ARTICLE IN PRESS

Building and Environment 43 (2008) 98–107

www.elsevier.com/locate/buildenv

Bond performance of polystyrene aggregate concrete (PAC) reinforced

with glass-fibre-reinforced polymer (GFRP) bars
W.C. Tang, T.Y. Lo, R.V. Balendran
Department of Building and Construction, City University of Hong Kong, Hong Kong, SAR China
Received 21 February 2005; received in revised form 25 October 2006; accepted 6 November 2006

Abstract

This paper concerns the bond performance of GFRP bars in polystyrene aggregate concrete (PAC) by the proposed concentric pullout
test. Identical specimens reinforced with mild steel bars were used for comparison. The bond performance including the mode of failure
and bond strength was studied with varying polystyrene aggregate content, concrete strength, embedment length, shape and surface
treatment of the bars. Empirical formulae were developed for the estimation of development length on the basis of the pullout test results.
The bond development length ðl d Þ determined according to ACI Building Code was used for comparison.
Considering the bond performance, the sand-coated GFRP bar gave the most promising results with the highest bond strength
attained. In particular, the bond strength was found increased with the compressive strength and concrete density of PAC. Besides, the
bond development length for PAC reinforced with GFRP bars can be predicted satisfactorily using the ACI equation.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: GFRP; PAC; Bond behaviour; Bond development length; Concentric pullout test

1. Introduction mation and mechanical properties of GFRP bars are

dissimilar to those of steel reinforcing bars. Therefore, the
The use of steel bars as reinforcement embedded in design guideline for steel bars may not fully manifest the
concrete is evidently susceptible to corrosion problems, case for GFRP bars. Before they can be recommended as
which still worry most engineers. Many efforts to prevent concrete reinforcement, it is necessary to characterize their
corrosion have been attempted but just geared towards the structural behaviour with concrete in order to establish a
protection of steel merely [1]. Nevertheless, these attempts comprehensive design guideline for engineers to follow. In
still may not provide a best solution to the problems, as their this research, the bond behaviour of a new GFRP bar in
long-term effectiveness is still questionable in the actual field various configurations is investigated with the concrete
experience. Lately, fibre reinforced polymer (FRP) reinfor- matrix using polystyrene aggregate concrete (PAC).
cing bars have been developed and used for slabs as PAC is a lightweight concrete consisting of expanded
prestressing tendons, in marine environment structures, and polystyrene aggregate with densities ranging from 600 to
in structures wherein non-magnetic properties are required 1200 kg=m3 normally. Polystyrene concrete, for its remark-
such as the large transformer, foundation pads etc. [2,3]. able thermal insulation and low permeability, can be used
Glass-fibre-reinforced polymer (GFRP) reinforcing bars for insulating screens and rendering, non-loadbearing
present many advantages such as high strength-to-weight components, and thermally loadbearing courses of high-
ratio, electromagnetic neutrality, and ease of handling. Most way pavements.
beneficial is that, they are not affected by the electro-
chemical corrosion [4–6]. Unfortunately, the surface defor- 2. Previous studies

Corresponding author. Tel.: +852 27844761, +852 27889446. In this paper, the literature reviews are stressed mainly
E-mail address: [email protected] (W.C. Tang). on the bond behaviour of GFRP bars in concrete.

0360-1323/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2006.11.030
 ARTICLE IN PRESS
W.C. Tang et al. / Building and Environment 43 (2008) 98–107 99

The relevant literature for PAC is reviewed by Tang [7]. load decreases quickly with a slip in the descending phase
Since the substantial material and surface deformation of which indicates that as slip increases, the bond decreases
GFRP bars are different from that of steel bars, it is sharply due to the shearing of surface deformation of the
expected that the bond behaviour between GFRP bars and GFRP reinforcing bars. Benmokrane et al. [15] and Ehsani
concrete is also different from that of steel reinforced et al. [18] found similar observation.
concrete. Faza and GangaRao [19] investigated the bond beha-
In steel reinforced concretes, the bond strength depends viour of GFRP bars by testing the cantilever beams and
on two main mechanisms: (1) adhesion between the pullout specimens. Their results indicated that the decrease
concrete and the reinforcing bar, and (2) bearing of the in bond strength with the increasing embedment length was
reinforcing bar deformations against the concrete [8,9]. due to the increase in perimeter area of the reinforcing
The major contribution is the bearing component. bars. Similar observations were reported in the study of
The deformations on GFRP reinforcing bars, however, Tighiouart et al. [20].
contain a large amount of resin with a low shear modulus
and thus the contribution of the ribs in GFRP bars to the 3. Experimental details
resistance mechanism is significantly lower when compared
to that of steel bars. Consequently the adhesion and 3.1. Details of concrete mixes
friction may then be the important bond stress components
in those concretes reinforced with GFRP bars. This also In this research, a normal concrete (C5) and four
seems to be the case for other types of FRP reinforcing PAC mixes noted as PA20, PA40, PA60 and PA80 with
bars [10–14]. different proportions of polystyrene aggregate were stu-
Benmokrane et al. [15] studied the bond behaviour of died. These four PAC mixes were proportioned by
GFRP bars in concrete by the pullout tests. In any case, the replacing 20%, 40%, 60% and 80% of normal aggregate
results showed that all the deformed GFRP specimens from reference concrete with an equal bulk volume of
failed in the pullout patterns. They explained this might be polystyrene aggregates, respectively. The polystyrene ag-
an indication of low bearing stresses produced in the gregate was made from small polystyrene beans or recycled
concrete by the action of the GFRP bar’s deformation. granulates, coated with a non-toxic and patented chemical
Precisely, the surface deformation of GFRP bars does not compound that was supplied by BST (East Asia) Ltd. The
possess the similar characteristics of steel bars (i.e. high mean diameter and bulk density of the polystyrene beads
shear strength and rigidity) to provide sufficient lateral were 4 mm and 24 kg=m3 , respectively. The cement used
confinement through the rib bearings and it follows that throughout this investigation was ordinary portland
there is lower bond strength for the GFRP reinforcing bars cement (OPC), which conformed to BS 12:1991 and Type
[8,14,16]. I ASTM C150-92, supplied by the China Cement Company
The maximum bond strength for GFRP reinforcing bars (Hong Kong) Ltd. Table 1 shows the mix details of this
is approximately 60–90% compared to that of steel investigation.
reinforcing bars, depending on the diameter of the bars
and the embedment length as well. Generally, it is expected 3.2. Details of reinforcing bars
that the ultimate bond stress of GFRP bars increases with
concrete compressive strength. However, this effect is In this study, three types of GFRP reinforcing bars were
insignificant as reported by Nanni et al. [17]. investigated: they were smooth and circular GFRP bar
In the study of Chaallal and Benmokrane [14], the bond (Gsc), smooth and elliptical GFRP bar (Gse) and sand-
behaviour of GFRP bar in concrete was characterized by coated circular GFRP bar (Gsct). Mild steel bar was
the pullout tests. Their results showed that in the case of used for comparison. In case of the sand-coated GFRP
longer embedment length, failure of concrete splitting bar, its surface was coated with uniform size of quartz
occurred. In contrast, pullout failure occurred for shorter sand using epoxy resin. The GFRP bar was made of
embedment length used. Beyond the peak load, the pullout continuous longitudinal glass fibre strands impregnated in

Table 1
Mix details of PAC and reference concrete (C5) for cubic meter of the mix

Mix code Cement (kg) 10 mm agg. Sand (kg) Water (kg) Bulk volume of PA Absolute vol. of PA Weight of Raw density
(kg) in liters (V) in mix (%) PA used (kg) (kg)

C5 390 1131 609 195 — — — 2325

PA20 390 905 609 195 152 9.7 3.65 2100
PA40 390 679 609 195 304 19.5 7.30 1880
PA60 390 452 609 195 456 29.2 10.95 1660
PA80 390 226 609 195 608 38.9 14.59 1435
 ARTICLE IN PRESS
100 W.C. Tang et al. / Building and Environment 43 (2008) 98–107

Table 2
Tension properties of GFRP and mild steel reinforcing bars

Rebar types Cross-sectional area, Yield stress ðN=mm2 Þ Ultimate stress Ultimate strain (%) Elastic modulus
Ab ðmm2 Þ ðN=mm2 Þ ðkN=mm2 Þ

Gsc 50 — 1150 2.45 47.0

Gse 65 — 1210 2.55 47.5
MS 50 340.4 460 0.18a 185.0
a
Strain at yield point.

3.3. Details of bond test and sample preparation

3.3.1. Bond test method adopted

A modified version pullout setup proposed in the
Danish Standard DS 2082 [21] has been used in United
States [22,23]. In this set-up, the specimen consists of a
concrete prism in which two rods are embedded at the
opposite ends. The two rods have different embedment
lengths so that pullout is restricted to the shorter one.
The major contribution of this set-up is that the compres-
sion on the concrete face can be removed. As a result, a
similar set-up was used for the present study as illustrated
in Fig. 3a.

3.3.2. Pullout test specimens preparation

In this study, concrete specimens of 100 mm diameter
by 200 mm long were cast with the GFRP bars. The effect
Fig. 1. Appearance of GFRP and MS bars. of embedment lengths of the shorter bar was investigated
(i.e. 50; 70 and 90 mm). A 12 mm diameter of deformed
HY steel bar was used as the longer bar to improve the
1400 bearing capacity and minimize the slip between the longer
GFRP bar and concrete. As a result, the pullout is designed to be
1200
GFRP UTS: 2.5 times of MS restricted to the shorter one. The adjustment of embedment
Elastic Modulus : 25 % of MS
1000 length and alignment of the bars were carefully made prior
Stress (N/mm2)

to the concrete casting as shown in Fig. 3b. All the

800 specimens were cured in water tank at 27  1  C until
testing.
600
MS
The bond slip between the concrete and the reinforcing
400
bar was measured using a Digimatic indicator. This
indicator was mounted on the GFRP bar and measured
200 from the reference plate (i.e. the fixed rig on the pullout
specimen). The actual slip was obtained by subtracting the
0 elongation of the shorter bar from the total slip measured.
0 2 4 6 8 10 12
The pullout-loading rate for all the pullouts was 10 N/s.
Strain (10-2) The slip was recorded in each increment of 0.02 kN load.
Fig. 2. Typical stress–strain curve of GFRP and mild steel bars. The average bond stress was computed assuming a uniform
bond stress distribution along the embedded length of the
bar. The equation used for the computation was
Pmax
a thermosetting polyester resin using the pultrusion f av:b ¼ (1)
process. The density of GFRP bar was 1950 kg=m3 which pd b l d
is one-fourth that of mild steel bar of similar diameter. The where f av:b is the average bond stress; d b the nominal bar
mechanical properties and appearance of GFRP bars diameter; Pmax the maximum applied tension load; l d the
studied are shown in Table 2 and Fig. 1, respectively. embedded length of the shorter bars. In the case of
The typical stress–strain curve of GFRP bar compared to elliptical GFRP bars, the bond strength is determined
MS bar is given in Fig. 2. regarding the circumference of the elliptical bar.
 ARTICLE IN PRESS
W.C. Tang et al. / Building and Environment 43 (2008) 98–107 101

characterized by the fact that the concrete surrounding

a Load
the reinforcing bar is well confined to withstand the
splitting stress, the composites generally failed when the
bars were pulled out from the concrete, with a little or
Short Bar no visible cracking of the surrounding concrete. This
Reinforcing Bar
may be an indication of low bearing stresses produced
by the action of a smooth GFRP reinforcing bar and
ld as long as the adhesion is broken, the bar moves as a
100 - ld
rigid body [8,25]. Fig. 5 gives the typical load–slip
200 mm relationship of the bond specimen that failed in bar pullout
compared to the one that failed in concrete splitting. It can
100 mm be seen that the specimen failed in pullout gives a more
PAC
ductile failure when compared to that failed in concrete
splitting. Similar observations were made by Hamza and
Naaman [26].
Long Bar
ld = Embedment length 4.1.2. Concrete tensile splitting failure
In all cases of specimens bonded with sand-coated
Load GFRP bar, an explosive concrete splitting type of failure
occurred. Obviously, this failure was mainly observed in
b sand-coated GFRP bars. Nevertheless, two different
concrete splitting crack patterns were observed: (1)
splitting failure with side longitudinal cracks and (2)
splitting failure without side longitudinal cracks as shown
in Fig. 4.
Referring to Table 3, it seems that the side longitudinal
cracks are more likely to occur at relatively high concrete
density and strength. The concrete specimens having
the nominal density of 1880 kg=m3 or above generally
showed side longitudinal cracks of failure patterns with a
great explosion. In contrast, concrete specimens of
1650 kg=m3 or below, the failure mode turned out in
concrete splitting with less explosion and no visible side
Fig. 3. (a) Concentric pullout test specimen used in this investigation. (b)
longitudinal cracks. It can be deduced that concretes of
Ancillary fitting for alignment and embedment length adjustment. higher density are more brittle than those concretes with a
great proportion of polystyrene aggregate content. PAC
appears to be a reliable energy-absorbing material because
4. Results and discussions of its low crushing strength combined with their large
deformation capacity. The amount of energy therefore
The bond performance of GFRP bars reinforced required to cause damage and the energy dissipated
concrete can be characterized by their mode of failure, increase with the increase in polystyrene aggregate content
bond strength and bond–slip relationship. In this paper, [27]. As a result, the failure becomes less brittle in low PAC
the discussions are focused on their mode of failure, bond density.
strength and development length as well. The bond–slip The occurrence of splitting failure indicates that there is
relationship can be found in Balendran et al. [24]. insufficient lateral resistance of the specimen to withstand
the pullout force applied. The GFRP bar coated with sand
4.1. Mode of failure creates sufficient radial force components that split the
concrete before bond failure [26,28]. Since the splitting
The observed types of failure are listed in Table 3. The occurred in a tension mode, the failure is brittle. As a
specimens in general showed either concrete tensile splitting result, the splitting failure can be generally characterized by
failure or bar pullout failure. Fig. 4 shows the typical failure the formation of tensile cracks and accompanied by a large
patterns after the pullout test. The details of each type of sudden drop in the load–slip curve as shown in Fig. 5.
failure modes are described and discussed as follows. Specimens that failed in concrete splitting, however, lead us
to conclude that the bond between the sand-coated GFRP
4.1.1. Pullout failure bars and the surrounding concrete is much promising and
The pullout type of failure occurred in the majority of effective than those bonded with the smooth GFRP and
specimens embedded with smooth bars. Pullout failure is mild steel bars.
 ARTICLE IN PRESS
102 W.C. Tang et al. / Building and Environment 43 (2008) 98–107

Table 3
Bond performances of PACs and reference concrete reinforced with GFRP and mild steel bars

Mix details Sample Comp. Embed. Bond Power Development Failure

notation strength length strength exponent length ðI bd Þ, mm mode
ðf cyl Þ ðN=mm2 Þ (mm) ðf b Þ ðN=mm2 Þ (n)a
Test (1) Calculated (2) Ratio of (1)/(2)

Density 2325 kg=mm3 C5-Gsc 55.0 50 2.35 0.33 979 595 1.64 P
70 2.05 0.33 1122 595 1.88 P
90 1.90 0.33 1211 595 2.03 P
C5-Gse 55.0 50 2.55 0.24 765 989 0.77 P
70 2.40 0.24 813 989 0.82 P
90 2.35 0.24 830 989 0.84 P
PA content 0% C5-Gsct 55.0 50 10.60 0.50 271 744 0.36 SS
70 8.75 0.50 329 744 0.44 SS
90 8.20 0.50 351 744 0.47 SS
C5-MS 55.0 50 4.90 0.28 139 176 0.79 P
70 4.20 0.28 162 176 0.92 P
90 4.10 0.28 166 176 0.94 P

Density 2100 kg=mm3 PA20-Gsc 32.0 50 1.85 0.33 1243 1015 1.23 P
70 1.80 0.33 1278 1015 1.26 P
90 1.75 0.33 1314 1015 1.30 P
PA20-Gse 32.0 50 2.30 0.24 848 1685 0.50 P
70 2.20 0.24 886 1685 0.53 P
90 2.20 0.24 886 1685 0.53 P
PA content 9.7% PA20-Gsct 32.0 50 7.35 0.50 391 1269 0.31 SS
70 5.80 0.50 496 1269 0.39 SS
90 5.05 0.50 569 1269 0.45 SS
PA20-MS 32.0 50 3.05 0.28 223 300 0.74 P
70 3.30 0.28 206 300 0.69 P
90 2.95 0.28 231 300 0.77 P

Density 1880 kg=mm3 PA40-Gsc 21.0 50 1.55 0.33 1484 1253 1.18 P
70 1.45 0.33 1586 1253 1.27 P
90 1.55 0.33 1484 1253 1.18 P
PA40-Gse 21.0 50 2.00 0.24 975 2081 0.47 P
70 1.90 0.24 1026 2081 0.49 P
90 1.80 0.24 1083 2081 0.52 P
PA content 19.5% PA40-Gsct 21.0 50 5.80 0.50 496 1566 0.32 SS
70 4.65 0.50 618 1566 0.39 SS
90 4.20 0.50 685 1566 0.44 SS
PA40-MS 21.0 50 2.85 0.28 239 370 0.64 P
70 3.10 0.28 219 370 0.59 P
90 2.85 0.28 239 370 0.64 P

Density 1650 kg=mm3 PA60-Gsc 14.0 50 1.45 0.33 1586 1534 1.03 P
70 1.45 0.33 1586 1534 1.03 P
90 1.40 0.33 1643 1534 1.07 P
PA60-Gse 14.0 50 1.90 0.24 1026 2548 0.40 P
70 1.85 0.24 1054 2548 0.41 P
90 1.75 0.24 1114 2548 0.44 P
PA content 29.2% PA60-Gsct 14.0 50 5.15 0.50 558 1918 0.29 S
70 4.35 0.50 661 1918 0.34 S
90 3.85 0.50 747 1918 0.39 S
PA60-MS 14.0 50 2.75 0.28 247 454 0.55 P
70 2.80 0.28 243 454 0.54 P
90 2.65 0.28 257 454 0.57 P

Density 1400 kg=mm3 PA80-Gsc 9.6 50 1.20 0.33 1917 1853 1.03 P
70 1.15 0.33 2000 1853 1.08 P
90 1.14 0.33 2018 1853 1.09 P
PA80-Gse 9.6 50 1.65 0.24 1192 3103 0.38 P
70 1.55 0.24 1269 3103 0.41 P
90 1.60 0.24 1229 3103 0.40 P
PA content 38.9% PA80-Gsct 9.6 50 4.05 0.50 710 2316 0.31 S
70 3.60 0.50 799 2316 0.34 S
90 3.20 0.50 898 2316 0.39 S
 ARTICLE IN PRESS
W.C. Tang et al. / Building and Environment 43 (2008) 98–107 103

Table 3 (continued )

Mix details Sample Comp. Embed. Bond Power Development Failure

notation strength length strength exponent length ðI bd Þ, mm mode
ðf cyl Þ ðN=mm2 Þ (mm) ðf b Þ ðN=mm2 Þ (n)a
Test (1) Calculated (2) Ratio of (1)/(2)

PA80-MS 9.6 50 2.55 0.28 267 548 0.49 P

70 2.65 0.28 257 548 0.47 P
90 2.55 0.28 267 548 0.49 P

Failure modes: P ¼ rebar pullout; S ¼ concrete splitting only; SS ¼ concrete splitting with side cracks.
a
The power expression, n varies with the type of rebars as given in Eqs. (2)–(5).

Fig. 4. Typical concrete splitting failure modes: (a) with side longitudinal cracks; (b) without side longitudinal cracks.

8 The resin rich surface layer of GFRP bars and their

Load
Concrete splitting faliure smoothness are the critical factors in accounting for the
less slip relatively low bond strength.
2. The Gse bars with higher contact bond area generally
6
Bond Stress (fb), N/mm2

showed higher bond strength values than the Gsc bars

for the same embedment length and concrete matrix.
The maximum bond strength for the Gse bar was about
4 10–40% higher than the Gsc bar. This bond strength
Load
difference, however, increases with the decrease of
concrete density.
3. The maximum bond strength of the Gse bar was about
2 55–75% of the MS rebar. In spite of the contact bond
area is greater than that of the MS bar, the bond
Rebar pullout faliure strength of the Gse bar is lower.
Large slip
4. As indicated in Table 3, Gsct bar attained the highest
0
0 2 4 6 8 10 12
bond strength values among the tested reinforcing bars.
Slip (mm)
The maximum bond strength for the Gsct bar was
approximately 180–350% greater than the Gsc bars.
Fig. 5. Typical bond–slip relationships of bond specimens failed in bar Besides, Gsct bar showed 100–310% and 25–140%
pullout and concrete tensile splitting. greater in bond strength when compared to the Gse bars
and MS bars, respectively. These bond strength differ-
4.2. Average bond strength
ences, however, decrease with the decrease in concrete
density.
Table 3 lists the bond strength results in this study. Based
5. A GFRP bar coated with sand can provide some lateral
on the results, the following observations were made:
confinement through the contribution of the sand-
coated surface. This resistance mechanism somehow is
1. The maximum bond strength for Gsc bar was about lower than that of the deformed steel reinforcing bars.
45–60% of the MS rebars, depending on the values of However, the highest bond strength developed as well as
the embedment length and the surrounding concrete. its concrete splitting failure pattern did evidently show
 ARTICLE IN PRESS
104 W.C. Tang et al. / Building and Environment 43 (2008) 98–107

that the interfacial bond between the Gsct bar and the For MS
surrounding concrete is very great and promising. Most
concrete specimens particular in lower concrete density f b:MS ¼ 1:31f 0:28
cyl ; R2 ¼ 89:4%. (5)
(i.e. high PA content) split first before the occurrence of It is widely accepted that for steel reinforcement, the
bond failure. bond strength is proportional to the tensile strength of
concrete [31]. Since the tensile strength of concrete is
proportional to the square root of its compressive strength,
4.3. Relation between bond strength and concrete strength therefore the bond expressions also include the ðf cyl Þ0:5
term according to ACI 318-95 [32]. However, the regres-
In general terms, bond strength is related to the quality sion lines shown in Fig. 6b suggest that the bond strengths
of the surrounding concrete, and the bond strength is for the smooth bars, neither GFRP nor mild steel bars do
approximately proportional to the compressive strength up vary linearly with ðf cyl Þ0:5 except for the Gsct bars.
to about 20 MPa [29]. Fig. 6a illustrates the relation As expressed in Eq. (4), the power exponent of 0.50
between the bond strength and the compressive strength for Gsct bar shows an excellent agreement with the value of
for PACs. The bond strength for most types of reinforcing 0.5 recommended by the design codes for steel bar. In
bar studied was found to increase with the compressive contrast, the power exponents yielded from Eqs. (2), (3)
strength. The rate of increase, however, decreased with the and (5) for Gsc, Gse and MS, respectively, did significantly
increase in strength, which is similar to the findings made deviate from the value of 0.5. It seems that the current
by Benfenier [27] and Gjorv et al. [30]. The regression codes can be satisfactorily applicable to the sand-coated
expressions derived from Fig. 6a, relating the compressive GFRP bars of this investigation as well as the conventional
strength ðf cyl Þ and the bond strength ðf b Þ for PAC steel bars.
reinforced with the GFRP and mild steel bars are expressed
accordingly as follows:
For Gsc 4.4. Influence of bar embedment length on bond strength
f b:Gsc ¼ 0:57f 0:33
cyl ;
2
R ¼ 95:3%. (2) The effect of embedment length on bond strength is
For Gse, shown in Table 3. It can be concluded that as the
embedment length increases, the applied pullout load
f b:Gse ¼ 0:94f 0:24
cyl ; R2 ¼ 94:8%. (3) approaches the tensile strength of the reinforcing bar.
For Gsct The average bond strength therefore diminishes and the
specimens with a shorter embedment length of the bars in
f b:Gsct ¼ 1:13f 0:50
cyl ; R2 ¼ 85:8%. (4) general develop higher bond strength.

a b
12 12
Legend Regression equations Legend Regression equations

Gsc fb.Gsc =0.570 fcyl

0.328 R2 =0.9529 Gsc fb.Gsc = 0.317fcyl0.5 R2 = 0.659
0.5
10 10 fb.Gse= 0.390fcyl R2 = 0.350
Gse fb.Gse =0.943 fcyl0.240 R2 =0.9476 Gse

0.502 R2 =0.8581 Gsct fb. Gsct = 1.153fcyl

0.5 R2 = 0.866
Gsct fb.Gsct = 1.126 fcyl
Bond Strength (fb), N/mm2
Bond Strength (fb), N/mm2

0.279 MS fb.MS = 0.621fcyl

0.5 R2 = 0.556
MS fb.MS = 1.308 fcyl R2 = 0.8943
8 8

6 6

4 4

2 2

0 0
0 10 20 30 40 50 60 0 1 2 3 4 5 6 7 8
Compressive Strength (f cyl), N/mm2 SQRT of Compressive Strength (f cyl)0.5, N/mm2

Fig. 6. Relation between bond strength and compressive strength.

 ARTICLE IN PRESS
W.C. Tang et al. / Building and Environment 43 (2008) 98–107 105

The decrease in bond strength with the increase in as follows:

embedment length might probably be due to the increase in 12Uf s abl
the perimeter area of the bars [16,19,20,25,33,34]. The fact I bd ¼ pffiffiffiffiffiffiffi for elliptical rebar. (8a)
that the bond strength varies with the embedment length 25p f cyl
indicates that the bond stresses along the bonded length are
not evenly distributed [35].
4.5.2. Development length results
It can be seen from Table 3, the bond development
4.5. Development length lengths required to develop the ultimate tensile strengths
for reinforcing bars, Gsc, Gse and Gsct ranged from 979 to
4.5.1. Evaluation of bond development length 2018, 765 to 1192 and 271 to 898 mm, respectively,
The force required to pull a reinforcing bar out of depending on concrete density and strength. For MS bars,
concrete increases as the length of the bar embedded in the corresponding values ranged from 139 to 267 mm.
concrete increases [14,36]. When the embedment length The results from Table 3 indicate that the difference in
becomes long enough the bar will reach its ultimate tensile development lengths between the GFRP and MS bars is
strength before it pulls out of the concrete. The bond very large, especially when Gsc bars are compared to MS
development length, I bd (also termed as optimal anchored reinforcing bars. The higher development lengths required
length) is the minimum embedment length necessary to for GFRP bars may be probably due to the fact that their
develop the ultimate tensile strength of the reinforcing bars ultimate tensile strengths are nearly three times that of MS
through the bond [20]. For conventional steel bars, it is the bars, the development length required to develop the
length required to develop their yield strength in tension. ultimate tensile strength for a GFRP bar is therefore higher
From equilibrium, the bond development length can be than that of a MS bar.
calculated by equating the bond force over the surface of
the bar obtained from the pullout test to the ultimate 4.5.3. Effect of concrete strength and density on
tensile load of the bar: development length
In both cases of GFRP and MS bars, the development
f b UI bd ¼ Ab f pu , (6) length decreases with the increase in the compressive
where f b is the maximum bond strength ðN=mm2 Þ; U the strength and the concrete density. The decrease, however,
circumference of the bar (mm); I bd the bond development tends to minimize with the increase in compressive strength
length (mm); Ab the cross-sectional area of the bar ðmm2 Þ; or concrete density. In other words, it seems that the
and f pu the ultimate tensile strength of GFRP rebar development lengths will be increased when the polystyrene
ðN=mm2 Þ. In the case of MS bar, the yield strength, f s was aggregate content in the mix increases. Fig. 7a shows
used instead of f pu . Eq. (6) can be rearranged to the that for the range of concrete strength tested, the
following expressions: development length does not change significantly for
compressive strength above 40 N=mm2 . Similar observa-
I bd ¼ d b f pu =4f b for circular rebar, (7a) tion for PAC bonded with high yield steel bars was
reported by Benfenier [27].
I bd ¼ Ab f pu =Uf b for elliptical rebar. (7b)
4.5.4. Comparison between test and calculated development
Results of the bond development length derived from the
lengths
pullout test results are presented in Table 3. The develop-
In this study, the development lengths obtained from the
ment lengths calculated according to the ACI Building
pullout test were compared to the ones calculated by the
Code requirements [32] for circular MS reinforcing bar as
ACI Building Code (Eq. (8)). In Table 3, the development
described in the equation below, are also given in Table 3
lengths for both GFRP and MS bars (except for circular
for comparison: GFRP bar) in general were over-estimated by the ACI
12f s abld b Code as compared to the experimental development
I bd ¼ pffiffiffiffiffiffiffi for circular rebar, (8) lengths. The ratio of experimental lengths to the calculated
25 f cyl
lengths for Gsc, Gse, Gsct and MS bars, ranged from 1.03
where a is the reinforcement location factor; b the coating to 2.03, 0.38 to 0.84, 0.31 to 0.47 and 0.47 to 0.94,
factor; l the lightweight concrete factor; f cyl the cylinder respectively. The ratio in general decreases with the
compressive strength ðN=mm2 Þ; and f s the yield strength of decrease of concrete density and concrete strength.
reinforcing bar (f pu for GFRP bar instead). According to Fig. 7b shows the relation between the ratio of the
the ACI code requirement, the factors of a and b used in experimental lengths to the calculated lengths and com-
this study were both 1.0. The factors of l for PAC and pressive strength, for various reinforcing bars. It can be
reference concrete (C5) were 1.3 and 1.0, respectively. To observed from Fig. 7b that the ratio tends to increase with
determine the development length for the elliptical GFRP the increase in compressive strength. For Gse, Gsct and
bar, the expression given in Eq. (8) needed to be rearranged MS bars, the ratio approaches 1.00 when concrete strength
 ARTICLE IN PRESS
106 W.C. Tang et al. / Building and Environment 43 (2008) 98–107

a b
2.5
2500
GFRPsc GFRPse
GFRPsc GFRPse

Test Length / Calculated Length (in Ratio)

GFRPsct MS GFRPsct MS
2000 2.0
Development Length (Ibd), mm

1500 1.5

1000 1.0

500 0.5

0 0.0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Concrete Strength (fcyl), N/mm2 Concrete Strength (fcyl), N/mm2

Fig. 7. Relation between development lengths and compressive strength based on pullout test results.

increases. In contrast, the ratio for Gsc bars exceeds 1.00 5. It should be noted that the power exponent of 0.50 for
by far when concrete strength increases. As a result, Eq. (8) Gsct bar given in Eq. (4) showed an excellent agreement
given by ACI Code may not be suitable to evaluate the with the value of 0.5 recommended by the design codes
development length for specimens in high concrete strength for conventional steel reinforcement.
reinforced with the Gsc bars. It, however, can be used to 6. The development lengths for PAC specimens particu-
predict conservatively the development lengths for speci- larly in high concrete strength reinforced with the Gse,
mens particularly in high concrete strength reinforcing with Gsct and MS bar can be predicted conservatively by the
the Gse, Gsct and MS bar. ACI Code as compared to the experimental lengths.

5. Conclusions
Acknowledgement
1. The bond specimens in general showed either concrete
tensile splitting failure or pullout failure. The occurrence The work described in this paper was fully supported
of splitting failure was mainly observed in sand-coated by a grant from City University of Hong Kong (Project
GFRP reinforcing bars. No. 7001408).
2. Two different concrete splitting crack patterns were
observed: (1) splitting failure with side longitudinal
References
crack and (2) splitting failure only. The failure pattern
changed from (1) to (2) as the density and strength of [1] Benmokrane B, Tighiouart B, Chaallal O. Bond strength and load
concrete decreased. distribution of composite GFRP reinforcing bars in concrete. ACI
3. Due to the resin rich surface layer and smoothness, the Materials Journal, 1996; May–June: 246–53.
bond strength developed for specimens bonded with the [2] Ballinger CF. Development of fibre-reinforced plastic products for
smooth GFRP bars were relatively lower than those the construction market—How has and can it be done. In: Advanced
composite materials in bridges and structures, first international
specimens bonded with the MS bars. Sand-coated conference, 1992. p. 3–14.
GFRP specimens, however, achieved the highest bond [3] Faruqi MA. Mechanical properties and endurance limit of fibre
strength. composite bars. Journal of Reinforced Plastics and Composites
4. Results showed that specimens with shorter embedment 1998;17(17):1512–24.
length develop higher bond strength. In addition, the [4] Saadatmanesh H. Fiber composites for new and existing structures.
ACI Structural Journal 1994;91(3):346–54.
bond strength increased with the increase in compressive [5] Taerwe L. Non-metallic reinforcement for concrete structures. In:
strength and concrete density. The rate of increase, Proceeding of international symposium on new technology in
however, decreased with the increase in concrete strength. structural engineering. IABSE, Zurich, 1997. p. 15–24.
 ARTICLE IN PRESS
W.C. Tang et al. / Building and Environment 43 (2008) 98–107 107

[6] Yost JR, Goodspeed CH, Schmeckpeper ER. Flexural performance [21] Danish Standards Organization, Pull Out Test (DS 2082), Copenha-
of concrete beams reinforced with FRP grids. Journal of Composites gen, Denmark, December 1980. p. 2.
for Construction 2001;5(1):18–25. [22] Ezeldin AS, Balaguru PN. Bond behaviour of normal and high-
[7] Tang WC. Engineering properties of polystyrene aggregate concrete strength fibre reinforced concrete. ACI Materials Journal 1989;
with and without fibre reinforced polymer bars, PhD thesis, City September–December: 515–24.
University of Hong Kong, 2002. [23] Chapman RA, Shah SP. Early-age bond strength reinforced concrete.
[8] Larralde J, Silva-Rodriguez R. Bond and slip of FRP reinforcing bars ACI Materials Journal 1987;84(6):501–10.
in concrete. Journal of Materials in Civil Engineering 1993;5(1): [24] Balendran RV, Tang WC, Leung HY, Nadeem A. Bond stress–slip
30–40. relationships of glass-fibre-reinforced polymer (GFRP) bars em-
[9] Treece RA, Jirsa JO. Bond strength of epoxy-coated reinforcing bars. bedded in polystyrene aggregate concrete (PAC). In: 28th interna-
ACI Materials Journal 1989;86(2):167–74. tional conference on our world in concrete & structures, August 2003.
[10] Daniali S. Development length for fibre reinforced plastic bars. p. 223–32.
Advanced composite materials, on bridges and structures. Canadian [25] Nanni A, Al-Zaharani MM, Al-Dulaijan SU, Bakis CE, Boothby TE.
Society for Civil Engineering, 1992. p. 179–88. Bond of FRP reinforcement to concrete-experimental results. In:
[11] Ehsani MR. Glass fibre reinforcing bars. In: Clarke JL, editor. Taerwe L, editor. Non-metallic (FRP) reinforcement for concrete
Alternative materials for the reinforcement and pre-stressing of structures. E & FN Spon; 1995. p. 135–45.
concrete; 1993. p. 34–54. [26] Hamza AM, Naaman AE. Bond characteristics of deformed
[12] Mashima M, Iwamoto K. Bond characteristics of FRP rod and reinforcing steel bars embedded in sifcon. ACI Materials Journal
concrete after freezing and thawing deterioration, In: ACI interna- 1996; November: p. 578–88.
tional symposium on FRP reinforcement for concrete structures, [27] Benfenier AS. Engineering properties of polystyrene aggregate
Vancouver, Canada, SP-138. Detroit: American Concrete Institute; concrete. PhD thesis, University of Technology, Sydney, Australia,
1993, p. 51–69. 1998.
[13] Ehsani MR, Saadatmanesh H, Tao S. Bond of GFRP reinforcing [28] Goto Y. Cracks formed in concrete around deformed tension bars.
bars to ordinary strength concrete. In: ACI international symposium Journal of American Concrete Institute 1971;68:244–51.
on FRP reinforcement for concrete structures, Vancouver, Canada, [29] Neville AM. Properties of concrete. London: Longman; 1995.
SP-138. Detroit: American Concrete Institute; 1993, p. 333–45. [30] Gjorv OE, Monteiro PJM, Mehta PK. Effect of condensed silica
[14] Chaallal O, Benmokrane B. Pullout and bond of glass-fibre rods fume on the steel-concrete bond. ACI Material Journal 1990;87(6):
embedded in concrete and cement grout. Materials and Structures 573–80.
1993;26:167–75. [31] Ferguson PM, Thompson JN. Development length of high strength
[15] Benmokrane B, Chaallal O, Masmoudi R. Flexural response of reinforcing bars. Journal of American Concrete Institute 1962;59(7):
concrete beams reinforced with FRP reinforcing bars. ACI Structural 887–921.
Journal 1996;93(1):46–55. [32] ACI Committee 318M-95, Building code requirements for reinforced
[16] Makitani E, Irisawa I, Nishiura N. Investigation of bond in concrete concrete (ACI 318M-95), and commentry-ACI 318RM-95. ACI
member with fibre reinforced plastic bars, In: ACI international Manual of Concrete Institute, Farmington Hills, MI, 1995. p. 371.
symposium on FRP reinforcement for concrete structures, Vancouver, [33] Al-Zahrani MM, Al-Dulaijan SU, Nanni A, Bakis CE, Boothby TE.
Canada, SP-138, Detroit: American Concrete Institute; 1993, p. 315–32. Evaluation of bond using FRP rods with axisymmetric deformations.
[17] Nanni A, Bakis CE, Boothby TE. Test method for FRP-concrete Construction and Building Materials 1999;13:299–309.
systems subjected to mechanical loads: state of the art review. Journal [34] Kachlakev DI, Lundy JR. Performance of hollow glass fibre-
of Reinforced Plastics and Composites 1995;14:524–59. reinforced polymer rebars. Journal of Composites for Construction
[18] Ehsani MR, Saadatmanesh H, Tao S. Bond behaviour of deformed 1999;3(2):87–91.
GFRP rebars. Journal of Composite Materials 1997;31(14):1413–30. [35] Lorenzis LD, Nanni A. Characterization of FRP rods as near-surface
[19] Faza SS, GangaRao HVS. Bending and bond behaviour of concrete mounted reinforcement. Journal of Composites for Construction
beams reinforced with plastic rebars. Transportation Research 2001; May: p. 114–21.
Record 1290, 1990. p. 185–93. [36] Xu H, Bellavance E, Benmokrane B. Bond strength of fibre
[20] Tighiouart B, Benmokrane B, Gao D. Investigation of bond in reinforced plastic (FRP) cement grouted anchor bolts. In: Taerwe
concrete member with fibre reinforced polymer (FRP) bars. L, editor. Non-metallic (FRP) reinforcement for concrete structures.
Construction and Building Materials 1998;12:453–62. E & FN Spon; 1995. p. 209–16.