You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272153288

Serviceability Analysis of Concrete Beams with Different Arrangements of GFRP


Bars in the Tensile Zone

Article in Journal of Composites for Construction · October 2014


DOI: 10.1061/(ASCE)CC.1943-5614.0000465

CITATIONS READS

51 520

5 authors, including:

Ronaldas Jakubovskis Gintaris Kaklauskas


Vilnius Gediminas Technical University Vilnius Gediminas Technical University
43 PUBLICATIONS 455 CITATIONS 182 PUBLICATIONS 2,397 CITATIONS

SEE PROFILE SEE PROFILE

Viktor Gribniak André Weber


Vilnius Gediminas Technical University Schöck Bauteile GmbH
171 PUBLICATIONS 2,389 CITATIONS 38 PUBLICATIONS 168 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Viktor Gribniak on 11 February 2015.

The user has requested enhancement of the downloaded file.


Serviceability Analysis of Concrete Beams with Different
Arrangements of GFRP Bars in the Tensile Zone
Ronaldas Jakubovskis 1; Gintaris Kaklauskas, Ph.D. 2; Viktor Gribniak, Ph.D. 3;
André Weber, Ph.D. 4; and Mantas Juknys 5

Abstract: Extensive experimental investigations were carried out to investigate the serviceability (deformations and crack opening width) of
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

fiber–reinforced polymers (FRP) reinforced concrete members. However, most of the tests were limited to conventional arrangement of bars
in a section. Considering glass fiber–reinforced polymer (GFRP) bars (one of the most frequently used types of composite reinforcement), this
study experimentally and analytically investigated the serviceability behavior of beams with different arrangements of reinforcement bars in
the tensile zone. With particular emphasis on tension-stiffening and bond behavior, the experimental program consisted of full-scale beams
and bending bond specimens. The tests revealed specific features in the behavior of the beams due to distribution of tensile bars in three layers
or enlarging the cover. A numerical procedure, based on local interaction of reinforcement and surrounding concrete, was applied for serv-
iceability analysis. Several bond stress-slip models were employed in the numerical analysis. Despite the satisfactory prediction of crack
opening and deformations in general, neither of the bond models was able to give accurate results for all the beams. DOI: 10.1061/(ASCE)
CC.1943-5614.0000465. © 2014 American Society of Civil Engineers.
Author keywords: Reinforced concrete; Glass fiber–reinforced polymer (GFRP) bars; Experimental investigation; Cracking; Bond models.

Introduction techniques have been proposed for predicting the deformation


and cracking response of such members with comprehensive
Since the early 1980s, fiber–reinforced polymers (FRP) have been reviews carried out by Bischoff (2005, 2007) and Torres et al.
considered a promising alternative to steel reinforcement, espe- (2012). The observed inaccuracies in the predictions were mainly
cially in concrete structures subjected to aggressive environments attributed to the simplicity of the empirically deduced models.
or to the effects of electromagnetic fields (Torres et al. 2012). With In contrast to the empirical models, numerical simulations based
high durability, FRP bars have a tensile strength up to 5–6 times on the stress transfer (also known as force transfer, partial inter-
higher than structural steel. However, the low (in comparison to the action, or discrete crack in finite element formulation) approach
steel) elastic modulus of some types of polymer bars generally most realistically deal with discrete cracking phenomena and are
leads to increased deformations of FRP reinforced concrete (RC) capable of adequately reflecting the complex nature of stress redis-
elements. This is characteristic of glass fiber–reinforced polymer tribution between cracked concrete and reinforcement. This ap-
(GFRP) bars as one of the most frequently used types of structural proach can be universally applied to any type of reinforcement
reinforcement (Mias et al. 2013; Mahmoud and El-Salakawy bars (steel or FRP) and concrete (fiber–reinforced, high perfor-
2013). Decreased sectional stiffness increases the importance of mance) (Lackner and Mang 2003; Wu and Gilbert 2009).
serviceability conditions (limitation of deformations and crack As Fig. 1 shows, bond mechanics can be investigated at different
opening width) in design of GFRP RC elements. A number of scales of consideration. Initially, stress transfer models were devel-
oped using a simplified assumption of constant bond stresses
1
Ph.D. Student, Dept. of Bridges and Special Structures, Vilnius (Floegl and Mang 1982). Later, this approach was extended by
Gediminas Technical Univ., Sauletekio av. 11, 10223 Vilnius, Lithuania. Choi and Cheung (1996), who introduced a stepwise bond stress
E-mail: Ronaldas.Jakubovskis@vgtu.lt distribution relating average bond stresses with bond strength. This
2
Habilitated Doctor of Science, Professor, Head of Dept. of Bridges and assumption was used by Marti et al. (1998) who developed the
Special Structures, Vilnius Gediminas Technical Univ., Sauletekio av. 11, tension chord model and applied it in the serviceability analysis
10223 Vilnius, Lithuania (corresponding author). E-mail: Gintaris of RC. Assumption of constant bond stress distribution along an
.Kaklauskas@vgtu.lt uncracked concrete block simplifies the solution of the bond
3
Senior Researcher, Civil Engineering Research Centre, Vilnius
modeling problem, however a rigorous formulation of the stress
Gediminas Technical Univ., Sauletekio av. 11, 10223 Vilnius, Lithuania.
E-mail: Viktor.Gribniak@vgtu.lt transfer approach requires the bond to be dependent on slip.
4
Doctor of Engineering, Head of Research and Development Division Somayaji and Shah (1981) assumed exponential bond stress distri-
of Glass Fiber Reinforcement, Schöck Bauteile GmbH, Vimbucher st. 2, bution function and developed an analytical procedure predicting
76534 Baden-Baden, Germany. E-mail: Andre.Weber@Schoeck.de the deformation and cracking process of tensile RC specimens.
5
Ph.D. Student, Dept. of Bridges and Special Structures, Vilnius Gedi- Balazs (1993) analyzed the cracking behavior of RC elements in
minas Technical Univ., Sauletekio av. 11, 10223 Vilnius, Lithuania. E-mail: crack formation and stabilized cracking phases using a nonlinear
Mantas.Juknys@vgtu.lt
bond stress-slip relationship recommended by Model Code
Note. This manuscript was submitted on July 30, 2013; approved on
December 12, 2013; published online on February 3, 2014. Discussion 1990. This model is also included into the newly issued Model
period open until July 3, 2014; separate discussions must be submitted Code 2010 (Balazs et al. 2013). Its application in serviceability
for individual papers. This paper is part of the Journal of Composites problems has been criticized by Wu and Gilbert (2009), who have
for Construction, © ASCE, ISSN 1090-0268/04014005(10)/$25.00. shown that the Model Code bond law does not always

© ASCE 04014005-1 J. Compos. Constr.

J. Compos. Constr.
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Different scales of bond modeling

guarantee satisfactory results, especially at higher loading stages, increasing cover and/or introducing a few layers of reinforcement
when degradation of tension-stiffening related to secondary crack- with closely distributed bars in the tensile zone.
ing and, consequently, deterioration of the bond takes place. This Although extensive experimental programs have been carried
effect is considered in the bond model proposed by Shima et al. out to investigate the serviceability of FRP reinforced members, most
(1987), which takes into account not only slip, but also strain in of the tests were limited to conventional arrangement of bars in a
the reinforcement. section. Focusing on GFRP reinforcement, the present study exper-
In most cases, ties were used to assess bond properties of RC imentally and analytically investigated the moment-curvature and
elements. Only a few studies were dedicated to bending members. cracking behavior of concrete beams with nonconventional arrange-
In this respect, the research conducted by Fantilli et al. (1998) and ment of reinforcement bars in the tensile zone. Experimental data of
Chiaia et al. (2009) can be mentioned. It should be also pointed out three GFRP RC beams with different concrete cover and distribution
that most of the proposed bond models were derived from the pull- of bars in a section are reported. A numerical procedure based on the
out tests by using a single well-confined bar. However, such con- stress transfer approach was applied to investigate the evolution of
ditions substantially differ from the tension zones of RC bending deformations and crack patterns with increasing load. Through
elements as the nominal cover required in the design codes may comparison of the numerical simulation results with experimental
be insufficient to secure effective bond-confinement (Hong and data, the applicability of widely used bond models for serviceability
Park 2012). analysis of GFRP reinforced members has been investigated.
The arrangement of reinforcement bars in a section might be
also an important but frequently neglected parameter controlling
structural stiffness. Extensive experimental investigations by the Stress Transfer Modeling
authors of bending members with different types of reinforcement
(Kaklauskas et al. 2012; Gribniak et al. 2013) have shown that the Present serviceability analysis of GFRP reinforced beams employs
tension-stiffening effect might be significantly enhanced by the stress transfer approach using the formulation proposed by

(a) (b)

Fig. 2. Deformation and cracking analysis of RC beam using stress transfer approach: (a) a block between two adjacent cracks; (b) sketch for the
analysis

© ASCE 04014005-2 J. Compos. Constr.

J. Compos. Constr.
Fig. 3. Stress and strain distribution in a RC section subjected to bending: (a) variation of neutral axis within the block section; (b) distribution of
strains and stresses in the cracked section; (c) uncracked section

Floegl and Mang (1982) and Fantilli et al. (1998). A beam shown in of tensile and compressive reinforcement, respectively; σc ðεÞ is the
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2(a) helps to illustrate this approach. Before a cracking load is compressive concrete stress; and σct is the tensile concrete stress.
reached, the linear elastic behavior of the beam with a perfect bond Other notations are evident from Fig. 3.
between reinforcement and concrete can be assumed. After crack The stress-strain state in each section can be specified by three
initiation, redistribution of internal forces takes place. In a cracked parameters, namely, by the distance yc and strains in the tensile
section, denoted as 1-1 in Fig. 2(a), the tensile force is carried by reinforcement and the tensile concrete (at reinforcement level).
reinforcement only. Due to bond action at a certain distance from As Eqs. (1) and (2) allow the definition of only two parameters,
the crack, reinforcement transfers a certain portion of tensile stress an additional equilibrium equation is written for the reinforcement
to the concrete. This phenomenon is known as tension- bar limited by two adjacent sections i and i þ 1. This equation
stiffening. The distance from the crack to the section where the relates the increment in reinforcement stress with the bond stress
bond transferred stress in concrete reaches the tensile strength is acting in the segment [Fig. 2(b)]
called transfer length and is denoted as ltr . Obviously, a new crack
can form only at a distance l ≥ ltr from an existing crack. Assuming 4 · n · Δx · τ̄ i
εf;i ¼ εf;i−1 þ ð3Þ
heterogeneous distribution of strength properties of concrete, D · Ef
the distance between cracks lcr falls into the specific interval
where n is the number of tensile bars; D is the bar diameter; Ef is
ltr ≤ lcr ≤ 2ltr . It has been reported by Borosnyoi and Balazs
the elasticity modulus of reinforcement; and τ̄ i is the average bond
(2005) that average crack spacing varies from 1.3ltr to 1.5ltr .
stress calculated as τ̄ i ¼ ðτ i þ τ i−1 Þ=2.
The algorithm is based on further assumptions and principles:
The solution of Eqs. (1)–(3) is not straightforward, as bond
1. Cracking bending moment M cr is calculated using linear elas-
stress τ̄ i is related to an unknown parameter, namely reinforcement
tic material properties under the assumption of perfect bond
strain εf;i . Therefore, analysis has to be performed iteratively,
between reinforcement and concrete;
assuming initial value for one of the unknown parameters.
2. After cracking, two characteristic sections are considered
The numerical analysis is carried out in two stages. During the
[Fig. 3(a)]: cracked (1-1) and uncracked (2-2). For both sec-
first, transfer length and distance between cracks are defined. The
tions, linear strain distribution, being valid for compressive
second stage deals with calculation of deformation and crack width
concrete and tensile reinforcement, is assumed. In a cracked
using the transfer length obtained in the first stage. For a given load,
section [Fig. 3(b)], contribution of tensile concrete due to
computations are performed through the following steps:
tension-softening is ignored as the stresses transmitted through
1. Calculation of transfer length (stage 1) is started from the cen-
the crack plane are negligible in comparison to the stresses
ter section of an indefinitely long element (or a block limited
transferred through the bond. In a section at a certain distance
by two adjacent cracks) assuming three initial conditions: slip
from the crack, strain distribution of the tensile concrete is also
and bond stress taken in the center section are set equal to zero
taken linear, but with a different profile from the strains of
[s0 ¼ 0; τ 0 ¼ 0, Fig. 2(b)], whereas concrete stress is assumed
compressive concrete and tensile reinforcement [Fig. 3(c)];
to be equal to the tensile strength σct ¼ fct . Solution of the
3. Linear elastic stress-strain diagram is assumed for reinforce-
equilibrium Eqs. (1) and (2) results in reinforcement strain
ment at all loading stages; and
εf;0 [Fig. 2(b)] and the neutral axis coordinate yc (Fig. 3)
4. Concrete in tension behaves in an elastic-brittle manner.
in the mid-section;
Nonlinear behavior of concrete in compression is modeled
2. With the strains in the mid-section (section number i ¼ 0),
using the stress-strain relationship recommended by Model
the analysis progresses to each consecutive section [i ¼ 1,
Code 2010.
2, 3, : : : , see Fig. 2(b)]. Assuming a trial value of reinforce-
To calculate stress and strain variation along the element, it is
ment strain [εf;i ¼ εf;i–1 , see Fig. 2(b)], coordinate yc;i and
divided into a number of sections [Fig. 2(b)]. Two equilibrium
tensile concrete stress σct;i are obtained from Eqs. (1) and (2);
equations of internal forces and bending moments can be written
3. Reinforcement slip si is determined as the difference in con-
for each of the sections:
crete and reinforcement elongations between two consecutive
Z sections, as shown in [Fig. 2(b)];
b σc ðεÞdy þ σs1 As1 þ σs2 As2 ¼ 0 ð1Þ 4. Bond stress τ i ðsi Þ is obtained from the assumed bond stress-
Ac
slip relationship (for instance, as shown in Fig. 1);
Z 5. Reinforcement strain εf;i is calculated using Eq. (3);
6. If the difference between the previously calculated (or
b σc ðεÞydy þ σs1 As1 ðd − yc Þ þ σs2 As2 ðyc − a2 Þ ¼ M ð2Þ
Ac
assumed) and current reinforcement strains in section i is
not within the assumed tolerance, i.e., convergence is not
where M is the external bending moment; yc is the distance from achieved, the analysis is repeated from step 2 using the current
the top of the section to the neutral axis; σs1 and σs2 are the stresses reinforcement strain value. Otherwise, the analysis from

© ASCE 04014005-3 J. Compos. Constr.

J. Compos. Constr.
step 2 is carried out in other sections (i ¼ i þ 1) until reinfor- Model Code model based on bond test results of GFRP reinforce-
cement strain εf;n reaches εf;cr [Fig. 2(b)]. The latter is de- ment reported in the next section.
fined as for a fully cracked section [i.e., assuming σct ¼ 0 The bond model proposed by Shima et al. (1987) can capture
in Eqs. (1) and (2)]. The corresponding section identifies bond degradation due to secondary cracking and concrete damage.
the transfer length calculated as ltr ¼ ðn × ΔxÞ; The effect of secondary cracking is related to the reinforcement
7. Stage 1 is completed by assuming crack distance lcr;avg ¼ strain. The following expression relating bond stress, slip, and
1.5ltr for deflection and lcr;max ¼ 2.0ltr for crack width reinforcement strain has been proposed by Shima et al. (1987):
analysis;
8. Stage 2 analysis is performed for a block of specified length, 0.73 · ½lnð1 þ 5SÞ3
τ ðs; εÞ ¼ · fcm ð7Þ
lcr (see step 7), divided into a number of sections. Computa- 1 þ ε · 105
tions are performed similarly as for stage 1 with the difference
that they are started from the cracked section. Assuming where S is the nondimensional slip (S ¼ 1,000 × s=D; s is the ac-
εf;0 ¼ εf;cr , σct ¼ 0, and initial slip value s0 ¼ sn (for in- tual slip; and D is the bar diameter). The relationship Eq. (7) has
stance, taking the slip from stage 1 analysis), calculations been derived not only for steel, but also for aluminum bars having
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

are further progressed to the block center. Computations are low modulus of elasticity (72 GPa).
performed iteratively (changing s0 ) until the no-slip condition
(si ¼ 0) is satisfied for the block mid-section;
9. From the obtained strain profile, average curvature of the Experimental Study
block [Fig. 2(b)] is calculated
The experimental study was aimed at deformation, cracking, and
X
n
εf;i bond behavior of GFRP RC flexural members. The test program
κ¼ ð4Þ consisted of three full-scale beams and 12 bending bond elements.
i¼1
d − yc;i
The members were reinforced with Schöck Combar ribbed GFRP
bars with 75% of glass content by volume. The bars were produced
where εf;i is the average strain in the tensile reinforcement; yc;i
from boron-free ECR glass fibers and synthetic vinyl ester-urethane
is the coordinate of neutral axis; n is the number of sections
(VEU) resin (Schöck 2006). Three different diameters (8, 12, and
[Fig. 2(b)]; and
16 mm) of the reinforcement were used (Fig. 4).
10. Crack width is calculated by the following equation:
The full-scale beams were of a rectangular section, with nominal
Z l Z l length 3,280 mm (span 3,000 mm), depth 300 mm, and width
cr cr
w¼ ½εf ðxÞ − εct ðxÞdx ≈ εf ðxÞdx ð5Þ 280 mm. The concrete mix proportion of C35/45 grade was taken
0 0
the same for all specimens. Ordinary Portland cement and crushed
where εf ðxÞ and εct ðxÞ are the strain distribution in GFRP re- granite aggregate (16 mm maximum nominal size) were used. The
inforcement and in tensile concrete, respectively. The latter is water/cement and aggregate/cement ratio by weight were taken as
negligible in respect to εf ðxÞ and can be neglected. 0.42 and 2.97, respectively.
The calculated transfer length ltr is load-dependent. Depending Table 1 lists the main parameters of the test beams used in
on a bond law, it may decrease or increase with loading. However, the analysis, where ffk and f fd are, respectively, the characteristic
due to the physical nature of the cracking phenomenon, crack dis- and design tensile strength of the reinforcement; Ef is the elastic
tance cannot increase with growing load. Therefore, the present modulus of the bar reinforcement; f cm is the mean ∅150 × 300 mm
algorithm imposes a condition that a crack distance at a given load cylinder strength; and p ¼ ½As1 =ðbdÞ × 100% is the tensile rein-
stage cannot exceed the minimum crack spacing obtained in the forcement ratio. Other notations are evident from Fig. 3. The beams
previous stages. had similar reinforcement ratios, but different bar diameters and
number of tensile reinforcement layers. Bars in beam S2-5nm were
arranged in three layers (Fig. 5) whereas the other two beams had a
single reinforcement layer. The effective depth d (Table 1) was
Bond Models Employed in the Analysis taken with respect to the centroid of the reinforcement area. The
It is generally accepted to assume bond stress τ being dependent on experimental program has also included variation in cover depth
slip s that represents the so-called bond stress-slip relationship taken as 20 mm for beams S2-4-2nm and S2-5nm and 50 mm
τ ðsÞ. Most of the refined analyses based on the stress transfer ap- for beam S2-6nm.
proach employ the bond model from Model Code 2010 derived for Being part of a larger test program (Kaklauskas et al. 2008)
steel bars. The present study investigates its applicability for serv- supported by the Research Council of Lithuania, the present study
iceability analysis of GFRP reinforced beams. As in deflection and
cracking analyses slip is generally below 1 mm, only the ascending
branch of bond stress-slip relationship becomes relevant (fib 2000):
 α
s
τ ðsÞ ¼ τ max · ð6Þ
s1

where τ max is the maximal bond stress [τ max ¼ 2.5 × ðfcm Þ0.5 ]; α is
the power (α ¼ 0.4); and s1 is the slip value corresponding to the
maximal bond stress (in this study s1 ¼ 1 mm).
Depending on the power α, the bond model Eq. (6) allows for
simulating a wide range of relationships: from fully rigid (α ¼ 0) to
linear (α ¼ 1) bond stress-slip behavior. The present study, along
Fig. 4. Samples of experimental bars
with the original Model Code 2010 bond law, use the calibrated

© ASCE 04014005-4 J. Compos. Constr.

J. Compos. Constr.
Table 1. Main Characteristics of Beams
Beam h (mm) b (mm) d (mm) a2 (mm) As1 (mm2 ) As2 (mm2 ) ffk (MPa) f fd (MPa) Ef (MPa) f cm , (MPa) p (%)
S2-4-2nm 303 276 272 27 452 57 1,000 435 64.4 47.2 0.60
S2-5nm 302 276 246 26 452 57 1,000 435 64.4 41.3 0.67
S2-6nm 303 273 243 29 402 57 1,000 435 64.4 41.3 0.61

employs original notations of the specimens. Letter “S” refers to the throughout all loading stages has demonstrated high-quality bond
type of elements (in Lithuanian “Sija” = “Beam”); the first number characteristics of the reinforcement used in the tests.
corresponds to the level of reinforcement ratio (“2” represents The tests were performed with small increments (2 kN) and
p ≈ 0.6%) and “nm” refers to nonmetallic reinforcement. paused for short periods (about 2 min) to take readings of gauges
The test specimens were loaded with a 100 kN hydraulic jack in and to fix the crack development. In total, it took from 40 to 80 load
a stiff testing frame. The testing equipment acting on the beam increments during total test duration. The experimental moment-
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

weighed 232 kg. The latter summed up with the beam’s own weight curvature diagrams are presented in Fig. 5, whereas the final crack
has induced a 3.5 kNm bending moment at the midspan. Fig. 6 pattern of the beams is shown in Fig. 7.
shows the experimental setup and the loading scheme. Deflections Bond characteristics of the GFRP bars were determined from
at five characteristic points of the beam and slip of the reinforce- bending tests as described in RILEM (1994) and shown in Fig. 8.
ment bars at the both ends of the beam have been also recorded. The required embedment length of 10D (where D is the bar diam-
Concrete surface strains were measured throughout the length of eter) was secured using plastic sleeves. To allow free movement of
the pure bending zone on a 200 mm gauge length, using mechani- the bar relative to the concrete, the end of each internal plastic
cal gauges. As shown in Fig. 6 (view “A”), four continuous gauge sleeve was glued to the GFRP bar with an elastic silicon sealant.
lines were located at different depths with two extreme lines placed The bond specimens were cast together with the full-scale beams.
along the top and bottom reinforcement. Measured strains were The bond tests were performed using a Tinius Olsen H75KS testing
averaged along each gauge line. As reported in Gribniak et al. machine. Displacement control allowed capturing the descending
(2013), LVDT devices were used to measure slip of the GFRP bars part of load-slip diagrams. Two LVDT transducers (Fig. 8) mea-
at both ends of the beams. Zero slippage recorded for all the beams sured the slip of the bar. Two different covers (respective to those
used in the full-scale beams) were considered. Nine bond speci-
mens (three of each diameter) were produced with a standard
50 mm cover as designated by RILEM. Three other specimens re-
inforced with 12 mm bars had a reduced cover of 20 mm.
Fig. 9(a) shows the cover effect on bond stresses for the spec-
imens reinforced with 12 mm bars. The test of the specimens with
standard cover produced bond stress-slip relationships with ascend-
ing curve and extended softening branch reaching 15 mm slip.
However, the specimens with reduced cover demonstrated a brittle
response: failure occurred at 0.15 mm slip, giving about 100 times
the smaller limit slip value than the specimens with standard covers.
A detailed view of the ascending branches of bond stresses is given
in Fig. 9(b). The specimens with reduced cover reached about 25%
smaller maximal bond stresses than the standard members. Insuf-
ficient cover resulted in opening of the splitting cracks that led to
sudden failure. These results are in agreement with experimental
and theoretical investigations of the cover effect on bond stresses
carried out by Tepfers (1979), Uijl and Bigaj (1996), Ogura et al.
(2008), Choi et al. (2012), and Caldentey et al. (2013). Further
analysis will use the bond stress-slip relationship obtained from
Fig. 5. Moment-curvature response of the experimental beams
the standard specimens.

P Load cell
Aj

5x200 mm
1000 mm 1000 mm 1000 mm
View A Upper reinforcement 1st line
2nd line
60
3rd line
60 4th line
Lower reinforcement

Fig. 6. Experimental setup and the measuring scheme of the beams

© ASCE 04014005-5 J. Compos. Constr.

J. Compos. Constr.
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Final crack pattern of the experimental beams

As Fig. 9(c) shows, the bond stress-slip relationship from the As noted earlier, Shima’s model is dependent not only on slip,
Model Code in general captures well the experimental descending but also on reinforcement strain [Eq. (7)]. The model is shown in
branch. As mentioned, slip in serviceability states rarely exceeds Fig. 9 assuming strain ε ¼ 0.005, which is close to the service con-
1 mm [the respective part of the bond stress-slip relationship is ditions of the considered GFRP bars. It gives the lower boundary
shown Fig. 9(d)]. The experimentally obtained bond stress-slip for the bond models considered, however it should be kept in mind
relationships show stiffer responses than the ones predicted by that it is rather sensitive to the reinforcement strain.
the Model Code. The obtained results are in agreement with the
findings reported by Cosenza et al. (1997) and Aiello et al. (2007)
stating that bond tests of ribbed GFRP bars produce more rigid Deformation and Cracking Analysis
behavior in the ascending branch than steel reinforcement. The
increased stiffness of the bond might be explained by specifics of Adequate assessment of tension-stiffening is the most challenging
the slip measurement equipment arranged in the unloaded part of issue in modeling short-term deformations of flexural concrete
the reinforcement. Such measurement does not record slip until members, particularly those with a small reinforcement ratio.
full damage of adhesion along the reinforcement and concrete inter- Dealing with composite reinforcement, there is another aspect that
face has taken place. Different aspects of bond testing techniques should be noted, namely, taking into account slippage of the
have been considered by Pecce et al. (2001), who concluded that reinforcement. A more refined analysis of members having rein-
slip recording at the free end of reinforcement is more represen- forcement with poor bond characteristics cannot be based on the
tative of the real behavior than tests with slip measurement at assumption of linear strain distribution across the section, as it is
the loaded end. not valid at more advanced stages of loading. However, this is not
The obtained experimental results were used to calibrate the applied to the GFRP reinforcement employed in the present study.
Model Code relationship. Best fit resulted in the following param- As was shown in the earlier section, good bond characteristics of
eters of Eq. (6): τ max ¼ 18 MPa; s1 ¼ 0.4 mm; and α ¼ 0.16. the bars were demonstrated directly, by the flexural bond tests,
As shown in Fig. 9(d) constant bond stresses were assumed in
the slip region between 0.4 and 1 mm. This relationship is further
called the calibrated Model Code bond model.
Bond stress, MPa Bond stress, MPa

P/2 P/2 A
A-A
bonded zone bonded zone
50 (or 25)

Slip, mm Slip, mm
180

LVDT 10D 10D LVDT


z

(a) (b)

Bond stress, MPa Bond stress, MPa


P/2 plastic sleeves plastic sleeves
a A P/2
375 50 375

Slip, mm Slip, mm

(c) (d)
reduced cover
Fig. 9. Average bond stress-slip relationships: (a) experimentally ob-
tained bond stress-slip relationship; (b) zoomed part up to 1-mm slip;
(c) bond models proposed in the literature; (d) zoomed part up to
Fig. 8. Bending bond tests 1-mm slip

© ASCE 04014005-6 J. Compos. Constr.

J. Compos. Constr.
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Predicted moment-curvature response using different bond stress-slip models

and indirectly, by the structural tests, when no slippage of the not secure sufficient confinement causing initiation of splitting
reinforcement bars was recorded at the ends of the beams. cracks and, consequently, sudden bond degradation. A similar ef-
The predicted deformation behavior of the tested beams, repre- fect was observed in the bending bond specimens with reduced
sented by the moment-curvature response, is shown in Fig. 10. The cover where splitting cracks have formed at a very small slip fol-
effect of tension-stiffening can be assessed through comparison of lowed by sudden failure of the specimen. On the contrary, a large
test (or analysis) curvature results with the cracked section analysis number of bars closely distributed in the tensile zone (S2-5nm) or
(CSA) shown by dotted lines. Two types of behavior can be ob- enlarged cover (S2-6nm) safeguarded high bond-stress capacity
served from the obtained test diagrams: beam S2-4-2nm has shown (and consequently tension-stiffening) throughout all loading stages.
progressive degradation of tension-stiffening with increasing load, Moreover, as Fig. 10 shows, the distribution of tensile reinforce-
whereas in beams S2-5nm and S2-6nm the tension-stiffening effect ment into a number of layers with bars rather closely spaced from
has not been practically degrading until failure. The cracked section each other also plays an important role in the stiffness of the RC
approach could be successfully used for analysis of GFRP rein- member, as it may significantly delay the cracking process.
forced beams with high reinforcement ratio (Pecce et al. 2000), Regarding the numerical simulation, negligible differences were
however, in moderately and lightly reinforced members, the effect obtained between the moment-curvature diagrams calculated using
of tension-stiffening cannot be neglected as it can be clearly seen the original Model Code 2010 and the calibrated bond models
from the moment-curvature response of the members. (Fig. 10). As bond stresses in these models grow with increasing
The differences in tension-stiffening behavior of the beams can slip, the calculated transfer length reduces with progressing load-
be explained by variations in the effective depth and arrangement of ing. As shown in Fig. 11, variation in the calculated transfer length
tensile bars, as these were the only differences between the beams. with growing load is highly different for the bond models under
The cover varied from 22 mm in beams S2-4-2nm and S2-5nm to consideration. Contrary to the Model Code and the calibrated
51 mm in beam S2-6nm. The latter two beams had almost equal model, the transfer length calculated using Shima’s bond law con-
effective depths (246 and 243 mm, respectively). For beam S2- tinuously increases with progressing load as the bond stresses
5nm, the effective depth was assumed as for the middle layer of degrade with growing reinforcement strain. The final cracking
tensile bars being the centroid of the reinforcement area. Obviously, scheme was obtained in the early post-cracking corresponding to
the cover of beam S2-4-2nm (conventional for such elements) did highest bond stresses and minimal transfer length (Fig. 11).

Bending moment, kNm

Bending moment, kNm

Bending moment, kNm

ltr, cm ltr, cm ltr, cm

Fig. 11. Variation of transfer lengths with increasing bending moment

© ASCE 04014005-7 J. Compos. Constr.

J. Compos. Constr.
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Reinforcement strain distribution within the transfer lengths at service loading

Despite variations in the transfer lengths, the differences in aver- bond properties: increased bond stress reduces the transfer length
age curvature were not significant. The mean curvature response and slip, but increases the number of cracks resulting in almost con-
of the beams was obtained by averaging the sectional curvatures stant flexural stiffness.
along the block calculated by Eq. (4). The distribution of reinforce- Neither of the bond models was able to predict the trend of
ment strains at the service load between two consecutive cracks curvature growth after cracking of beam S2-4-2nm, which had
for each of the beams is shown in Fig. 12 [the service load was conventional cover and arrangement of the bars. As noted, the rapid
assumed to be at 60% of the design bending moment Md that degradation of tension-stiffening occurred mainly due to splitting
corresponded to the design strength of reinforcement taken as cracks reaching the tensile surface of the beam. Although Shima’s
435 MPa (Gribniak et al. 2013)]. It can be seen from Fig. 12 that model takes into account the influence of secondary cracking, the
the mean reinforcement strains were rather close to what also re- predicted degradation of tension-stiffening was not sufficiently
sulted in similar predictions of average curvatures (Fig. 10). Analo- quantified. As can be observed from Table 2, the curvature predic-
gous results were reported by Oehlers et al. (2012) concluding that tions at service load (in Fig. 10, referred as Mser ) were underesti-
mean deformations are not significantly influenced by the assumed mated by about 20% by all the models. On the contrary, the
respective curvatures of the beams S2-5nm and S2-6nm were sig-
nificantly overestimated—the prediction errors by the considered
methods ranged from 55 to 125%. Fig. 10 shows that all the models
Table 2. Experimentally and Numerically Obtained Curvatures
captured rather well the deformation behavior of the latter beams at
Average curvature κ at service loading, more advanced loading stages represented by stabilized cracking.
km−1 ½ðκcalc − κobs Þ=κobs  However, the specific arrangement of reinforcement in the beam
Calculated, κcalc S2-5nm caused delayed cracking and considerably increased ten-
Observed, Model code Model code Shima sion-stiffening at the service load (Fig. 10). Similar results were
Beam κobs (fib 2010) (calibrated) et al. (1987) obtained in the tests of tensile members reinforced with a large
number of closely spaced bars (Rostásy et al. 1976; Hwang and
S2-4-2nm 15.8 12.2 (–22.8%) 12.1 (–23.4%) 13.2 (–16.5%)
Rizkalla 1983; Williams 1986; Purainer 2005). Such members have
S2-5nm 7.1 14.4 (102.8%) 14.4 (102.8%) 16.0 (125.3%)
S2-6nm 8.9 13.9 (56.2%) 13.8 (55.1%) 15.7 (76.4%)
demonstrated significantly increased stiffness, often showing no
signs of degradation in tension-stiffening with increasing load.

Pure bending zone, L = 100 cm

Main cracks
Secondary
cracks Longitudinal
cracks

Bar

Effective concrete Tensile stress


area in tension distribution

Fig. 13. Final crack pattern of the beam S2-6nm

© ASCE 04014005-8 J. Compos. Constr.

J. Compos. Constr.
Table 3. Experimentally and Numerically Obtained Crack Widths Model Code relationship based on the flexural bond tests reported
Maximal crack width at service loading, mm herein.
½ðwcalc –wobs Þ=wobs  These main conclusions were drawn from the study.
1. Deformation behavior and tension-stiffening as well as
Calculated, wcalc
crack pattern and crack width effect are highly related to
Observed, Model code Model code Shima the arrangement of tensile bars within the beam section.
Beam wobs (fib 2010) (calibrated) et al. (1987) An enlarged cover safeguards the high bond-stress capacity
S2-4-2nm 0.52 0.45 (–13.5%) 0.23 (–55.8%) 0.48 (–7.7%) (and consequently tension-stiffening) throughout all loading
S2-5nm 0.26 0.37 (42.3%) 0.15 (–42.3%) 0.31 (19.2%) stages. The study also revealed specific features in the crack-
S2-6nm 0.44 0.54 (22.7%) 0.28 (–36.4%) 0.64 (45.5%) ing and deformation behavior of the beams having a number of
layers of tensile reinforcement with closely spaced bars. This
kind of bar arrangement causes delayed cracking and consid-
Although transfer length and distance between cracks have a erably increased the tension-stiffening effect;
2. A choice of bond model is of minor significance in deforma-
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

minor influence on deformation analysis, it becomes the governing


tion analysis but it becomes vitally important in the calculation
criterion in calculation of crack widths. According to Eq. (5),
of transfer length and crack widths. Maximal crack widths
the crack opening is obtained by integrating the distribution of
of the conventionally reinforced beam with regular crack pat-
reinforcement strain along the crack spacing expressed in terms
tern were rather accurately predicted by the Model Code and
of transfer length. However, assessment of crack distances for some
Shima’s bond models. However, these bond laws significantly
of the test beams was not so straightforward. Conventionally rein-
overestimated crack widths in the beams with enlarged cover.
forced beam S2-4-2nm had a relatively regular final crack pattern
This could have been caused by the appearance of a large num-
with 10 major cracks in the pure bending zone (average crack spac- ber of spectacular secondary cracks forming around the pri-
ing around 110 mm). Whereas formation of cracks in beam S2-6nm mary cracks, thus reducing the widths of the primary cracks;
with enlarged cover (or rather reduced effective depth) was far more 3. Care should be taken when experimental bond-slip relation-
complicated: a large number of minor cracks (seemingly initiated ships are used for cracking simulation. The present analysis
by the secondary cracks) formed around the primary cracks (dis- has shown that deduction of bond properties from the conven-
tanced at about 25 cm from each other), thus redistributing the tional test methods may not assure adequate constitutive
strains and reducing widths of the primary cracks. The large bottom modeling, as the bond-slip relationship is influenced by struc-
cover safeguarded the bottom surface from the formation of split- tural parameters such as concrete cover and distribution of
ting cracks, however the 20 mm cover from the rear sides appeared reinforcement in the section. Moreover, the standard recom-
not to be sufficient to avoid splitting cracks. Fig. 13 shows a fas- mended embedment length of reinforcement (10D) may be
cinating crack pattern, making it difficult to deduce the crack too rough to represent local interaction of reinforcement and
spacing. concrete; and
The calculated maximal crack widths at service load are 4. Further studies should be dedicated to explore tension-
presented in Table 3 and checked against the experimental ones. stiffening and bond-related effects in the members with in-
The test results of the conventionally reinforced beam S2-4-2nm creased cover (or reduced effective depth) and a large number
were quite accurately (with slight underestimation) predicted using of closely spaced bars in the tensile zone.
the Model Code and Shima’s bond models. However, the calculated
maximal crack widths for the unconventionally reinforced beams
(S2-5nm and S2-6nm) were substantially overestimated. These dis- Acknowledgments
crepancies can be explained by the complex nature of the crack
formation in these members not taken into account by the models The authors wish to express their sincere gratitude for the financial
under discussion. The calibrated bond model underestimated by support of the Research Council of Lithuania (Research Project No
almost twice the crack widths for all the beams. These errors were MIP-083/2012). Viktor Gribniak also wishes to gratefully acknowl-
made due to inaccurate assessment of transfer length, in turn being edge the financial support provided by the European Social Fund
dependent on the assumed bond model. Deduction of bond proper- (Project No. 2013/0019/1DP/1.1.1.2.0/13/APIA/VIAA/062).
ties from the conventional test methods may not assure adequate
constitutive modeling: the standard recommended embedment
length of reinforcement (10D) can be too rough to represent local References
interaction of reinforcement and concrete.
Aiello, M. A., Leone, M., and Pecce, M. (2007). “Bond performances of
FRP rebars-reinforced concrete.” J. Mater. Civ. Eng., 10.1061/(ASCE)
0899-1561(2007)19:3(205), 205–213.
Concluding Remarks Balazs, G. L. (1993). “Cracking analysis based on slip and bond stresses.”
ACI Mater. J., 90(4), 340–348.
This paper reports experimental and analytical results on moment- Balazs, G. L., et al. (2013). “Design for SLS according to fib Model Code
deflection and cracking behavior of concrete beams with different 2010.” Struct. Concr., 14(2), 99–123.
arrangement of GFRP bars in the tensile zone. With particular em- Bischoff, P. H. (2005). “Reevaluation of deflection prediction for concrete
phasis on tension-stiffening and bond behavior, the experimental beams reinforced with steel and fiber reinforced polymer bars.” J.
program consisted of full-scale structural beams and bending bond Struct. Eng., 10.1061/(ASCE)0733-9445(2005)131:5(752), 752–767.
Bischoff, P. H. (2007). “Rational model for calculating deflection of rein-
specimens to determine the bond stress-slip relation of the bars.
forced concrete beams and slabs.” Can. J. Civ. Eng., 34(8), 992–1002.
Theoretical investigations were based on the stress transfer ap- Borosnyoi, A., and Balazs, G. L. (2005). “Models for flexural cracking in
proach using three bond laws: (1) Model Code 2010, as the most concrete: The state of the art.” Struct. Concr., 6(2), 53–62.
widely used bond stress-slip relationship; (2) Shima’s model, being Caldentey, A. P., Peiretti, H. C., Iribarren, J. P., and Soto, A. G. (2013).
dependent on slip and reinforcement strain; and (3) the calibrated “Cracking of RC members revisited: Influence of cover, ϕ=ρs; ef and

© ASCE 04014005-9 J. Compos. Constr.

J. Compos. Constr.
stirrup spacing—an experimental and theoretical study.” Struct. Concr., Mias, C., Torres, L., Turon, A., and Barris, C. (2013). “Experimental study
14(1), 69–78. of immediate and time-dependent deflections of GFRP reinforced
Chiaia, B., Fantilli, A. P., and Vallini, P. (2009). “Evaluation of crack width concrete beams.” Compos. Struct., 96, 279–285.
in FRC structures and application to tunnel linings.” Mater. Struct., Oehlers, D. J., Visintin, P., and Haskett, M. (2012). “Consequences and
42(3), 339–351. solutions to our abysmal neglect of bond-slip behaviour in reinforced
Choi, C. K., and Cheung, S. H. (1996). “Tension stiffening model for planar concrete.” Proc. of the 4th Bond in Concrete Conf.: Bond, Anchorage,
reinforced concrete members.” Comput. Struct., 59(1), 179–190. Detailing, Publisher Creations, Brescha, Italy, 39–46.
Choi, D.-U., Chun, S.-C., and Ha, S.-S. (2012). “Bond strength of glass Ogura, N., Bolander, J. E., and Ichinose, T. (2008). “Analysis of bond split-
fibre-reinforced polymer bars in unconfined concrete.” Eng. Struct., ting failure of deformed bars within structural concrete.” Eng. Struct.,
34, 303–313. 30(2), 428–435.
Cosenza, E., Manfredi, G., and Realfonzo, R. (1997). “Behavior and Pecce, M., Manfredi, G., and Cosenza, E. (2000). “Experimental response
modeling of bond of FRP rebars to concrete.” J. Compos. Constr., and code models of GFRP RC beams in bending.” J. Compos. Constr.,
10.1061/(ASCE)1090-0268(1997)1:2(40), 40–51. 10.1061/(ASCE)1090-0268(2000)4:4(182), 182–190.
Fantilli, A. P., Ferreti, D., Ivori, I., and Vallini, P. (1998). “Flexural deform- Pecce, M., Manfredi, G., Realfonzo, R., and Cosenza, E. (2001).
ability of reinforced concrete beams.” J. Struct. Eng., 10.1061/(ASCE) “Experimental and analytical evaluation of bond properties of GFRP
Downloaded from ascelibrary.org by University Of Sheffield on 05/20/14. Copyright ASCE. For personal use only; all rights reserved.

0733-9445(1998)124:9(1041), 1041–1049. bars.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899-1561(2001)13:4


fib (International Federation for Structural Concrete). (2000). “Bond of (282), 282–290.
reinforcement in concrete: State of the art report.” Bulletin No 10, Purainer, R. (2005). “Last- und Verformungsverhalten von Stahlbetonflä-
Lausanne, 434. chentragwerken unter zweiaxialer Zugbeanspruchung.” Ph.D. thesis,
fib (International Federation for Structural Concrete). (2010). “Model Code Universität der Bundeswehr München, Munich, 207 (in German).
2010, first complete draft—Volume 1.” Bulletin No 55, Lausanne, 317. RILEM. (1994). “Bond test for reinforcement steel. 1. Beam test.”
Floegl, H., and Mang, H. A. (1982). “Tension stiffening concept based on Recommendations for the testing and use of constructions materials,
bond slip.” J. Struct. Div., 108(12), 2681–2701.
E & FN SPON, London, 213–217.
Gribniak, V., Kaklauskas, G., Torres, L., Daniunas, A., Timinskas, E.,
Rostásy, F. S., Koch, R., and Leonhardt, F. (1976). “Zur Mindestbewehrung
and Gudonis, E. (2013). “Comparative analysis of deformations and
für Zwang von Außenwänden aus Stahlleichtbeton.” Deutscher
tension-stiffening in concrete beams reinforced with GFRP or steel bars
Ausschuß für Stahlbeton 267, Beuth, Berlin, 116 (in German).
and fibers.” Compos. B Eng., 50, 158–170.
Schöck, (2006). Technical information Schöck Combar, Schöck Bauteile,
Hong, S., and Park, S. K. (2012). “Uniaxial bond stress-slip relationship of
Baden Baden, Germany, 23.
reinforcing bars in concrete.” Adv. Mater. Sci. Eng., 2012, 12.
Hwang, L. S., and Rizkalla, S. H. (1983). “Behavior of reinforced concrete Shima, H., Chou, L., and Okamura, H. (1987). “Micro and macro models
in tension at post-cracking range.” Engineering Rep., Dept. of Civil for bond in reinforced concrete.” J. Facul. Eng. B, 39(2), 133–194.
Engineering, Univ. of Manitoba, Winnipeg, 124. Somayaji, S., and Shah, S. P. (1981). “Bond stress versus slip relationship
Kaklauskas, G., Christiansen, M. B., Bacinskas, D., and Gribniak, V. and cracking response of tension members.” ACI J. Proc., 78(3),
(2008). “Deformation model of reinforced concrete members taking 217–225.
into consideration shrinkage and creep effects at the pre-loading stage.” Tepfers, R. (1979). “Cracking of concrete cover along anchored deformed
Final Rep. No T-1025/08, Vilnius Gediminas Technical Univ., Vilnius, reinforcing bars.” Mag. Concr. Res., 31(106), 3–12.
47 (in Lithuanian). Torres, L., Neocleous, K., and Pilakoutas, K. (2012). “Design procedure
Kaklauskas, G., Gribniak, V., Jakubovskis, R., Gudonis, E., Salys, D., and and simplified equations for the flexural capacity of concrete members
Kupliauskas, R. (2012). “Serviceability analysis of flexural reinforced reinforced with fibre-reinforced polymer bars.” Struct. Concr., 13(2),
concrete members.” J. Civ. Eng. Manage., 18(1), 24–29. 119–129.
Lackner, R., and Mang, H. A. (2003). “Scale transition in steel-concrete Uijl, J. A., and Bigaj, A. (1996). “A bond model for ribbed bars based on
interaction. Part I: Model.” J. Eng. Mech., 10.1061/(ASCE)0733- concrete confinement.” Heron, 41(3), 201–226.
9399(2003)129:4(393), 393–402. Williams, A. (1986). “Test on large reinforced concrete elements subjected
Mahmoud, K., and El-Salakawy, E. (2013). “Shear strength of GFRP- to direct tension.” Technical Rep. 562, Cement and Concrete Associ-
reinforced concrete continuous beams with minimum transverse ation, Wexham Springs, Cement and Concrete Association, Slough,
reinforcement.” J. Compos. Constr., 10.1061/(ASCE)CC.1943-5614 Buckinghamshire, 56.
.0000406. Wu, H. Q., and Gilbert, R. I. (2009). “Modeling short-term tension stiff-
Marti, P., Alvarez, M., Kaufmann, W., and Sigrist, V. (1998). “Tension ening in reinforced concrete prisms using a continuum-based finite
chord model for structural concrete.” Struct. Eng. Int., 8(4), 287–298. element model.” Eng. Struct., 31(10), 2380–2391.

© ASCE 04014005-10 J. Compos. Constr.

View publication stats J. Compos. Constr.

You might also like