
same graphs is also reported the fracture energy, GFPcrit , which has have been already presented in Carloni et al. (2017a) and it was ob-
been back-calculated from Eq. (17), substituting Ptheor with Pcrit served that steel FRP strips applied to the B face of the specimen
for each specimen: had a Mode-II fracture energy greater than strips applied on the S
2 face of the specimen. Due to the geometry of the molds and the di-
Pcrit
GFPcrit = (19) rection of casting [Fig. 1(c)], a large amount of aggregates distrib-
2bf Ef t *f ,A
2
utes near the B face of the specimen with respect to the S face. The
Considering the polynomial strain function, the average value of large amount of aggregates influences the debonding phenomenon,
 F for S-face specimens is equal to 365, 330, 390, and 357 N/m
G which in turn might require a greater amount of energy to propagate

© ASCE 04020024-13 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c) (d)

Fig. 10. Comparison between the fracture energy evaluated through a direct procedue based on the fitting of the strain profile using a polynomial
function [Eq. (7) in Carloni et al. (2017a)] and the fracture energy obtained from Eq. (19) for (a) S-face specimens; and (b) B-face specimens. Com-
parison between the value of Leff obtained from the polynomial function [Eq. (7) in Carloni et al. (2017a)] that approximates the strain profle and
through Eq. (20) proposed by Chen and Teng (2001) for (c) S-face specimens; and (d) B-face specimens.

the interfacial cohesive crack. The presence of more and larger composite strip is applied do not affect the effective bond length.
aggregates near the B face could also explain the larger scatter of A similar observation is reported in Carloni et al. (2017a). In Figs.
the results in terms of fracture energy when compared to the 10(c and d), the value of Leff determined using the formula proposed
S-face results. The reader might be confused by this result, that is, by Chen and Teng (2001) is reported for comparison:
the fracture energy G F varies between B-face and S-face specimens. 
Here, G  F is the energy required to extend the crack by a unitary Ef t *f ,A
eff =
LC−T ′  (20)
amount along the surface of concrete (roughly 1–2 mm underneath f c
the surface). Thus, if the surface changes (from S face to B face) ′
the energy is not supposed to be the same because it is similar to test- where the cylindrical compressive strength at 28 days fc was indi-
ing a different material (i.e., a different interface). On the other hand, rectly evaluated from Rcm (European Standards 2004)
the fracture energy of concrete corresponds to the energy required to
f ′c = 0.83Rcm (21)
extend a crack within the bulk of the concrete. This energy must be
constant because it is a property of the bulk of the concrete. From Figs. 10(c and d), it can be observed that the value of LC−T
Figs. 10(c and d) show the value of L eff together with the standard obtained from Chen and Teng (2001) formula is in good agreement
eff

deviation obtained using the polynomial function for S-face and with the average value of L eff determined for S-face and B-face
B-face specimens, respectively. The average values of the effective specimens for different values of the ratio bf/b, using the polyno-
bond length are quite scattered when plotted versus the ratio bf/b. mial fitting function. A good agreement is observed also using
For S-face and B-face specimens, the value of L eff evaluated through the Dai et al. (2005) fitting function (Table 5). Using the sinusoidal
the polynomial function is comprised within the range 83 mm–111, function, the values of Leff seemed to be slightly different with re-
and 88–107 mm, respectively [Figs. 10(c and d)]. From the available C−T
spect to the value of Leff obtained from Chen and Teng (2001)
results, it appears that the ratio bf/b and the face to which the (see Table 5).

© ASCE 04020024-14 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c)

Fig. 11. P Y,C,E


per unit width for different bf/b ratios: (a) experimental results using the actual width of the FRP strip; (b) predictive behavior from the
crit
literature; and (c) best fitting using Eq. (22) and Eq. (23).

Fig. 11(a) shows the plot of P Y,C,E


for each width and face di- the rate used for the other tests. The value of P Y,C,E
for these three
crit crit
vided by the actual width of the composite bf,actual as a function specimens was 21.6% higher than S-face specimens tested at stan-
of the ratio bf/b. It can be observed that for both S-face and dard rate. Given these observations, it appears that an accurate pre-
B-face specimens, high values of Pcrit per unit width are reached dictive formula can be defined only if homogeneous specimens
when the steel FRP composite strip is narrow (15 mm). As the tested with a standard procedure are considered.
width of the strip increases, the value of Pcrit per unit width tends In Fig. 11(b), the value of P  Y,C,E per unit width (using the nom-
crit
to plateau, that is, no substantial variations of Pcrit per unit width inal width) for S-face and B-face specimens is shown, together with
are observed when the ratio bf/b is increased. This behavior can the most important models proposed in literature (Lin et al. 2017;
be explained considering the strain analysis performed in the “Lon- Neubauer and Rostasy 1997; Brosens and Van Gemert 1999;
gitudinal Strain across the Composite Width” section. For speci- Chen and Teng 2001; Wu and Jiang 2013; Lu et al. 2005) (previ-
mens with a 15 mm-wide strip, the presence of a mixed ously described in the “Background on the Width Effect” section)
Mode-II–Mode-III condition entailed for a high amount of energy that define the load-carrying capacity of the FRP–concrete interface
needed to propagate the crack. When the ratio bf/b increases, the as a function of the ratio bf/b. The choice of using the nominal
area across the FRP width characterized by a pure Mode II condi- width to divide the load in Figs. 11(b and c) is related to the design
tion becomes larger than the region with a mixed mode condition, implications that can be derived from these plots. The curves cor-
causing a decrease of the value of Pcrit per unit width that tends to responding to the models from literature were plotted using
an approximately constant value when the ratio bf/b becomes equal Young’s modulus of the fibers, Ef, and the equivalent thickness
to or higher than 0.33. Results plotted in Fig. 11(a) indicate that re- of the fibers, t*f ,A , instead of the values referred to the composite,
gardless of the width of the composite strip, specimens with the that is, EFRP and tFRP, respectively. It can be observed that most
composite strip applied to the B face reached higher values of of the models are not able to predict the experimental behavior of
Pcrit than specimens with the steel FRP strip applied to the S face. this campaign. It is worth noting that the majority of the models
Fig. 11(a) includes the results of three S-face specimens with a proposed in literature are based on experimental data in which
width equal to 50 mm (bf/b = 0.33) tested at a rate equal to 10 times the testing conditions are not homogeneous. For example, in

© ASCE 04020024-15 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Chen and Teng (2001) the database used to calibrate the predictive observation can be made if the face of application of the composite
model considered specimens with different test rates and specimens strip is considered. Fig. 12(c) shows the longitudinal strains across
for which the face of application of the composite strip was not re- the width for specimen DS_300_50_HD_D_S_1. For this speci-
ported. Among the different predictive models shown in Fig. 11(b), men, the width of the region with constant longitudinal strains is
only the model proposed by Lin et al. (2017) is in relatively good equal to 30 mm, that is, the same value observed for specimen
agreement with the experimental data of S-face specimens. DS_300_50_HD_D_B_2. It appears that both the stiffness of the
It should be noted that Eq. (13) implies that the width factor composite strip and the face of application of the FRP strip do
is influenced by the product between Young’s modulus and the not affect the strain profile across the width of the strip, and there-
thickness of the composite. The influence of the stiffness of the fore the width factor should not be influenced by these parameters.
composite strip on the width factor should entail for a different dis- Based on this experimental evidence, the authors propose a new
tribution of ɛyy across the width of the composite strip when differ- formula for the width factor, which is obtained by fitting the exper-
ent densities of the steel fiber sheet are employed. Figs. 12(a and b) imental results reported in Fig. 11(b). The load-carrying capacity
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

provide a comparison between the profile of ɛyy across the per unit width was evaluated through Eq. (17) introducing the re-
composite width for specimens DS_300_50_HD_D_B_2 vised width factor kw
and DS_300_50_UHD_D_B_5, respectively. Specimen ΓF 
DS_300_50_UHD_D_B_5 is part of the experimental campaign Ptheor
= kw 2ΓF Ef t*f ,A (22)
presented in Carloni et al. (2017a) and it was strengthened with a bf
ultra-high density (UHD) steel fiber sheet, characterized by a
Young’s modulus of the fibers, Ef, and an equivalent thickness,
kw = 1 + (1 − bf /b)0.5 /(1 + 0.01bU
f ) (23)
t*f ,UHD , equal to 190 MPa and 0.381 mm, respectively. Fig. 12
shows that, regardless of the stiffness of the composite strip, the ΓF is the value of the fracture energy evaluated as
region of the steel FRP strip characterized by a pure Mode-II con-

dition (see “Longitudinal Strain across the Composite Width” sec-
ΓF = α f ′ c fctm (24)
tion) has always the same width for a 50 mm-wide strip. A similar

(a) (b)

(c)

Fig. 12. Profile of ɛyy across the width of the composite strip for specimens (a) DS_300_50_HD_D_B_2; (b) DS_300_50_UHD_D_B_5; and
(c) DS_300_50_HD_D_S_1.

© ASCE 04020024-16 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


where U and α = parameters obtained from a best fit procedure of the strip is involved in the shear transfer. It is possible that when
the experimental data (Fig. 11(c)). Instead of using GF, the symbol the width of the composite strip is equal or very close to the width
ΓF is adopted to indicate the fracture energy, since the value deter- of the concrete prism, Eq. (17) needs to be modified or it is not ap-
mined in Eq. (24) is obtained from an empirical formula and not plicable at all. The difference between points W and V and the plot of
from experimental tests. Similarly, the load-carrying capacity is in- function (22) is the reason why ΓF and G  F are different.
ΓF
dicated as Ptheor in Eq. (22) since it is evaluated using ΓF. Eq. (22) is
used to fit the experimental data of S-face specimens and B-face
specimens separately, since they show a different trend. The values Conclusions
of U and α for S-face specimens are equal to 1.78 and 0.0698, re-
spectively, while for B-face specimens are equal to 1.75 and 0.102, In this paper, direct single-lap shear tests were performed to deter-
respectively. The trend of S-face specimens has been then fitted mine the load-carrying capacity of steel FRP–concrete joints. Tests
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

again, keeping the parameter U equal to the value determined for were performed varying the bonded width of the composite strip
B-face specimens (1.75). This choice is related to the observation in order to investigate the influence of the bf/b ratio on the load-
that the main fracture parameter that differs between S-face speci- carrying capacity. The face of the concrete prism to which the com-
mens and B-face specimens is the fracture energy, GF, while results posite strip was applied was also varied in order to study the effect of
of Fig. 12 indicate that the width factor should be independent of concrete surface quality on the bond behavior. Three specimens were
the face to which the composite strip is applied. A value of α tested at a rate that was 10 times higher than the rate used for the re-
equal to 0.0685 is obtained from this new fitting for S-face speci- maining specimens. DIC was employed to study the longitudinal
mens (U is kept equal to 1.75). Fig. 11(c) shows the comparison strain profile on the surface of the composite strip during the debond-
between the two different fittings made for S-face specimens. It ing phenomenon. Three different functions were used to fit the ex-
can be noted that the two fittings are almost identical, therefore perimental strain profile. The following observations can be made.
the last one (U = 1.75 and α = 0.0685) is chosen to represent the be- 1. The load-carrying capacity per unit width is affected by the
havior of S-face specimens, in order to maintain the same width width of the composite strip. When the bf/b ratio is lower than
factor both for S-face and B-face specimens. Since the FRP shear 0.33, the load-carrying capacity per unit width tends to decrease
reinforcements are usually applied to the side faces of a concrete for increasing bf/b ratios. For bf/b ratios equal to or higher than
structural element (e.g., beam), the predictive formula obtained 0.33, the load-carrying capacity per unit width is nominally
for S-face specimens (U = 1.75 and α = 0.0685) shall be used to de- constant.
sign the load-carrying capacity of FRP composites used for shear 2. The fracture energy GF is not affected by the bf/b ratio, while it
strengthening. On the other hand, since the FRP flexural reinforce- varies as the composite is bonded to different faces of the con-
ments are usually applied to the bottom face of a structural element, crete prism.
the predictive formula obtained for B-face specimens (U = 1.75 and 3. The effective bond length seems not to be affected by the bf/b
α = 0.102) shall be used to design the load-carrying capacity of ratio and by the face of the concrete prism to which the compos-
FRP composites used for flexural strengthening. The value of ΓF ite strip is applied.
for S-face and B-face specimens resulted equal to 479 and 703 4. The loading rate affects the load-carrying capacity of the steel
N/mm. These values are slightly higher than the average values FRP–concrete joint and therefore should be always reported
of G F , obtained through the polynomial strain function considering in order to compile a database.
all the different widths [red dotted lines in Figs. 10(a and b)], that 5. A new formula for the width factor is proposed. Different coef-
resulted equal to 364 and 589 N/mm for S-face and B-face speci- ficients are proposed for the width factor that depends on
mens, respectively. whether the steel FRP strip is used for shear strengthening or
Eq. (22) is characterized by an asymptotic behavior that levels flexural strengthening.
off when the ratio bf/b is greater than 0.4. Using Eq. (17), it is pos-
sible to predict the value of the load-carrying capacity per unit
width when the ratio bf/b is equal to 1 for both S-face and B-face Acknowledgments
specimens [point W in Fig. 11(c)]. To this end, the value of GF em-
ployed in Eq. (17) is set equal to the average of the values of G F Technicians of the laboratory LISG (Laboratory of Structural and
determined from the polynomial fitting (Table 5) for all S-face Geotechnical Engineering) at University of Bologna are gratefully
and B-face specimens, respectively. For S-face and B-face speci- acknowledged. Kerakoll S.p.A. of Sassuolo, Italy, is gratefully ac-
mens, the load-carrying capacity per unit width when bf/b = 1 knowledged for providing the composite materials.
from Eq. (17) is lower than the value predicted by Eq. (22). It
should be observed once again that for the evaluation of Ptheor
and GF, the fracture mechanics formulation requires the use of Notation
Young’s modulus of the composite material instead of Young’s
modulus of the fibers. However, if Young’s modulus of the com- The following symbols are used in this paper:
posite referred to the bare fibers (E HD f ,SRP = 258 GPa) is employed Acord = area of a steel cord comprising five steel filaments;
to evaluate the load-carrying capacity per unit width when bf/b = 1 b = width of the concrete prism (Fig. 1);
via Eq. (17), a much higher value is obtained compared with the bf = width of the FRP strip (Fig. 1), which is also the width
one predicted by Eq. (22) [point V in Fig. 11(c)]. This circumstance of the bonded area in single-lap shear tests;
can be in part explained by referring to what was previously ob- bf,actual = actual width of the steel FRP composite obtained as
served, that is, Eq. (17) does not take into account the high gradients the average of three measurements along the strip;
of the longitudinal strain component close to the edges of the com- CF = calibration factor for the fracture energy used in
posite strip. In addition, it should be noted that the analytical evalu- Neubauer and Rostasy (1997), Brosens and Van
ation via Eq. (17) of the load per unit width when bf/b = 1 is valid Gemert (1999), and Brosens (2001);
only under the hypothesis that the fracture process occurs in a thin Dmax = maximum diameter of the aggregates in the concrete
layer of concrete and only a small portion of the substrate beneath mix;

© ASCE 04020024-17 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


Ef = Young’s modulus of the bare fibers;  Y,C,E = average of Pcrit for specimens characterized by the
P crit
EFRP = Young’s modulus of the FRP composite; same width (Y), the same face to which the
f ,SRP = Young’s modulus of the steel FRP composite referred
EHD composite is applied (C), and the same loading rate (E)
to the area of the HD fiber sheet ASTM D3039 (Table 4);
(ASTM 2008); Ptheor = theoretical bond capacity (or load-carrying capacity)
fh = surface tensile strength of concrete [EN 1542 (CEN based on the Mode-II fracture approach proposed in
1999)]; Täljsten (1996) [see Eq. (17)];
fctm = tensile strength of concrete [EN 12390-6 (CEN  theor = theoretical bond capacity when GF = G
P  F . There are
2009)]; three values of P  theor for each specimen corresponding
ff,u = average tensile strength of the steel FRP composite to the three fitting functions of the strain profile used in
strip; Carloni et al. (2017a);
f ′c = cylindrical compressive strength of concrete; Ptheor,WU = symbols used for the maximum bond force (bond
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

f fA,u = tensile strength of the steel fiber sheet provided by the capacity or load-carrying capacity) in Wu and Jiang
manufacturer (Kerakoll 2018). A specifies the density (2013);
of the sheet; Pu = symbols used for the maximum bond force (bond
,u = tensile strength of the high-density steel fiber sheet
f fHD capacity or load-carrying capacity) in Chen and Teng
provided by the manufacturer (Kerakoll 2018); (2001);
g = global slip, that is, average of the readings of the P* = peak load in the response of single-lap shear
LVDT a and b that are mounted at the beginning of the tests;
ΓF
bonded area. An alternative name for g is loaded-end Ptheor = bond capacity expressed in terms of the width factor
slip; bw proposed by the authors and employing Eq. (17)
g1 = value of the global slip that defines the beginning of with GF = ΓF;
the range of values of g used to compute the average of Rcm = cubic compressive strength of concrete [EN 12390-3
the load, which corresponds to Pcrit; (CEN 2001)];
g2 = value of the global slip that defines the end of the s = interfacial slip;
range of values of g used to compute the average of the tFRP = thickness of the FRP composite;
load, which corresponds to Pcrit; tFRP,actual = actual thickness of the steel FRP composite
GF = interfacial fracture energy that corresponds to the area obtained as the average of three measurements along
under the τxy(s) curve; the strip;
G F = average of the fracture energy for one specimen t *f ,HD = equivalent thickness of the high density (HD) steel
obtained using a fitting curve for the strain profile and fiber sheet, which is equal to 0.254 mm;
ten DIC images within the range [g1,g2]; t*f ,UHD = equivalent thickness of the ultrahigh density (UHD)
GFPcrit = fracture energy back-calculated from Eq. (17), steel fiber sheet, which is equal to 0.381 mm;
substituting Ptheor with Pcrit for each specimen; t*f ,A = equivalent thickness of the steel fiber sheet. A
kb = symbol used for the width effect factor in Neubauer specifies the density of the sheet. For example, A =
and Rostasy (1997), Brosens and Van Gemert (1999), HD for high density;
and Brosens (2001); Tu,max = symbols used for the maximum bond force (bond
kw = symbol used for the width effect factor by the authors capacity or load-carrying capacity) in Neubauer and
in this paper; Rostasy (1997) [see Eq. (1)];
kc = parameter used in Brosens (2001) to take into account wc = backward displacement of the concrete prism
the effect of concrete surface condition on the measured by LVDT c;
interfacial fracture energy; wd = backward displacement of the concrete prism
Leff = effective bond length that corresponds to the length of measured by LVDT d;
the fully established stress transfer, that is, the  c = average of wc within the range [g1,g2];
w
minimum bonded length to obtained the bond capacity  d = average of wd within the range [g1,g2];
w
Ptheor; x = Cartesian coordinate along the width of the composite
eff = average of the effective bond length for one specimen
L (Fig. 1);
obtained using a fitting curve for the strain profile and y = Cartesian coordinate along the direction of the fibers
ten DIC images within the range [g1,g2]; of the composite (Fig. 1);
LC−T
eff = effective bond length proposed by Chen and Teng βp = symbol used for the width effect factor in Chen and
(2001); Teng (2001);
l = length of the bonded area; βw = symbol used for the width effect factor in Lu et al.
P = applied load to the FRP strip in single-lap shear (2005);
tests; κ WUw = symbol used for the width effect factor in Wu and
Pcrit = plateau load, which corresponds to the bond capacity Jiang (2013);
obtained experimentally from single-lap shear test as κLINw = symbol used for the width effect factor in Lin et al.
the average of the applied load P within the range [g1, (2017);
g2]; ɛf,u = ultimate deformation of the steel fiber sheet provided
εyy
Pcrit = bond capacity obtained as the integral of the by the manufacturer (Kerakoll 2018);
longitudinal strain over the width of the composite at ɛyy = longitudinal strain in the composite in the direction of
the loaded end (see Eq. 18); the fibers;
εyy εyy
Pcrit = average of the bond capacity Pcrit considering several ɛxy = shear strain in the composite in the plane of the
values of y near the loaded end and five points of the composite;
load response (Table 7); ɛmax = maximum value of ɛyy fitting function;

© ASCE 04020024-18 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


ΓF = fracture energy expressed in terms of the properties of Ceroni, F., and M. Pecce. 2007. “Cracking behaviour of RC beams exter-
concrete (Eq. 24); and nally strengthened with emerging materials.” Constr. Build. Mater.
τxy = interfacial shear stress in debonding Mode-II problems 21 (4): 736–745.
Chen, J. F., and J. G. Teng. 2001. “Anchorage strength models for FRP
that assume a fictitious zero-thickness interface.
and steel plates bonded to concrete.” J. Struct. Eng. 127 (7): 784–
791. https://doi.org/10.1061/(ASCE)0733-9445(2001)127:7(784).
Dai, J., T. Ueda, and Y. Sato. 2005. “Development of the nonlinear bond
References
stress–slip model of fiber reinforced plastics sheet–concrete interfaces
with a simple method.” J. Compos. Constr. 9 (1): 52–62. https://doi
Ali-Ahmad, M., K. Subramaniam, and M. Ghosn. 2006. “Experimental in- .org/10.1061/(ASCE)1090-0268(2005)9:1(52).
vestigation and fracture analysis of debonding between concrete and El-Hacha, R., and M. A. Mashrik. 2012. “Effect of SFRP confinement on
FRP sheets.” J. Eng. Mech. 132 (9): 914–923. https://doi.org/10.1061 circular and square concrete columns.” Eng. Struct. 36: 379–393.
/(ASCE)0733-9399(2006)132:9(914). Elices, M., V. G. Guinea, and J. Planas. 1992. “Measurement of the fracture
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

ASTM (American Society for Testing and Materials). 2008. Standard test energy using three-point bend tests: Part 3—influence of cutting the P-δ
method for tensile properties of polymer matrix composite materials. tail.” Mater. Struct. 25 (6): 327–334.
ASTM D3039. West Conshohocken, PA: ASTM International. European Standards. 2004. BS EN 1992-1-1:2004: Eurocode 2. Design of
Barton, B., E. Wobbe, L. R. Dharani, P. Silva, V. Birman, A. Nanni, T.
concrete structures. General rules and rules for buildings. UK: British-
Alkhrdaji, J. Thomas, and G. Tunis. 2005. “Characterization of rein-
Adopted European Standard.
forced concrete beams strengthened by steel reinforced polymer and
Ghorbani, M., D. Mostofinejad, and A. Hosseini. 2017. “Experimental in-
grout (SRP and SRG) composites.” Mater. Sci. Eng., A 412 (1–2):
vestigation into bond behavior of FRP-to-concrete under mixed-mode I/
129–136.
II loading.” Constr. Build. Mater. 132: 303–312.
Baž ant, Z. P., and J. Planas. 1997. Vol. 16 of Fracture and size effect in
Grande, E., M. Imbimbo, and E. Sacco. 2015. “Investigation on the bond
concrete and other quasibrittle materials. London: CRC Press.
behavior of clay bricks reinforced with SRP and SRG strengthening
Bencardino, F., and A. Condello. 2015. “SRG/SRP–concrete bond–slip
systems.” Mater. Struct. 48 (11): 3755–3770.
laws for externally strengthened RC beams.” Compos. Struct. 132:
Hawileh, R., J. Abdalla, W. Nawaz, A. Alzeer, R. Muwafi, and A. Faridi.
804–815.
2014. Strengthening reinforced concrete beams in flexure using hard-
Benvenuti, E., N. Orlando, D. Ferretti, and A. Tralli. 2016. “A new 3D ex-
wire steel fiber sheets. Vancouver, BC, Canada: CICE.
perimentally consistent XFEM to simulate delamination in
Hillerborg, A. 1985. “The theoretical basis of a method to determine the
FRP-reinforced concrete.” Composites Part B 91: 346–360.
fracture energy GF of concrete.” Mater. Struct. 18 (4): 291–296.
Brosens, K. 2001. “Anchorage of externally bonded steel plates and CFRP
Holzenkämpfer, P. 1994. “Ingenieurmodelle des Verbunds geklebter
laminates for the strengthening of concrete elements.” Dissertation,
Dept. of Civil Engineering, Katholieke, Univ. Leuven. Bewehrung für Betonbauteile.” Ph.D. dissertation, Technische Univ.
Brosens, K., and D. Van Gemert. 1999. “Anchorage design for externally of Braunschweig.
bonded carbon fiber reinforced polymer laminates.” ACI Spec. Publ. Hoover, C. G., and Z. P. Baž ant. 2013. “Comprehensive concrete fracture
188: 635–646. tests: Size effects of types 1 & 2, crack length effect and postpeak.” Eng.
Carloni, C., M. Santandrea, and I. A. O. Imohamed. 2017a. “Determination Fract. Mech. 110: 281–289.
of the interfacial properties of SRP strips bonded to concrete and com- Kerakoll. 2018. Accessed May 2, 2018. http://www.kerakoll.com.
parison between single-lap and notched beam tests.” Eng. Fract. Mech. Kim, Y. J., A. Fam, A. Kong, and R. El-Hacha. 2005. “Flexural strength-
186: 80–104. ening of RC beams using steel reinforced polymer (SRP) composites.”
Carloni, C., M. Santandrea, and R. Wendner. 2017b. “An investigation on ACI Spec. Publ. 230: 1647–1664.
the ‘width and size effect’ in the evaluation of the fracture energy of Lin, J. P., Y. F. Wu, and S. T. Smith. 2017. “Width factor for externally
concrete.” Proc. Struct. Integrity 3: 450–458. bonded FRP-to-concrete joints.” Constr. Build. Mater. 155: 818–829.
Carloni, C., and K. V. Subramaniam. 2010. “Direct determination of cohe- Lu, X. Z., J. G. Teng, L. P. Ye, and J. J. Jiang. 2005. “Bond–slip models for
sive stress transfer during debonding of FRP from concrete.” Compos. FRP sheets/plates bonded to concrete.” Eng. Struct. 27 (6): 920–937.
Struct. 93 (1): 184–192. Mazzotti, C., M. Savoia, and B. Ferracuti. 2008. “An experimental study on
Carloni, C., S. Verre, L. H. Sneed, and L. Ombres. 2017c. “Loading rate delamination of FRP plates bonded to concrete.” Constr. Build. Mater.
effect on the debonding phenomenon in fiber reinforced cementitious 22 (7): 1409–1421.
matrix-concrete joints.” Composites Part B 108: 301–314. Mitolidis, G. J., T. N. Salonikios, and A. J. Kappos. 2012. “Tests on RC
Carrara, P., and D. Ferretti. 2013. “A finite-difference model with mixed beams strengthened at the span with externally bonded polymers rein-
interface laws for shear tests of FRP plates bonded to concrete.” forced with carbon or steel fibers.” J. Compos. Constr. 16 (5): 551–
Composites Part B 54: 329–342. 562. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000281.
Carrara, P., D. Ferretti, F. Freddi, and G. Rosati. 2011. “Shear tests of car- Napoli, A., and R. Realfonzo. 2015. “Reinforced concrete beams strength-
bon fiber plates bonded to concrete with control of snap-back.” Eng. ened with SRP/SRG systems: Experimental investigation.” Constr.
Fract. Mech. 78 (15): 2663–2678. Build. Mater. 93: 654–677.
Casadei, P., A. Nanni, T. Alkhrdaji, and J. Thomas. 2005. “Performance of Napoli, A., and R. Realfonzo. 2016. “Compressive behavior of concrete
double-T prestressed concrete beams strengthened with steel reinforce- confined by SRP wraps.” Constr. Build. Mater. 127: 993–1008.
ment polymer.” Adv. Struct. Eng. 8 (4): 427–442. Napoli, A., G. de Felice, S. De Santis, and R. Realfonzo. 2016a. “Bond be-
CEN (European Committee for Standardization). 1999. Products and sys- haviour of steel reinforced polymer strengthening systems.” Compos.
tems for the protection and repair of concrete structures. Test methods. Struct. 152: 499–515.
measurement of bond strength by pull-off. EN 1542. Brussels, Belgium: Napoli, A., R. Realfonzo, M. Petracca, F. Candeloro, G. Camata, and P.
CEN. Casadei. 2016b. “Flexural strengthening of RC slabs with SRP/SRG: An
CEN (European Committee for Standardization). 2001. Testing hardened experimental-numerical comparison.” Appl. Mech. Mater. 847: 381–390.
concrete—Part 3: Compressive strength of test specimens. EN Neto, P., J. Alfaiate, and J. Vinagre. 2016. “Assessment of the dependence
12390-3. Brussels, Belgium: CEN. of CFRP-concrete behaviour on the width of the bonded materials.”
CEN (European Committee for Standardization). 2009. Testing hardened Composites Part B 91: 448–457.
concrete—Part 6: Tensile splitting strength of test specimens. EN Neubauer, U., and F. S. Rostasy. 1997. “Design aspects of concrete struc-
12390-6. Brussels, Belgium: CEN. tures strengthened with externally bonded CFRP-plates.” In Proc., 7th
Ceroni, F., M. Ianniciello, and M. Pecce. 2016. “Bond behavior of FRP car- Int. Conf. on Structural Faults and Repair, 109–118. Edinburgh,
bon plates externally bonded over steel and concrete elements: Scotland: ECS Publications.
Experimental outcomes and numerical investigations.” Composites Pecce, M., F. Ceroni, A. Prota, and G. Manfredi. 2006. “Response predic-
Part B 92: 434–446. tion of RC beams externally bonded with steel-reinforced polymers.”

© ASCE 04020024-19 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024


J. Compos. Constr. 10 (3): 195–203. https://doi.org/10.1061/(ASCE) Sneed, L. H., S. Verre, C. Carloni, and L. Ombres. 2016. “Flexural behav-
1090-0268(2006)10:3(195). ior of RC beams strengthened with steel-FRCM composite.” Eng.
Prota, A., K. Y. Tan, A. Nanni, M. Pecce, and G. Manfredi. 2006. Struct. 127: 686–699.
“Performance of shallow reinforced concrete beams with externally Subramaniam, K. V., C. Carloni, and L. Nobile. 2007. “Width effect in the
bonded steel-reinforced polymer.” ACI Struct. J. 103 (2): 163. interface fracture during shear debonding of FRP sheets from concrete.”
Rabinovitch, O. 2008. “Debonding analysis of fiber-reinforced-polymer Eng. Fract. Mech. 74 (4): 578–594.
strengthened beams: Cohesive zone modeling versus a linear elastic Subramaniam, K. V., C. Carloni, and L. Nobile. 2011. “An understand-
fracture mechanics approach.” Eng. Fract. Mech. 75 (10): 2842–2859. ing of the width effect in FRP–concrete debonding.” Strain 47 (2):
Rabinovitch, O. 2012. “Dynamic debonding in concrete beams strengthened 127–137.
with composite materials.” Int. J. Solids Struct. 49 (26): 3641–3658. Täljsten, B. 1996. “Strengthening of concrete prisms using the plate-
Rostasy, F. S., and U. Neubauer. 1997. “Bond behaviour of CFRP-laminates bonding technique.” Int. J. Fract. 82 (3): 253–266.
for the strengthening of concrete members.” In Proc., Int. Conf. of Teng, J. G., S. T. Smith, J. Yao, and J. F. Chen. 2003. “Intermediate crack-
Composite Construction-Conventional and Innovative, 717–722. induced debonding in RC beams and slabs.” Constr. Build. Mater.
Downloaded from ascelibrary.org by CASE WESTERN RESERVE UNIV on 04/24/20. Copyright ASCE. For personal use only; all rights reserved.

Zurich: IABSE. 17 (6–7): 447–462.


Salomoni, V., G. Mazzucco, C. Pellegrino, and C. Majorana. 2011. Wobbe, E., P. Silva, B. L. Barton, L. R. Dharani, V. Birman, A. Nanni, T.
“Three-dimensional modelling of bond behaviour between concrete Alkhrdaji, J. Thomas, and G. Tunis. 2004. “Flexural capacity of RC
and FRP reinforcement.” Eng. Comput. 28 (1): 5–29. beams externally bonded with SRP and SRG.” In Proc., Society for
Santandrea, M., I. A. O. Imohamed, C. Carloni, C. Mazzotti, S. de Miranda, the Advancement of Material and Process Engineering 2004
and F. Ubertini. 2016. “An investigation of the debonding mechanism Symposium, 16–20. Long Beach, CA: Society for the Advancement of
in steel FRP- and FRCM-concrete joints.” In Proc., 4th Workshop on Material and Process Engineering.
the New Boundaries of Structural Concrete, 289–298. Italy: Wu, Y. F., and C. Jiang. 2013. “Quantification of bond-slip relationship for
University of Naples Federico II - ACI Italy Chapter, Edizioni Imready. externally bonded FRP-to-concrete joints.” J. Compos. Constr. 17 (5):
Sneed, L. H., F. Ravazdezh, M. Santandrea, I. A. O. Imohamed, and C. 673–686. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000375.
Carloni. 2017. “A study of the compressive behavior of concrete col- Xu, T., Z. J. He, C. A. Tang, W. C. Zhu, and P. G. Ranjith. 2015. “Finite
umns confined with SRP jackets using digital image analysis.” element analysis of width effect in interface debonding of FRP plate
Compos. Struct. 179: 195–207. bonded to concrete.” Finite Elem. Anal. Des. 93: 30–41.

© ASCE 04020024-20 J. Compos. Constr.

J. Compos. Constr., 2020, 24(4): 04020024

You might also like