Available online at www.sciencedirect.

com

Composites: Part B 39 (2008) 258–270

www.elsevier.com/locate/compositesb

Interfacial bond strength of glass fiber reinforced polymer bars

in high-strength concrete
J.-Y. Lee a, T.-Y. Kim a, T.-J. Kim a, C.-K. Yi b,*
, J.-S. Park c, Y.-C. You c, Y.-H. Park c

a
Department of Architectural Engineering, Sungkyunkwan University, 300, Cheoncheon-dong, Jangan-gu, Suwon 440-746, South Korea
b
Department of Civil, Environmental and Architectural Engineering, Korea University, Anam-dong, Seongbuk-gu, Seoul 136-713, South Korea
c
Building Research Department, Korea Institute of Construction Technology, 2311, Daehwa-dong, Ilsan-gu, Goyang 411-712, South Korea

Received 29 August 2006; accepted 3 March 2007

Available online 26 July 2007

Abstract

This paper presents the results of an experimental study on the interfacial bond strength of glass fiber reinforced polymer (GFRP)
bars in high-strength concrete cube. The experimental program consisted of testing 54 concrete cube specimens prepared according to
CSA S802-02 standard. Two main parameters were considered in the experimental investigation: the compressive strength of concrete
(from 25.6 MPa to 92.4 MPa) and the type of rebar (steel, sand-coated GFRP, and helically wrapped GFRP). The test results showed
that the interfacial bond strength of the GFRP bars increased as the compressive strength of concrete increased. However, the increasing
rate of the bond strength of the GFRP bars with respect to the concrete strength was much smaller than that of the steel bars. The con-
crete specimens were sawn in half after the test for a closer investigation of the actual mode of bond failure. Visual examination of
the specimens showed that bond failure of the steel bar was caused by concrete crushing against the face of the ribs, while bond failure
of the GFRP bars occurred not only in the concrete but also in the bars by delamination of the resin-rich outer layer from the fiber core.
The average area of the delaminated resin-rich layer of the GFRP bar increased with increasing compressive strength of concrete.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: A. Glass fiber; B. Adhesion; C. Interface/interphase; Interfacial bond strength

1. Introduction strength-to-weight ratio. Various structural applications of

FRP bars have been implemented in the areas of severe
Although steel bars remain as the most widely used type environment and shown satisfactory performance in service
of concrete reinforcement in construction industry, struc- thus far [1].
tures exposed to highly aggressive environments, such as Good performance of reinforced concrete requires ade-
sea water and deicing salts, have shown that steel reinforce- quate interfacial bond between the reinforcing material
ments accelerate concrete deterioration due to expansive and the concrete because the load applied must be trans-
corrosion. Increasing concern on durability of the rein- ferred from the matrix to the reinforcement. Although
forced concrete structure has encouraged researchers to the interfacial bond behavior of the steel bars in concrete
seek alternative type of reinforcement whose physicochem- has been established and clearly addressed in various
ical properties surpass those of the conventional steel bars. design codes, direct application of the codes to FRP bars
For last two decades, the use of fiber-reinforced polymer would be imprudent act because the interfacial bond
(FRP) bars has been gaining popularity in the civil engi- behavior of FRP bars to concrete is expected to vary from
neering community for their non-corrosive nature and high that of conventional steel bars. The variation arises from
the fact that FRP bars have different key parameters that
*
Corresponding author. Tel.: +82 2 3290 3329; fax: +82 2 928 7656. influence bond performance from those of conventional
E-mail address: chongku@skku.edu (C.-K. Yi). steel bars, such as surface condition of the bars, modulus

1359-8368/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesb.2007.03.008
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 259

Table 1
Composition and characteristics of concrete
fc0 ðMPaÞ w/c (%) Fine/coarse Slum Cement Water Fine aggregate Coarse aggregate Silica fume Air content
aggregate (%) (mm) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (%)
25.6 53 40 131 319 172 868 945 – –
35.3 45 40 123 394 177 697 1040 – –
40.6 55 43 125 400 220 745 995 – –
56.3 40 35 120 523 209 569 1040 53 –
75.7 32 38 115 530 187 629 1026 53 10.6 (2%)
92.4 29 38 120 530 169 629 1026 53 15.9 (3%)

of elasticity, shear stiffness, and tensile strength. Hence, ables included types of rebars (AFRP, CFRP, GFRP,
there is need to establish clear understanding of the interfa- and steel) and compressive strength of concrete (29–
cial bond behavior of FRP bars in concrete. 60 MPa). Based on the experimental results, Okelo et al.
Many studies have been conducted to investigate the proposed that the average bond strength of the FRP bars
interfacial bond behavior of FRP bars. In particular, the was proportional to 1/2 power of the concrete compressive
effects of the shape of the bar cross section [2,3], surface strength (29–60 MPa).
texture [4–11], bar deformations [12], and elastic modulus This paper presents the results of an experimental study
of the bars [13,14] on the interfacial bond in FRP compos- on the bond performance of two types of GFRP bars and
ites have been reported. Furthermore, attempts to formu- steel bars in high-strength concrete. The experimental pro-
late analytical models to predict the interfacial bond gram consisted of testing 150 mm concrete cube specimens
behavior of FRP bars were made by Cosenza et al. [15], prepared according to CSA S806-02 standard [18]. Six
Achillides et al. [13]. different concrete mix designs were prepared to produce
However, detailed investigations available on the inter- compressive strengths of 25.6 MPa, 35.3 MPa, 40.6 MPa,
facial bond behavior of FRP bars in high-strength concrete 56.3 MPa, 75.7 MPa, and 92.4 MPa. Applied load and slip
have been relatively limited and, to a certain degree, con- were recorded during pullout tests to evaluate bond stress–
troversial. Achillides [13] studied the effect of concrete slip behavior of the bars. The test results were then com-
strength on the bond strength of EUROCRETE bars and pared with some of the existing empirical models available
noted that, for concretes with compressive strengths in the technical literature.
greater than 30 MPa, the bond strength of the FRP bar
is not controlled much by concrete strength but appears
to be influenced by the interlaminar shear strength just 2. Test program
below the resin rich surface layer of the bar. On the other
hand, in an investigation by Shima et al. [16], the bond 2.1. Materials
strength of CFRP bar was proportional to between 1/2
and 2/3 power of the concrete compressive strength in 2.1.1. Concrete
the range of 38–84 MPa. Similarly, Okelo and Yuan [17] Normal- and high-strength concrete specimens were pre-
tested 151 specimens containing 6, 8, 10, 16, 19 mm rebars pared in the laboratory. The composition of concrete and
embedded in a 203 mm concrete cube. Experimental vari- the average concrete strength of each specimen at the time

100
Compressive stress of concrete (MPa)

80

60

40 25.6MPa
35.3MPa
40.6MPa
20 56.3MPa
75.7MPa
92.4MPa
0
0 0.001 0.002 0.003 0.004 0.005
Compressive strain of concrete
Fig. 1. Compressive stress–strain curves of concrete. Fig. 2. Surface deformations of GFRP rebars.
260 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

Table 2
Material properties of rebars
Fiber/rebar D (mm) A (mm2) Fiber content (%) Specific gravity (g/cm3) Surface fy (MPa) ffy (GPa) Es (GPa)
Steel 12.7 126.7 – 7.90 – 410 560 200.0
GFRP-SC 12.7 129.0 70 2.04 Sand coating – 690 42.0
GFRP-HW 12.7 129.0 70 2.00 Helically wrapped – 617 40.8

12.7mm

810mm
150mm

150mm
50.8mm
Fig. 3. Test specimen.

150mm

Thickness
6mm

150mm
A A
Thickness
10mm

Plan View Section A-A

Fig. 4. Mold for bond test specimens for vertical bars.

Table 3
Test specimens
Rebars Compressive strength of concrete (MPa)
25.6 35.3 40.6 56.3 75.7 92.4
Steel 1 F25A1 F35A1 F40A1 F55A1 F75A1 F90A1
2 F25A2 F35A2 F40A2 F55A2 F75A2 F90A2
3 F25A3 F35A3 F40A3 F55A3 F75A3 F90A3
GFRP-SC 1 F25B1 F35B1 F40B1 F55B1 F75B1 F90B1
2 F25B2 F35B2 F40B2 F55B2 F75B2 F90B2
3 F25B3 F35B3 F40B3 F55B3 F75B3 F90B3
GFRP-HW 1 F25C1 F35C1 F40C1 F55C1 F75C1 F90C1
2 F25C2 F35C2 F40C2 F55C2 F75C2 F90C2
3 F25C3 F35C3 F40C3 F55C3 F75C3 F90C3
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 261

LVDT

Steel plate

Concrete
cube

Spherical seat
Testing
machine

2 LVDTs
FRP rods

Grip

Fig. 5. Test setup.

of parent concrete cylinder test are given in Table 1 and behavior up to failure. The material properties of the
Fig. 1, respectively. The nominal coarse aggregate size rebars are shown in detail in Table 2.
was limited to 13.0 mm to ease the placement of concrete
inside the small formwork. The concrete was mixed in a 2.2. Test specimens
0.06 m3 effective capacity rotating mixer. The fine and
coarse aggregates were first dry-mixed together, and then A total of 54 cube specimens were tested. The 150 mm
the cement was added slowly to ensure thorough mix. cube specimens were prepared according to CSA S806-02
Finally, water was added and allowed to mix until good standard [18]. Each specimen consisted of a concrete cube
uniformity was obtained. The concrete cylinder tests were with a single rebar embedded vertically along a central axis.
carried out according to ASTM C39, Standard Test The bonded length of the rebar was set to 50.8 mm, four
Method for Compressive Strength of Cylindrical Concrete times the diameter of the rebars. In order to minimize effect
Specimens. Three cylinders were tested for each design mix, of the stress from the loading plate, the bar was sheathed
and the average concrete compressive strengths ranging with soft PVC tube to prevent bonding between the bar
from 25.2 MPa to 93.2 MPa were obtained. and concrete on the loaded end side (see Fig. 3). Concrete
was placed in four layers of approximately equal thickness,
2.1.2. Bars and each layer was rodded 25 times with 16 mm diameter
Two types of GFRP bars and one type of steel bar sup- tamping rod. The concrete was cast with the rebar in the
plied by international manufacturers were used. The GFRP vertical position inside the steel formwork that was pre-
rebars are made of continuous longitudinal glass fibers pared in accordance with ASTM Standard C234 [19] as
glued together with a thermosetting resin. The nominal shown in Fig. 4. After molding, the specimens were initially
diameter of the bars was 12.7 mm (#4). The surface of cured by covering them with plastic sheet, which prevented
the GFRP bars was treated to improve bond by sand coat- moisture loss for 24 h. Immediately after the removal of the
ing (GFRP-SC, by Pultrall Inc.) and helical wrapping with molds, specimens were cured in accordance with ASTM
sand coating (GFRP-HW, by Hughes Brothers Inc.) as Standard C511 [20] until the time of test. During this cur-
shown in Fig. 2. The fiber volume fraction and density of ing period, they were sprayed with water two times a day to
all GFRP rebars were 70% and approximately 2.0 g/cm3, maintain moisture on the surfaces at all times. Two param-
respectively. The fracture strength of GFRP bars ranged eters were considered in this investigation: the type of
from 617 MPa to 690 MPa, and the elastic modulus was rebars (steel or two types of GFRP bars) and compressive
200 GPa for the steel and 40.8–42.0 GPa for the GFRP strength of concrete. Three nominally identical specimens
bars. All of the GFRP rebars followed a linear stress–strain were prepared for each specimen type as shown in Table 3.
262 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

2.3. Test method and measurements Table 4

Interfacial bond strength and corresponding slip
The pullout tests were performed by using a UTM with Specimens sb,max (MPa) Sm (mm)
a capacity of 1000 kN. Load was applied to the rebar at a F25 series F25A1 24.59 0.66
rate of about 20,000 N/min. The specimen was mounted on F25A2 22.77 0.56
the testing machine such that the face of the cube with long F25A3 24.79 0.47
F25B1 20.40 0.31
end of the bar was in contact with the bearing block assem- F25B2 18.70 0.47
bly. A spherically seated bearing block was manufactured F25B3 20.17 0.40
to transfer the reaction from the block to the weighting F25C1 17.67 5.50
table of the testing machine without exerting transverse F25C2 20.41 4.70
load on the bar. The projecting rebar was placed through F25C3 18.59 3.71
the bearing block assembly, and the bar end was gripped F40 series F40A1 25.36 0.83
for tension by the jaws, with round wedges, of the testing F40A2 28.49 1.08
F40A3 27.44 1.08
machine as shown in Fig. 5.
F40B1 19.34 0.58
The load was measured with the electronic load cell of F40B2 23.60 0.25
the machine. The slips of the rebar relative to concrete at F40B3 20.59 0.43
the loaded end and at the free end were measured with F40C1 22.62 4.83
three linear displacement transducers (LVDTs). The read- F40C2 20.87 5.10
F40C3 20.99 4.60
ings of the applied load and the corresponding LVDTs
were recorded automatically through a data logger at F75 series F75A1 37.53 1.63
specified load intervals. The test was terminated when F75A2 35.50 –
F75A3 34.90 –
the value of the applied load dropped to about 85% of
F75B1 23.36 0.09
the maximum-recorded load in the post-peak descending F75B2 24.67 0.08
branch. F75B3 25.20 0.15
F75C1 23.03 3.53
3. Test results F75C2 25.00 3.57
F75C3 23.02 4.81

All of the specimens failed by pullout of the rebars. F35 series F35A1 26.65 1.13
None of ‘‘failure at the anchorage,’’ ‘‘rupture of the rebar,’’ F35A2 24.85 1.03
F35A3 26.20 1.02
and ‘‘splitting of the enclosing concrete’’ was observed. The F35B1 23.92 0.30
interfacial bond strength and the corresponding free end F35B2 – –
slip are shown in Table 4. The test results of specimen F35B3 19.09 0.68
F35B2 could not be recorded due to errors in the data log- F35C1 19.20 7.00
ger. The letter ‘‘F’’ was used to designate the compressive F35C2 19.70 3.00
F35C3 20.28 5.70
strength of concrete while the letters ‘‘A, B, C’’ were used
to designate the type of rebar (A – deformed steel, B – sand F55 series F55A1 29.28 0.72
F55A2 30.28 0.74
coated GFRP, and C – GFRP with deformations in helical
F55A3 31.31 1.08
pattern). The digits following the letter ‘‘F’ imply the com- F55B1 22.15 0.21
pressive strength of concrete while the digit after the letter F55B2 19.39 0.39
‘‘A’’, ‘‘B’’, ‘‘C’’ indicates the specimen number in each set F55B3 22.63 0.43
of specimens. For example, F90B2 denotes the second F55C1 21.45 5.15
F55C2 18.51 6.81
specimen with the sand coated GFRP bar in 92.4 MPa
F55C3 19.90 3.09
compressive strength concrete.
Failure was defined at the point of maximum pullout F90 series F90A1 39.96 –
F90A2 39.20 0.40
load during the test. Corresponding maximum nominal F90A3 40.40 1.05
bond stress and slip values were then determined as those F90B1 23.38 0.10
values occurring at the point of failure. The test results gen- F90B2 25.44 0.10
erally indicated that the interfacial bond strengths of the F90B3 26.20 0.13
steel bar and the GFRP bars increased as the compressive F90C1 25.05 0.38
F90C2 28.05 0.35
strength of concrete increased. However, the increase in F90C3 25.20 0.50
bond strength of the GFRP bars with respect to concrete
strength was smaller than that of the steel bar. The post-
peak interfacial bond behavior of rebars also varied with
the type of rebar. The bond stress of the steel and the 3.1. Interfacial bond stress vs. slip curves of the steel bars
GFRP-SC bars dropped rapidly after attaining the maxi-
mum bond stress, while GFRP-HW exhibited a gradual Fig. 6 shows the interfacial bond stress vs. free end slip
reduction of bond stress. curves of the steel bars in normal- and high-strength
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 263

concrete. The observed bond behavior of the steel bar was eventually form shear failure plane and lead to the pullout
similar to that obtained from the pull-out tests by Elige- failure of the bar.
hausen et al. [21]. At the beginning of loading, no measur- The influence of concrete strength on the interfacial
able slip was observed, and it was attributed to efficient bond–slip relationship is demonstrated in Fig. 6e and f.
bond through chemical adhesion between the steel bar These figures show that the interfacial bond resistance
and the concrete. As the load increased, the slip became is enhanced considerably in high-strength concrete com-
measurable with developments of transverse and longitudi- pared with normal strength concrete. For example, the
nal cracks in the concrete around the bars. The cracks average bond strength in the normal strength concrete

Fig. 6. Interfacial bond stress vs. slip curves of test specimens (a) F25 series, (b) F35 series, (c) F40 series, (d) F55 series, (e) F75 series, and (f) F90 series.
264 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

ðfc0 ¼ 25:6 MPaÞ is 23.95 MPa, while that in high-strength

concrete ðfc0 ¼ 92:4 MPaÞ is 39.85 MPa. The test results
also show that the average bond strength increases with
the concrete compressive strength in the range considered
for this study.

3.2. Interfacial bond stress vs. slip curves of the sand coated
GFRP bars

The curves of the interfacial bond stress vs. free end slip
of the sand coated GFRP bars (GFRP-SC) for different
compressive strengths of concrete are shown in Fig. 6.
The maximum bond stress and the corresponding slip of
the GFRP-SC bars were smaller than those of the steel
bars. After attaining the maximum bond stress, the
GFRP-SC bars show dramatic decrease in the bond stress
as seen in the case of steel bars. However, the interfacial
bond strength of the GFRP-SC bars augments slightly, rel-
ative to the steel bars, with increase in the compressive
strength of concrete. The slip corresponding to the maxi-
mum bond strength of the GFRP-SC bars also show little
enhancement as the concrete strength increased.

3.3. Interfacial bond stress vs. slip curves of the helically

wrapped GFRP bars

Fig. 6 shows the curves of the interfacial bond stress vs.

free end slip of the helically wrapped GFRP bars (GFRP-
HW) for different compressive strengths of concrete. In
normal strength concrete, the bond stress of the GFRP-
HW bars continues to rise after breakage of initial chemical
adhesion and then reaches the peak at approximately 4 mm
of the free end slip, while the bond stress in high-strength
concrete reaches the peak in less than 1/2 mm of the free
end slip. Comparing the interfacial bond behaviors of the
GFRP-HW with those of the steel bar or the GFRP-SC
bar, it is noted that the post-peak reduction of bond stress
of the GFRP-HW bar is much more gradual than the steel
bar or the GFRP-SC bar. In addition, the slip correspond-
ing to the bond strength of the GFRP-HW bar in normal
strength concrete is significantly larger than that of the steel
bar or the GFRP-SC bar in normal strength concrete. The
interfacial bond strength of the GFRP-HW bars is compa-
rable to that of the GFRP-SC bars, but smaller than that of
the steel bars. The bond strength of the GFRP-HW bar
increases moderately with increasing compressive strength
of the concrete.
Fig. 7. Interfacial bond stress vs. slip curves of rebars in normal- or high-
strength concrete (a) steel, (b) GFRP-SC, and (c) GFRP-HW.
4. Influence of high-strength concrete on interfacial bond
behavior of GFRP bars
92:4 MPaÞ. At the beginning of loading, the curves show
4.1. Interfacial bond stress vs. slip curves of rebars in virtually infinite slope for both normal and high-strength
normal- or high-strength concrete concrete, indicating that the applied load is transferred to
the concrete mainly by chemical bond between the bar
Fig. 7a–c shows the curves of interfacial bond stress vs. and the concrete. As the load increases, the chemical bond
free end slip of the rebars in normal strength concrete breaks, and slip commences at the loaded end of the bar.
ðfc0 ¼ 25:6 MPaÞ and in high-strength concrete ðfc0 ¼ The slip gradually increases with further increase in applied
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 265

Fig. 8. Interfacial bond failure of the steel bars after test.

load, and the bearing stress or friction against the sur- observed in both normal- and high-strength concrete. Since
rounding concrete become responsible for the load resis- the bond failure of the steel bar seems to be the outcome of
tance. The point of assumed complete debonding of concrete failure, the bond strength is likely governed by the
chemical adhesion is denoted as ‘‘Point A’’ and ‘‘Point compressive strength of concrete. On the other hand, the
B’’ in Fig. 7 for the normal- and high-strength concrete, bond failure of the GFRP bars occurred partly on the sur-
respectively. face between concrete and resin and partly on the surface
The bond stresses of three types of rebar at measurable between resin and bar fiber due to peeling of the surface
slip in the normal strength concrete (Point A in Fig. 7) layer of the bar, as shown in Fig. 9. The failure at interface
show the similar value, about 15 MPa regardless of the 1 occurs when the shear strength between concrete and
bar type. The assumed chemical adhesion improves with resin is smaller than the shear strength between resin and
higher concrete strength (Point B in Fig. 7), and the bar fiber, while the failure at interface 2 occurs when the
improvement is more noticeable in steel bar than in GFRP shear strength between resin and bar fiber is smaller than
bars. The values of slip at the maximum bond stress of the the shear strength between concrete and resin. Hence, the
rebars, on the other hand, is found greater in the normal bond strength and failure mode of the GFRP bars depends
strength concrete than in the high-strength concrete.
The bond stress–slip curves show different post-peak
behaviors for the different types of rebar with different con-
crete strengths. The curves of the steel bars and the heli-
cally wrapped GFRP bars show ductile characteristics in Delaminated area at
both normal- and high-strength concretes. But in case of interface 1
the sand coated GFRP bars, the ductile behavior is only (concrete and resin)
observed in the normal strength concrete. The GFRP-SC
bars in the high-strength concrete loose about 50% of the Delaminated area at
bond strength immediately after the peak stress as indi- interface 2
cated by the straight lines in Fig. 7b. The sudden drop in (resin and fiber)
post-peak bond stress suggests that the bond failure of
GFRP-SC bars is associated with very different mecha-
nisms from those of the other types of bars, when used in
high-strength concrete.

4.2. Delaminated area of GFRP

The cubes were split after the tests for a closer investiga-
tion of the actual mode of bond failure. Figs. 8–10 show
damages associated with the failures of the specimens.
Fig. 8 reveals that, in the case of the steel bar, ultimate
bond failure occurs due to concrete crushing against the
bar deformations, and the identical failure mode is Fig. 9. Interfacial bond failure of the GFRP bars after test.
266 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

Fig. 10. Interfacial bond failure of the GFRP bars in normal- or high-strength concrete.
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 267

Table 5
Delaminated area of GFRP rebars at surface 2
Specimens Gross bonding area (mm2) Delaminated area at surface 2 (mm2) Ratio (%)
F25 series F25-B1 3005 640 21.3
F25-B2 3105 1557 50.14
F25-C1 3116 665 21.34
F25-C2 2849 422 14.81
F35 series F35-B1 2693 1326 49.24
F35-B3 2547 1148 45.07
F35-C1 2740 1352 49.34
F35-C2 2796 1156 41.34
F40 series F40-B1 2774 1518 54.72
F40-B2 2713 1234 45.48
F40-C1 3070 1450 47.23
F40-C2 2927 1942 66.35
F55 series F55-B1 2880 1526 52.99
F55-B2 3019 1877 62.17
F55-C1 2626 1477 56.24
F55-C2 3088 2178 70.53
F75 series F75-B1 2872 2743 95.50
F75-B2 3182 2990 93.96
F75-C1 3041 2671 87.83
F75-C2 2869 2415 84.17
F90 series F90-B1 2823 2744 97.23
F90-B2 2823 2766 98.00
F90-C1 2895 2763 95.45
F90-C2 2820 2745 97.35

100 100
Delaminated area at surface 2 (%)

GFRP-SC GFRP-HW
Delaminated area at surface 2 (%)

80 80

60 60

40 40

20 20

0 0
0 25 50 75 100 0 25 50 75 100
Compressive strength of concrete (MPa) Compressive strength of concrete (MPa)
Fig. 11. Delaminated area at surface 2 of GFRP rebars vs. compressive strength of concrete.

on the relative shear strengths of interface 1 and 2. Fig. 10a total surface area of the GFRP rebars were determined and
and b for the normal strength concrete specimens show presented in Table 5. Fig. 11 shows that the ratio of the del-
concrete pieces still attached to the bars over the embed- aminated area at interface 2 to the total surface area of the
ment length, indicating that the interfacial bond failure GFRP rebars increases as the compressive strength of
occurs mainly at the interface 1. Fig. 10c–l shows the dam- concrete increases. The measured ratio in high-strength
aged core fibers due to interlaminar delamination, between concrete ðfc0 ¼ 92:4 MPaÞ and in normal strength of
the outer layer and the fiber core of the GRFP bars, at the concrete ðfc0 ¼ 25:6 MPaÞ is approximately 97% and 27%,
interface 2. Greater area of the interlaminar delamination respectively.
is observed with higher concrete strength. Fig. 12 shows the interfacial bond strength vs. the ratio
The delaminated areas at interface 2 were digitally mea- of the delaminated area to the total surface area of the
sured using a graphic software. The ratios of the area to the GFRP rebars. The ratio tends to increase with the increase
268 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

30 30
GFRP-SC GFRP-HW
Interfacial bond strength (MPa)

Interfacial bond strength (MPa)

25 25

20 20

15 15
20 40 60 80 100 20 40 60 80 100
Delaminated area at surface 2 (%) Delaminated area at surface 2 (%)
Fig. 12. Delaminated area at surface 2 of GFRP rebars vs. interfacial bond strength.

in the interfacial bond strength of which value depends on where sb,max is the bond strength; fc0 is the compressive
the shear strengths of the interface 1 and 2. For interfacial strength of concrete; db is the bar diameter; and b is the
bond strength greater than 25 MPa, the bond failures were coefficient related to fc0 (b = 1/2 has been adopted in
almost exclusively associated with the delamination of the CEB-FIP code for the sb,max in case of ‘‘good’’ bond
resin rich layer from the core fibers. conditions).
To investigate the influence of the compressive strength
of concrete on the bond strength of the GFRP bars, Olelo
4.3. Interfacial bond strength of steel and GFRP bars
and Yuan proposed Eq. (3) based on 151 test specimens
containing 6, 8, 10, 16, and 19 mm rebars embedded in a
The interfacial bond strength and the slip corresponding
203 mm concrete cube for different compressive strengths
to the bond strength are presented in Table 4. In addition,
of concrete (29–60 MPa). The average bond strength of
varying interfacial bond strengths of the rebars with differ-
FRP bars was proportional to square root of concrete com-
ent compressive strengths of concrete are shown in Fig. 13.
pressive strength.
The results show that the interfacial bond strength of the pffiffiffiffi
rebars increases with the compressive strength of concrete, fc0
and the effect of concrete strength on the interfacial bond sb;max ¼ 14:70 ðin MPaÞ ð3Þ
db
strength is much greater for the steel bars than the GFRP.
As mentioned earlier, the steel bar pullout failure is conse- On the other hand, Achillides indicated that for con-
quence of the concrete crushing in front of the bar defor- cretes with compressive strengths greater than 30 MPa,
mations, thus it is only just that the bond strength of the interfacial bond strength of FRP bars is not controlled
steel bar is deeply affected by the concrete strength. In case by the concrete strength because the bond failure, which
of the GFRP bars, on the other hand, the bond failure occurs on the interlaminar surface inside the FRP bars. It
occurred partly on the surface between concrete and resin is noted that the conclusions by both Olelo et al. and Achil-
and partly on the surface between resin and fiber. There- lides are based on the bond strength of FRP bars tested in
fore, the effect of concrete strength is compromised by the concrete with compressive strengths lower than 60 MPa.
the bond failure at surface 2. Here, we investigate the influence of the compressive
ACI design code and CEB-FIP model code adopt a con- strength of concrete on the bond strength of the GFRP
cept in which the bond strength increases in proportion to bars as a function of concrete strength up to 90 MPa based
the concrete tensile strength, which is related to the square on the experimental results reported above. First we
root of the compressive strength of concrete. assume a typical equation for the evaluation of bond
strength as given in
[ACI code]
b
pffiffiffiffi sb;max ¼ aðfc0 Þ ðin MPaÞ ð4Þ
fc0
sb;max ¼ 20:23 ðin MPaÞ ð1Þ Fig. 14 shows the relationship of the interfacial bond
db
strength of the rebars vs. compressive strength of concrete.
[CEB-FIP code] In this figure, the solid line represents the average bond
 0 b strength calculated by Eq. (4) assuming b = 0.5, which is
f a well-established value for steel bars, while the dotted line
sb;max ¼ 13:50 c ðin MPaÞ ð2Þ
30 represents the estimated bond strength assuming b = 0.3.
 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270 269

45 Test-Steel
Steel 45 Test-GFRP-SC
40 Test-GFRP-HW
40 Eq.(4) - Steel, β = 0.5
Eq.(4) - GFRP-SC, β = 0.5
Bond strength (MPa)

35 Eq.(4) - GFRP-HW,β = 0.3

35

Bond strength (MPa)

30
30
25
25
20
20
15
20 40 60 80 100
Compressive strength of concrete (MPa) 15
20 40 60 80 100
Compressive strength of concrete (MPa)
30
GFRP-SC Fig. 14. Increase of interfacial bond strength of steel and GFRP bars.
Bond strength (MPa)

25 of the GFRP bars, especially for high-strength concrete.

Because the parameter did not reflect the interlaminar fail-
ure which is common in the GFRP bars but nil in the steel
bars. Therefore, we here propose b = 0.3 to be used in Eq.
20 (4) to predict the bond strength of GFRP bars. The bond
strengths predicted using b = 0.3 show reasonable agree-
ment with the experimental bond strengths of the GFRP
bars as shown in Fig. 14.
15
20 40 60 80 100 5. Conclusions
Compressive strength of concrete (MPa)
The interfacial bond behavior of GFRP bars in normal-
30
or high-strength concrete was studied experimentally in this
GFRP-HW
research. Based on the results of this study the following
conclusions are made:
Bond strength (MPa)

25
(1) The bond strength of GFRP bars tended to increase
at a constant rate as the compressive strength of con-
crete increased. However, the increasing rate of the
bond strength of the GFRP bars was smaller than
20
that of the steel bars.
(2) The steel bar failed by concrete crushing in front of
the bar deformations and the bond strength was con-
trolled mainly by the shear strength of concrete. In
15
20 40 60 80 100 case of the GFRP bars, bond failure occurred partly
Compressive strength of concrete (MPa) on the surface between concrete and resin and partly
on the surface between resin and bar fiber. The bond
Fig. 13. Interfacial bond strength vs. compressive strength of concrete (a)
steel, (b) GFRP-SC, and (c) GFRP-HW.
failure of the GFRP bars in normal strength concrete
occurred mainly due to slip at interface 1 (interface
between concrete and resin), while that in the high-
In Eq. (4), the coefficient a was assumed as 4.1 for the steel strength concrete occurred mainly due to interlami-
bars and 3.3 for the GFRP bars in order to fit the bond nar delamination at interface 2 (interface between
strength of the rebars at fc0 ¼ 25:0 MPa. The interfacial resin and fiber).
bond strength of the steel bar predicted by Eq. (4) using (3) The increasing rate of the bond strength of the GFRP
b = 0.5 shows good agreement with the experimental bond bars with respect to concrete strength depended on
strength of the steel bar for the range of concrete strength the ratio of the delaminated area at surface 2 to the
considered. However, using the same parameter value total surface area of GFRP rebars. The results
b = 0.5, Eq. (4) clearly overestimates the bond strength showed that the bond strength of the GFRP bars
270 J.-Y. Lee et al. / Composites: Part B 39 (2008) 258–270

increased proportionally to the 0.3 power of concrete [10] Al-Zahrani MM, Nanni A, Al-Dulaijan SU, Bakis CE. Bond of FRP
compressive strength. to concrete in reinforcement rods with axisymmetric deformations.
In: El-Badry MM, editor. Advanced composite materials in bridges
and structures. Montreal (Que.): Canadian Society for Civil Engi-
neering; 1996. p. 853–9).
References [11] Larrelde J, Mueller-Rochholz J, Schneider T, Willmann J. Bond
strength of steel, AFRP and GFRP bars in concrete. In: Proceedings
[1] Rostasy FS. FRP: The European perspective. In: Proceedings of the of the ICCI’98, second international conference on composites in
first international conference on composites in infrastructure, Tuscon infrastructure, Tucson (AZ). January 1998. p. 92–101.
(AZ). 1996. p. 12–20. [12] Malvar LJ. Bond stress–slip characteristics of FRP rebars. In:
[2] Achillides Z, Pilakoutas K, Waldron P. Bond behaviour of FRP bars Technical Report TR-2013-SHR. Port Hueneme (CA): vch; 1994.
to concrete. Non-metallic (FRP) reinforcement for concrete struc- p. 45.
tures. In: Proceedings of the international symposium, Sapporo. [13] Achillides Z. Bond behavior of FRP bars in concrete. Thesis, Centre
October 1997. p. 341–8. for Cement and Concrete, Department of Civil and Structural
[3] Kettil P. Composite beams of fibre reinforced plastic profile and Engineering, The University of Sheffield, July 1998.
concrete. Thesis, Division of Building Technology, Chalmers Uni- [14] Tepfers R. Bond of FRP reinforcement in concrete. In: A state-of-
versity of Technology, Work no. 8, G} oteborg. September 1995. p. 80. the-art in preparation. Division of Building Technology, Chalmers
[4] Itoh S, Maruyama T, Nishiyama H. Study of bond characteristics of University of Technology, Work No. 15, G} oteborg. January 1997.
deformed fibre reinforced plastic rods. Proc Jpn Concr Inst p. 16.
1989;11(1):777–82 [in Japanese]. [15] Cosenza E, Manfredi G, Realfonzo R. Analytical modelling of bond
[5] Daniali S. Development length for fibre-reinforced plastic bars. In: between FRP reinforcing bars and concrete. In: Proceedings of the
Proceedings of the 1st international conference on advanced com- second international RILEM symposium (FRPRCS-S), London.
posite materials in bridge and structures, Sherbrooke (Que.), Canada. August 1995. p. 164–71.
[6] Molander I, Thalenius K. Icke metallisk armoring i betong (Non [16] Shima H, Suga T, Honma M. Local bond stress–slip relationship of
metallic reinforcement in concrete) Examensarbete. Division of continuous fibre reinforcing materials obtained by pull bond test with
building Technology, Chalmers University of Technology, Work long embedment. Trans Jpn Concr Inst 1993;15:297–304.
No. 770, G} oteborg. January 1992. p. 71. [17] Okelo R, Yuan RL. Bond strength of fiber reinforced polymer rebars
[7] Makitani E, Irisawa I, Nishiura N. Investigation of bond in concrete in normal strength concrete. J Compos Constr 2005;9(3):203–13.
member with fibre reinforced plastic bars. In: Nanni A, Dolan CW, [18] CSA. Design and construction of building components with fibre-
editors. Proceedings of the international symposium on fibre reinforced polymers. S806-02; 2002.
reinforced plastic reinforcement for concrete structure, ACI SP-138. [19] ASTM standard test method for comparing concrete on the basis of
Vancouver, Canada; 1993. the bond developed with reinforcing steel. C234; 1991.
[8] Hattori A, Inoue S, Miyagawa T, Fujii M. A study on bond creep [20] ASTM standard specification for moist cabinets, moist rooms, and
behavior of FRP rebars embedded in concrete. In: Proceedings of the water storage tanks used in the testing of hydraulic cements and
second international RILEM symposium (FRPRCS-S), London. concretes. C511-92; 1992.
August 1995. p. 172–9. [21] Eligehausen R, Popov EP, Bertero VV. Load bond stress–slip
[9] Jerrett CV, Ahmad SH. Bond tests of carbon fibre reinforced plastic relationships of deformed bars under generalized excitations. Report
(CFRP) rods. In: Proceedings of the second international rilem No. UCB/EERC82-83. Berkeley (CA): Earthquake Engineering
symposium (FRPRCS-S), London. August 1995. p. 180–91. Research Center, University of California; October 1983.