Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract: Bond behavior is an important subject in the design and performance of reinforced concrete structures. In this research, the bond
property between sand-coated glass fiber–reinforced polymer (GFRP) bars, a corrosion-resistant substitute to steel bars, and fly ash–based
geopolymer cement (GPC) concrete, a more environmental friendly alternative to ordinary portland cement (OPC) concrete, is investigated.
Pullout test specimens containing GFRP bars embedded in GPC and OPC concrete cylinders with 100-mm diameter and 170-mm height were
prepared. Three different embedment lengths were tested: three, six, and nine times the bar diameter. Average concrete compressive strengths
of approximately 25 and 45 MPa and GFRP bar diameters of 12.7 and 15.9 mm were the other test parameters. For each specimen, the test
results include the bond failure mode, the average bond strength, the slip at the loaded and free end, and the bond-slip relationship curves. The
test results showed that GFRP-reinforced GPC concrete has similar bond strength as that of GFRP-reinforced OPC concrete. The increase in
embedment length resulted in the decrease of the bond strength as well as a change in the failure mode of the specimens. Furthermore, the
experimental results were used to generate a constitutive bond-slip law. Finally, finite-element modeling is performed by using the
constitutive bond-slip law to investigate strain and bond distribution along the embedment length of the bar. DOI: 10.1061/(ASCE)CC
.1943-5614.0000685. © 2016 American Society of Civil Engineers.
Author keywords: Geopolymer concrete; Glass fiber–reinforced polymer (GFRP) bars; Bond properties; Development length; Bond slip.
Introduction and/or alkali silicate) to react with silica (SiO2 )- and alumina
(Al2 O3 )-rich natural materials, such as metakaolin, or with indus-
Steel-reinforced ordinary portland cement (OPC) concrete is a suc- trial by-products, such as fly ash (FA), silica fume (SF), rice husk
cessful construction material because of its affordability, strength, ash (RHA), or slag (Hardijito and Rangan 2005). In addition to the
and ease of construction. However, concrete, especially when ex- low carbon footprint, using industry by-product materials such as
posed to aggressive environments such as those encountered in fly ash, which otherwise would be disposed of, saves a great deal of
coastal areas or chemical plants, deteriorates because of the ingress virgin materials that would otherwise be used for concrete produc-
of deleterious agents through its pores. In addition to attacking con- tion. These binders have been used to produce what is now known
crete, the deleterious agents also result in the corrosion of the steel as geopolymer cement (GPC) concrete. In addition to environmen-
bars, reducing the service life of the structure. To deal with these tal benefits, various desirable characteristics were reported by using
problems, various methods such as concrete additives, galvaniza- GPC concrete. Such characteristics include rapid rate of strength
tion, epoxy coating, and using cathodic protection systems have development, resistance to sulphate attack, acid resistance, little
been applied, but they still are not the ultimate solution (Kessler and drying shrinkage, low creep, improved resistance to fire, and pro-
Powers 1998). Moreover, production of OPC concrete comes at a longed handling time (Hardijito and Rangan 2005; Junaid et al.
great risk to the environment because of the large CO2 emissions 2014, 2015b).
associated with the production of OPC during lime calcination. Fiber-reinforced polymer (FRP) reinforcing bars can be used
The high environmental footprint of reinforced concrete and instead of steel bars in reinforced concrete structures. They are
its durability issues have inspired scientists and engineers to find mainly based on thermoset polymers vinyl ester and glass (GFRP)
alternatives for both OPC and steel reinforcement for more sustain- or carbon fibers (CFRP) and are characterized by high tensile
able construction. Davidovits (1988) proposed an alternative binder strength, high durability, light weight, and electromagnetic per-
produced from by-product materials that are rich in silica and alu- meability (Bank 2006). Because of lack of well-stablished stan-
mina, such as fly ash and rice husk ash, and named it geopolymer dards, wide varieties of FRP bars are available. If GFRP bars and
because of the polymerization process involved. These binders are GPC concrete are to be used as alternatives for steel reinforcement
produced using alkali liquids (usually a soluble metal hydroxide and OPC concrete, they need to display similar, or better, physical
characteristics. High ranking among these are bond properties. In-
1
Graduate Student, School of Engineering and Information Technology, deed, reinforced concrete is itself a composite material. Maintain-
UNSW Canberra, Campbell, ACT 2612, Australia. ing this composite action requires transfer of load between the
2
Senior Lecturer, School of Engineering and Information Technology, concrete and the reinforcing bar. It is this load transfer, called bond,
UNSW Canberra, Campbell, ACT 2612, Australia (corresponding author). which makes reinforced concrete structures effective. Depending
E-mail: a.khennane@adfa.edu.au on the surface condition of the FRP bar, the bond mechanism
3
Associate Professor, School of Engineering and Information Technol-
can differ. For straight FRP bars (smooth bars and bars with no
ogy, UNSW Canberra, Campbell, ACT 2612, Australia.
Note. This manuscript was submitted on October 27, 2015; approved on ribs or indentation), adhesion and friction have been found to be
December 29, 2015; published online on March 3, 2016. Discussion period responsible for the bond resistance (Cosenza et al. 1997). These
open until August 3, 2016; separate discussions must be submitted for properties in turn depend on the bar surface condition and the type
individual papers. This paper is part of the Journal of Composites for Con- of concrete used. Conversely, deformed bars (helically wound
struction, © ASCE, ISSN 1090-0268. spiral outer surface, indented, braided, or with ribs) gain their
GG-15.9-5d
20
GG-15.9-10d
GG-15.9-15d
10
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 03/29/16. Copyright ASCE. For personal use only; all rights reserved.
c
Experimental tensile strength and elastic modules calculated using nominal bar diameter.
d
Effective values (including sand coating).
manufactured by using the pultrusion process and are reinforced by fabrication, ease of repeatability, and most importantly, its satisfac-
continuous E-glass fibers with a minimum volume of 65%, whereas tory results when used for comparative purposes. This test is mainly
the binding material is modified vinyl ester with a maximum vol- used to compare the bond strength of different materials under
ume of 35%. Diameter, tensile strength, elastic modulus, and failure similar conditions. It is composed of a concentrically placed rein-
strain of the bar were tested according to ACI 440.3R-04 (ACI forcement bar in a concrete cylinder. Each bar was embedded in
2004). The mechanical and physical properties of the bar are sum- a 100 × 170-mm concrete cylinder, and the embedment lengths
marized in Table 1. were three, six, and nine times the bar diameter. All the bars were
The mix proportions of the GPC and OPC concretes are given in 1,000 mm long with a special anchorage on one end to avoid pre-
Table 2. The desired strengths of GPC concrete were attained by mature failure of the GFRP bars caused by the grip. To achieve the
varying the curing period. Class F fly ash and general-purpose ce- desired embedment length, contact between the concrete and the
ment complying with ASTM C618 (ASTM 2012) and AS 3972 bar was avoided through the use of PVC tubing, as shown in Fig. 3.
(AS 2010) were used for GPC and OPC concretes, respectively. Three nominally identical specimens were tested for each embed-
The alkaline liquid used for GPC concrete was a combination ment length. Cylindrical steel molds with plywood top and bottom,
of sodium hydroxide and sodium silicate solutions. The sodium shown in Fig. 3, were used to cast the concrete around the bar. In
hydroxide solution had a concentration of 12 M and was prepared the GPC concrete mixes, a thin sheet of plastic is laid inside the
by dissolving 98% caustic soda flakes in distilled water to the pre- steel mold for ease of demolding.
scribed molarity. The SiO2 =Na2 O ratio of the sodium silicate The mixing and curing of GPC concrete was based on previous
solution used was 1.95–2.05. The nominal maximum coarse aggre- research (Junaid et al. 2015a). The pullout specimens were cast in a
gate size was 14 mm, and the fine aggregate size interval was vertical position and vibrated with a poker vibrator in two layers.
0–4.75 mm. The aggregates were brought to saturated surface dry Standard 100 × 200-mm concrete cylinders were also cast for com-
(SSD) condition before mixing the concrete by adding the SSD pressive [ASTM C39/C39M (ASTM 2015a)], elastic modulus
water, which is determined from the absorption and the moisture [ASTM C469/C469M (ASTM 2014)], and splitting tests [ASTM
content tests described in ASTM C127 (ASTM 2015b) and ASTM
C496/C496M (ASTM 2011)]. The specimens were then allowed
C566 (ASTM 2013), respectively. Midrange water-reducing ad-
to rest for 24 h, after which they were placed to cure in an oven
mixture and viscosity modifier were used to improve the workabil-
ity of the GPC concrete. The workability was measured by using a
standard slump test in ASTM C143/C143M (ASTM 2015c), and a
medium workability was obtained according to the classification of
GPC concrete workability in Junaid et al. (2015b).
Preparation of Specimens
There are different tests for investigating the bond behavior of
reinforcing bars with concrete. Pullout tests, beam-end tests, beam
anchorage, and splice tests are the most commonly used. In this
program, the pullout specimen test is used because of its ease of
FRP-reinforced GPC concrete and FRP-reinforced OPC concrete plotted in Figs. 6–9. The bond-slip curves display the characteristic
(Maranan et al. 2015; Okelo and Yuan 2005; Cosenza et al. stages of the typical bond-slip curve shown in Fig. 1. For low val-
1997). For the same embedment length and diameter, a change ues of bond stress, no significant slip is observed resulting in a
in concrete compressive strength resulted in a change in failure steep initial slope. Once the failure occurred, i.e., at the maximum
mode. G13-6d1 specimens failed by pullout caused by their lower bond stress point, softening of the bond stress follows. This is ac-
compressive strength, whereas G13-6d2 specimens showed a split- companied by a large slip at both free and loaded ends. This divides
ting failure mode. This shows the dependence of failure mode on the bond-slip curve into linear, nonlinear, and descending branch.
compressive strength. Smaller bar diameters were also found to re- The high initial stiffness observed is caused by the adhesion and
sult in pullout failure because of the relatively higher amount of mechanical interlock between the concrete and the bar. The slip
confinement provided in smaller bar diameters. When the confine- at the free end lags that of the loaded end, showing the nonlinear
ment provided by the concrete is not enough to support the radial bond stress across the embedment length. The loaded-end bond-
stresses generated by the bond, longitudinal cracks that initiate at slip curve shows a lower stiffness than the unloaded-end curve
the interface propagate to the surface of the concrete, resulting in (Figs. 6–9). The difference represents the lengthening of the
splitting failure. Both GPC and OPC concrete specimens’ splitting reinforcing bar between the two points of slip measurements. As
failures were observed to be brittle. This was also reported by Cui the pullout load increases, the stiffness of the curve reduces, and
and Kayali (2013), Sarker (2011), and Sofi et al. (2007).
nonlinear response is observed. This continues up to the peak bond
Furthermore, a close investigation of the failure faces (Fig. 5)
stress. This lower slope is caused by the progress of bond damage
revealed that the GFRP bar was covered with a thin layer of crushed
as the pullout load increases. At this stage, the friction and the
concrete in both the OPC and GPC concrete specimens, and no
mechanical interlock are responsible for the bond strength. After
peeling off of the sand coating is observed.
this, either pullout (Fig. 6) or splitting failure (Figs. 8 and 9) occurs
depending on the specimen. When enough confinement from the
Bond-Slip Relationship Curves surrounding concrete, the bond-slip curve will continue until shear
The median curves from groups of three identical samples are cracks are initiated between the concrete and the bar. Once the peak
chosen to be representative of the bond-slip behavior and are bond stress is reached, for pullout failure mode, friction reduces as
Fig. 5. Failure faces: (a) O16-6d2-1 concrete surface after splitting failure; (b) G16-6d3-1 concrete surface after splitting failure; (c) bar surface of
G16-3d3-1 (pullout failure specimen) after hand splitting
20 20
18 18
16 16
14 14
12 12
10 10
8 8
G16-3d3-1 free end G16-6d3-1 free end
6 G16-3d3-1 loaded end 6
G16-9d3-3 free end
4 O16-3d2-2 free end 4 G16-6d3-1 loaded end
O16-3d2-2 loaded end G16-9d3-3 loaded end
2 2
0 0
0 2000 4000 6000 0 200 400 600 800 1000
Slip (micron) Slip (micron)
Fig. 6. Bond stress–slip curve for 15.9-mm 3d specimens Fig. 8. Bond stress–slip curve for 15.9-mm 6d and 9d GPC concrete
specimens
20
18 18
16 16
Bond stress (MPa)
14 14
Bond stress (MPa)
12 12
10 10
8 G13-6d2-2 free end 8
6 G13-6d2-2 loaded end 6 O16-6d2-2 free end
4 O13-6d1-2 free end O16-9d2-1 free end
O13-6d1-2 loaded end 4 O16-6d2-2 loaded end
2
2 O16-9d2-1 loaded end
0
0
0 1000 2000 3000 4000 5000 6000 0 200 400 600 800 1000
Slip (micron) Slip (micron)
Fig. 7. Bond stress–slip curve for 12.7-mm 6d specimens Fig. 9. Bond stress–slip curve for 15.9-mm 6d and 9d OPC concrete
specimens
0 5 10 15 20 ment did not show the same trend. However, the results of this ex-
Loaded end slip (mm)
periment were not sufficient to quantify this behavior of the bond.
Fig. 10. Extended bond stress–slip curve for 15.9-mm GPC concrete The observed splitting failure of some of the specimens also shows
3d specimens the dependence of the bond strength on the compressive strength
as splitting failure cannot happen unless the tensile strength of the
concrete is reached, which in turn depends on the compressive
strength of the concrete. In case of nondependence on compressive
20
strength, like in some deformed FRP bars where the indentation on
18
the bar is relatively weak, the bond collapses because of detachment
16 of the ribs or spirals and the concrete remains uncrushed (Cosenza
Bond stress (MPa)
14 et al. 1997).
12 O16-3d2-1 Bond strength decreases with the increase of bar diameter. This
10 O16-3d2-2 tendency is observed by various researchers in GFRP bars with
O16-3d2-3
8 OPC concrete (Baena et al. 2009; Cosenza et al. 1997; Okelo and
6 Yuan 2005). The failure mode difference between O13-6d1 and
4
O16-6d2 shows the effect of bar diameter on bond behavior of
the specimens. The nonlinear bond stress distribution along the em-
2
bedment length was given as the possible cause of this behavior
0
(Lee et al. 2013; Baena et al. 2009). Poisson effect has also been
0 5 10 15 20
Loaded end slip (mm) suggested as another factor for the reduction of bond strength as bar
diameter increases. As the bar is subjected to pullout stress, a slight
Fig. 11. Extended bond stress–slip curve for 15.9-mm OPC concrete reduction in diameter occurs. This reduction in the bar diameter
3d specimens increases with diameter and leads to the reduction of friction and
mechanical locking stress between the bar and the concrete.
softening branches of the mBPE model are given in Eqs. (5) and bar. Eq. (7) is valid if the GFRP bar has a linear elastic constitutive
(6), respectively. The CMR model, given in Eq. (1), only describes law in the longitudinal direction and the displacement of the con-
the ascending branch of the bond-slip curve because most structural crete points at the interface between the concrete and GFRP bar are
problems are dealt with at the serviceability limit state level. These negligible compared with the displacements of the bar points;
analytical models contain unknown parameters, which are cali- i.e., the deformation of the concrete are neglected, and thus sðxÞ
brated by using the experimental test results. The constitutive will be the displacement of the GFRP bar at x (Focacci et al. 2000).
law developed in such a way can then be used to carry out a numeri- By integrating Eq. (7), a relationship between slip and tensile
cal simulation of the experimental tests. stress can be obtained. A closed-form solution of this differential
The modified version of the mBPE law is expressed as equation can be determined only if the embedment length of the bar
α is longer than the development length, in which case the free-end
s slip will be zero. This equation, as obtained by Pecce et al. (2001),
τ ðsÞ ¼ τ m : ð5Þ is as follows:
sm
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8·E
τ ðsÞ ¼ τ m · 1þp−p
s
ð6Þ σðsl Þ ¼ · Aτ ðsl Þ ð8Þ
sm db
where τ m = maximum bond stress; sm = slip at the maximum bond where Aτ ðsl Þ = area under the τ − s curve. As stated previously,
stress; and α, p, β, and sr = unknown parameters. The CMR law is Eq. (8) can only be used if the free-end slip is zero. However, as
given in Eq. (1). the free-end slip is greater than zero for all the specimens in the
The parameters for the bond-slip constitutive laws are deter- present case, the closed-form solution cannot be used. Therefore,
mined by assuming a constant bond stress distribution along the Eq. (7) is integrated by using a numerical method with the boun-
embedment length of the bar and measuring the slip at the free dary conditions as shown in Eqs. (9) and (10)
end. This procedure works well for steel bars because of the neg- ds
ligible difference between the slips at the loaded and free end; ¼ εðxÞ; sð0Þ ¼ sf ð9Þ
dx
hence, the constant bond stress distribution is acceptable. However,
for GFRP bars, there is a significant difference between the slips at dε 4
the two ends because of the lower elastic modulus of the bar. These ¼ · τ ½sðxÞ; εð0Þ ¼ 0 ð10Þ
dx E · db
results in two bond-slip curves for each test: free- and loaded-end
curves. Thus, the procedure of finding a constitutive law for the where ε = strain along x; and sf = slip at the free end. With these
bond strength of GFRP bars is not as straightforward as for steel boundary conditions and a set of assumed tentative values for the
bars (Pecce et al. 2001). parameters (τ m , α, sm , and p for mBPE and τ m , sr , and β for
The differential equation of the problem of a reinforcing bar em- CMR), Eq. (7) is solved by using Euler’s method. For each loading
bedded in a concrete block can be derived by considering an infini- step of the experiment, the theoretical values of slip and strain at the
tesimal segment of embedded reinforcing bar as shown in Fig. 12. loaded end of the embedment length of the bar are estimated. These
The bond stress along the bar is given by τ ¼ τ ðsÞ, where τ = shear theoretical values of slip and stress at the loaded end are compared
stress acting on the contact surface between the bar and the con- with experimental values to obtain the optimal values of the param-
crete; and s ¼ sðxÞ = relative displacement between the bar and the eters of the bond-slip relationship. The error function, Eq. (11),
used by Pecce et al. (2001), is used to optimize these parameters
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
th th ffi
u
uPn σl − σel 2 sl − sel 2
u i¼1 þ
t σel sel
e¼ ð11Þ
n
where N th th e e
l , sl , N l and sl = theoretical and experimental load and
slip at the loaded end, respectively; and n = number of loading
steps, i.e., number of data points from the experiment. The afore-
mentioned procedure is applied for all the specimens, and the re-
sults are summarized in Tables 5 and 6.
The mBPE and CMR bond-slip relationships determined by
averaging the parameters for each set of specimens are also re-
ported in Tables 5 and 6. The average parameters for all specimens,
except G13-6d1 specimens, are reported in Table 7 (only the as-
Fig. 12. Infinitesimal segment of anchored GFRP
cending part is reported for mBPE). These values represent the
bond-slip laws resulting from the average behavior of all the con- Table 7. Average Bond-Slip Parameters
sidered specimens. G13-6d1 specimens (24.66 MPa compressive
mBPE CMR
strength) are not included in the average bond-slip parameter cal-
culation because of the much lower value of the bond strength, Specimen τ m (MPa) sm (mm) α τ m (MPa) sr β
which comes from the difference in compressive strength. This GPC concrete 17.3 0.14 0.49 17.6 0.03 1.23
introduces higher variation to the bond-slip parameters, especially OPC concrete 15.2 0.18 0.28 15.2 0.03 0.71
the maximum bond strength. Thus, the parameters of these spec-
imens are measured separately from the higher-strength specimens
(45 MPa compressive strength on average). Furthermore, as shown in between the free-end and loaded-end bond-slip curves for both
in Tables 5 and 6, different bond-slip relationships are found from numerical and experimental results.
different specimens because of the difference in bar diameter and The developed CMR bond-slip law appeared to be more close
embedment length. As a result of these differences, the average to the free-end curve than that of the loaded end for both GPC and
bond-slip law (Table 7) only gives general indications on the bond- OPC concretes. In addition, in terms of shape, the numerical re-
slip behavior of the GFRP bar. sults from the CMR law were found to be more accurate than those
By using these bond-slip law, the numerical values of the from the mBPE. This is shown in Figs. 13 and 14, in which the
loaded-end slip and the stress at the loaded end, which are then CMR constitutive law shows a decrease in slope and becomes
used to calculate the average bond stress along the embedment almost zero as it approaches the peak bond stress, which is also
length, are determined. The numerical and experimental bond- observed in the experimental results. Conversely, the mBPE still
slip relationships are shown in Figs. 13 and 14. For most of the has a significant slope around the peak bond stress. As for the ex-
ascending branch, the bond-slip constitutive law appears to be perimental results, the average bond stress of GPC concrete was
Fig. 13. Numerical and experimental bond-slip curves with the bond-slip law: (a) GPC-mBPE; (b) GPC-CMR
Fig. 14. Numerical and experimental bond-slip curves with the bond-slip law: (a) OPC-mBPE; (b) OPC-CMR
found to be greater than that of OPC concrete in both mBPE and reinforcement are different, which causes loss of bond. In such sit-
CMR models. uations, the real bond behavior should be modeled separately. In
Further verification of the constitutive bond-slip law is done by ABAQUS, this can be done either by using cohesive elements or
finite-element modeling in the next section. The bond-slip constit- surface-based cohesive behaviors. In this study, the bond is mod-
utive law is used with other material models to simulate the pullout eled by using surfaced-based cohesive behavior because the inter-
specimens. face thickness is negligibly small. This behavior assumes a linear
elastic traction-separation law before damage and also assumes fail-
ure of the cohesive bond to be characterized by progressive deg-
Finite-Element Modeling radation of the cohesive stiffness, which is driven by a damage
process. Because the loading, boundary, and material properties
offer symmetry of revolution as shown in Fig. 15, the problem
Description of the Model
can be modeled as axisymmetric. A meshed model consisting of
The experimental results only provided data on what happened out- four-node bilinear axisymmetric quadrilateral elements with a re-
side the embedment length. These include free-end slip, loaded-end duced integration and hourglass control (CAX4R) is also shown on
slip, and the load value. The actual bond development along the Fig. 15. The same element is used for both the concrete and the
embedment length can only be understood through modeling. For reinforcement.
this purpose, the bond between the reinforcement and the concrete
is modeled through a bond-slip relationship to define its behavior.
Concrete and Reinforcement Models
In a perfect bond situation, the nodes of the reinforcement elements
and of the concrete elements displace by the same amount acting as The concrete damage plasticity model (CDPM) available in
a single material. However, this is only valid in areas with low ABAQUS is used to model the concrete. As the complete stress-
stress transfer. In high-stress areas, the strain in the concrete and the strain behavior of the geopolymer concrete is not available, it is
Fig. 15. Pullout model: (a) full model; (b) meshed axisymmetric model Bond-Slip Interaction
As described previously, the bond-slip behavior in this model is
Downloaded from ascelibrary.org by UNIVERSITY OF NEW SOUTH WALES on 03/29/16. Copyright ASCE. For personal use only; all rights reserved.
50
45
40
Traction
35
Stress (MPa)
30
25
20
15
10
5
0 0
0 0.002 0.004 0.006 0.008 0
Strain Separation
Fig. 16. GPC concrete (49.33 MPa) stress-strain curve Fig. 17. ABAQUS typical traction-separation response
40%Pmax
0.004 and free ends, and the bond-slip relationship curves were deter-
60%Pmax
mined. Furthermore, analytical and numerical analyses were made
0.003 80%Pmax
to determine a bond-slip relationship and investigate the bond dis-
90%Pmax tribution along the embedment length. The results of this study can
0.002
100%Pmax be summarized as follows:
0.001 1. The bond-slip behavior of sand-coated GFRP-reinforced GPC
concrete is similar to that of OPC concrete reinforced with
0 FRP or steel in the sense that it display similar stages consisting
0 20 40 60 80 100 of initial stiffening, nonlinear behavior before peak stress, and
Distance from free-end (mm) softening when pullout failure takes place;
Fig. 19. Modeling result for strain distribution along the embedment 2. The bond performance of sand-coated GFRP-reinforced GPC
length for G16-9d3 specimen concrete is better than that of the OPC concrete counterpart.
This is shown by the higher failure loads and bond strengths
of GFRP-reinforced GPC concrete specimens. For all the speci-
20 mens involved, average bond strengths of 15.8 MPa for GPC
concrete and 14.3 MPa for OPC concrete were found;
18
3. Pullout load increases with increasing embedment length, but
16 20%Pmax the average bond strength decreases because of the splitting fail-
14 40%Pmax ure mode and the nonlinear distribution of the bond along the
Bond stress (MPa)