Professional Documents
Culture Documents
Flexural Behavior and Serviceability of Normal-And High-Strength Concrete Beams Reinforced With Glass Fiber-Reinfor...
Flexural Behavior and Serviceability of Normal-And High-Strength Concrete Beams Reinforced With Glass Fiber-Reinfor...
Flexural Behavior and Serviceability of Normal-And High-Strength Concrete Beams Reinforced With Glass Fiber-Reinfor...
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/256287944
CITATIONS READS
24 793
3 authors:
Brahim Benmokrane
Université de Sherbrooke
380 PUBLICATIONS 4,418 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Tension Stiffening and Cracking of Concrete Ties Reinforced with Deformed and Sand Coated GFRP
bars View project
All content following this page was uploaded by Amr Maher Elnemr on 04 March 2014.
This paper investigated the flexural behavior and serviceability distinction between both modes of failure is achieved through
performance of glass fiber-reinforced polymer (GFRP)-reinforced the balanced reinforcement ratio rfb, which is influ-
concrete (GFRP-RC) beams fabricated with normal- and high- enced by the mechanical properties of the FRP bars and
strength concretes (NSCs and HSCs). The beam specimens concrete strength.
measured 4250 mm long x 200 mm wide x 400 mm deep (167.0 x Due to the lower modulus of elasticity of the FRP bars
8 x 16 in.). Three GFRP products with moduli of elasticity ranging
from 48.7 to 69.0 GPa (7100 to 10,000 ksi) with sand-coated
compared to that of steel, for the same reinforcement ratio
and helically grooved surface textures were employed. A total of rf, FRP-reinforced concrete (RC) members will exhibit
12 full-scale beams reinforced with GFRP bars and two reinforced larger deflections and crack widths. Thus, the design of
with steel bars, serving as control specimens, were tested to failure FRP-RC members is usually governed by serviceability.
in four-point bending over a clear span of 3750 mm (148 in.). The For deflection, a number of studies6-8 proposed modifying
test parameters were: 1) type and ratio of the GFRP reinforcement; Branson’s original equation9 for the effective moment of
2) surface configuration of the GFRP bars; 3) concrete strength; inertia and introduced modification factors for FRP-RC
and 4) bar diameter. The test results were reported in terms of members. Other studies10-12 proposed an equivalent moment
deflection, crack width, strains in concrete and reinforcement, of inertia derived from curvatures. For cracking, modifica-
flexural capacity, and mode of failure; they were also employed tions were proposed to the original Gergely and Lutz13 equa-
to assess the accuracy of the current deflection and crack-width
tion or strain limits were introduced to control the crack
prediction equations in the FRP-RC codes. The test results revealed
that the crack widths were affected by the bar diameter and surface
width.14 Furthermore, Feeser and Brown15 concluded
configuration while the deflections were not significantly affected. that sections reinforced with multiple layers of glass FRP
In addition, the bond coefficient kb value of 1.4 was very conserva- (GFRP) reinforcement might be more attractive and should
tive for both of the sand-coated and helically grooved GFRP bars be explored. They also proposed that multiple layers of rein-
in NSC and HSC. forcement would enable a GFRP-RC section to accommo-
date larger rf for compression-controlled failure. Besides,
Keywords: beam; crack width; deflection; fiber-reinforced polymer; flexure; using higher concrete strengths to make more efficient use
high-strength concrete; normal-strength concrete; reinforcing bars; serviceability.
of the GFRP tensile strength is more applicable to designs
with multiple layers of reinforcement.
INTRODUCTION This paper presents the flexural behavior and serviceability
Fiber-reinforced polymer (FRP) has been commer- of 12 full-scale beams reinforced with three commercially
cially available as reinforcement for concrete over the available GFRP bars and two full-scale beams reinforced
last 15 years.1 Due to the lack of well-established standards, and steel bars fabricated using normal- and high-strength
a wide variety of FRP bars are currently marketed, ranging concretes (NSCs and HSCs). Moreover, the test results were
from the simple smooth and helically deformed bars to bars employed to assess the accuracy of the current design equa-
treated with external features such as sand-coating. Nowadays, tions in North American FRP codes and guidelines (CAN/
over 10 million meters of FRP reinforcement are used in CSA S806,3 ACI 440.1R,4 ISIS Manual No. 3,14 and the
construction every year.1 These FRP products have a wide ACI 440-H Subcommittee on FRP-Reinforced Concrete).
range of mechanical properties and different surface config-
urations. Most of the design codes and guides use equations RESEARCH SIGNIFICANCE
for the steel-reinforced members with some modifications Due to lack of standards, a wide variety of FRP products
to account for the differences in the mechanical proper- are commercially available. Thus, the need to investigate
ties between FRP and steel bars. Furthermore, service- serviceability-related issues and validate the accuracy of
ability checks such as crack widths and deflections involve design codes and guidelines is paramount. This paper investi-
assessing the tension stiffening effect, which directly arises gates the flexural behavior and serviceability performance of
from bond behavior.2 Thus, employing an equation to predict NSC and HSC beams reinforced with different types, diam-
the performance may yield reasonable predictions with one eters, ratios, and configurations of GFRP bars. Moreover, the
type of FRP bars but discrepancies with another.
At the ultimate limit state, most of the design
codes and guides recommend the over-reinforced section ACI Structural Journal, V. 110, No. 6, November-December 2013.
design,3,4 where concrete crushing is the dominant mode MS No. S-2011-321.R4 received January 31, 2013, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
of failure. This is preferred because it is less catastrophic reserved, including the making of copies unless permission is obtained from the
and yields higher deformability before failure. Some codes, copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the September-October 2014 ACI Structural Journal if the discussion is
however, permit the under-reinforced section design.5 The received by May 1, 2014.
was installed to continuously capture the crack width varia- strength which, in turn, is a function of compressive strength,
tion. Furthermore, the crack formation was visually observed increasing the concrete compressive strength is expected to
during the test and was marked on one side of each beam and yield higher cracking moments. The cracking moment of
the corresponding loads were recorded. Figure 2(b) shows a the HSC beams ranged from 17.49 to 23.78 kN.m (12.9 to
beam specimen during testing. 17.5 kip.ft) with an average of 21.61 kN.m (15.9 kip.ft). The
ratio of the average cracking moment of the HSC beams to
TEST RESULTS AND DISCUSSION that of the NSC beams was 1.63. This ratio is close to the
First cracking moment ratio between the square root of the average compressive
During the test, the beams were observed visually until strength of the HSC beams (66.2 MPa [9.6 ksi]) and that of
the first crack appeared and the corresponding load was the NSC beams (32.5 MPa [4.7 ksi]), which was 1.43. The
recorded. The cracking load was also verified from the load- cracking moments were predicted using Eq. (1), where fr is
deflection and load-strain relationships. Table 3 provides the calculated from Eq. (2) for ACI 440.1R4, and from Eq. (3)
cracking moments of the tested beams (excluding the self- for CSA codes3,5 and ISIS.14
weight of the beams). The cracking moment of the NSC
beams ranged from 12.20 to 14.15 kN.m (9.0 to 10.4 kip.ft) Mcr = fr /ytIg (1)
with an average of 13.27 kN.m (9.8 kip.ft). Because the
cracking moment is directly related to concrete tensile where
Fig. 3—Moment-to-maximum concrete and reinforcement strain relationships. (Note: 1 kN.m = 23,730 lb.ft.)
GFRP bars after cracking at the same load level. In addition, exceeding 0.005/d generally failed to satisfy either the deflec-
GFRP-RC beams with close Ef Af are expected to yield the same tion or the crack width criteria for beams with span-depth
moment-strain relationship. The GFRP-RC beams in Series III ratios of 8 to 13.20 The service load level for the GFRP-RC
had the same Ef Af as the steel-reinforced ones. After cracking, members, however, has no fixed definition in the design codes
however, they showed lower strains in the GFRP bars than in and guides. ISIS14 recommended a value of 2000 me as a limit
the steel bars. This may be related to the good bond behavior of for the strain in the GFRP bars under service load. Other
the GFRP bars, which yielded more cracks, leading to smaller researchers assume that the service load is approximately
stains at the same load level. 30% of their nominal flexural capacity,21 0.3Mn. The former
Figure 3 also shows that Series I beams with low rf (0.38 and often satisfies the serviceability requirements but, in some
0.56%) experienced very high strain increases at cracking. cases, the 0.30Mn yields very high strain levels, such as in case
The sharp increase at the cracking was more than 3000 me in of N2#13G2 and N3#13G1 (5349 and 4378 me, respectively).
the NSC and HSC beams. While three of these beams were Kassem et al.17 also reported strain values as high as 4119 me
designed as over-reinforced, that did not prevent the large in the GFRP bars and 3850 me in the CFRP bars at 0.3Mn,
increase due to poor energy absorption at cracking. Main- which is also high and could not be considered as strains at
taining minimum practical reinforcement may be of interest service. Furthermore, at a curvature limit of 0.005/d, strains
to ensure that the section behaves reasonably after cracking. in FRP bars were typically as high as 4500 me.20 Thus, in this
Table 4 provides the curvature at failure as a function of 1/d. study, the key results were presented at 2000 me in the GFRP
The curvature ranged from 0.008/d to 0.015/d for compres- bars at 0.3Mn and 0.67Mn.
sion-controlled failure (excluding N2#13G2), which agrees
with the range19 of 0.0089/d to 0.012/d and the range17 of Deflection behavior
0.009/d to 0.014/d. On the other hand, the tension-controlled Figure 4 provides typical bilinear load-deflection relation-
failure of the H3#13G1 beam yielded a very high curva- ships. Each curve represents the average deflection obtained
ture of 0.020/d. Based on the moment-curvature diagrams from two LVDTs mounted at beam midspan. The relation-
of 50 GFRP-RC beams,20 the maximum unified curvature at ships (Fig. 4) revealed that rf had a direct impact on the stiff-
a service load that satisfies both deflection and crack-width ness of the beams and, hence, on the load-deflection behavior.
serviceability should be limited to 0.005/d. A curvature limit Series I beams (with the lowest rf) showed the highest
deflection compared to Series II and III. Due to the differ- not affected by GFRP bar properties or surface configura-
ence in the modulus of elasticity of the GFRP bars employed tion when the same Ef Af was achieved. This was confirmed
herein, Ef Af was used as reference. Series I beams had Ef Af by comparing the load-deflection relationships of Series II
around 18 MN (4047 kip), while Series II and III beams had and III (GFRP-RC) against Series IV (steel-RC), which had
Ef Af of approximately 60 MN (13,488 kip). Thus, Series I almost the same Ef Af. Providing the same Ef Af with GFRP
beams exhibited the largest deflection compared to Series II and steel bars yielded the same load-deflection relationships
and III at the same load level. Figure 4 indicates that using HSC up to yielding of the steel bars. Table 5 summarizes the deflec-
while maintaining the same Ef Af contributed to enhancing the tion of the tested beams at 0.3Mn and 0.67Mn. The deflection
deflection of the GFRP-RC beams in addition to enhancing values from Table 5 were plotted against Ef Af and rf /rfb, as
the load-carrying capacity of Series II and III beams. Series I shown in Fig. 5. This figure shows that the deflection values
beams with low rf and the same Ef Af, however, did not show at 0.3Mn and 0.67Mn followed the same trend. The higher the
significant differences in load-deflection relationships. Ef Af and rf /rfb, the lower the deflection values. Assuming that
Comparing Beams N2#25G3, N5#15G3, N5#15G2, beam deflection limit at service load is span/240 (L/240), as
and N6#15G1 (which had the same Ef Af) showed that the provided for by CAN/CSA A23.322—which yields 15.63 mm
beams had the same load-deflection relationships. This (0.6 in.)—all the beams except H2#13G2 exhibited deflec-
was confirmed by comparing H5#15G2 and H6#15G1. tions smaller than 15.63 mm (0.6 in.) at 0.30Mn.
Furthermore, comparing the load-deflection relationships of
N2#25G3 and N5#15G3 reveals that increasing the GFRP Crack propagation and crack width
bar diameter from 15 to 25 mm (0.6 to 1.0 in.) while main- All beam specimens were initially uncracked before
taining the same Ef Af did not affect the load-deflection rela- testing, except N6#15G1, which had a hair crack in the pure
tionships. In addition, comparing N5#15G3, N5#15G2, and bending zone (this crack was not considered in the analysis).
N6#15G1 showed that the load-deflection relationships were Flexural cracks appeared when the concrete’s tensile strength
and, consequently, the crack moment was reached in the pure crack widths except in case of H6#15G1, which seems to be
bending zone. Table 6 gives the initial crack width of the affected by the pre-existing hair crack.
first crack in each beam. Figure 6† (refer to Appendix) shows Comparing Series II and III beams, which have almost the
the crack pattern of the tested GFRP-RC beams of Series II same Ef Af but different types of GFRP bars (sand-coated:
and III at two loading levels: 0.3Mn and 0.67Mn. The former G1; G2 and helically grooved: G3), revealed fewer cracks
corresponds to a service load reported by other research21; the in beams reinforced with helically grooved GFRP bars.
latter presents the threshold at which cracks almost stabilized. This tends to confirm that the sand-coated GFRP bars have
Looking at Fig. 6 reveals that increasing rf in the normal- better bond characteristics than the helically grooved ones.
and high-strength GFRP-RC beams increased the number Figure 6 also pointed out the effect of concrete strength on
of cracks and, consequently, reduced the average crack cracking behavior. Given the same load level, the higher
spacing. Moreover, the crack depth also decreased when concrete strength resulted in more cracks with closer
rf increased. Comparing Series I and II beams, which had spacing and smaller crack width than beams with lower
the same reinforcing bars (sand-coated GFRP bars: G1 and concrete strength. Moreover, HSC produced higher cracking
G2), it can be noted that increasing rf or Ef Af resulted in moments, which enhance the comparable behavior. Similar
a higher number of cracks and also yielded smaller initial behavior was also reported23 in FRP-RC slabs where using
HSC resulted in closer cracks with smaller widths.
†
Most design codes specify a flexural crack-width limit for
The Appendix is available at www.concrete.org in PDF format as an addendum to
the published paper. It is also available in hard copy from ACI headquarters for a fee steel-reinforced concrete structures to protect the reinforcing
equal to the cost of reproduction plus handling at the time of the request. bars from corrosion and to maintain the structure’s aesthetic
appearance. Unlike steel reinforcement, FRP is corrosion- bars instead of five No. 15 GFRP bars increased the crack
resistant. Therefore, the serviceability limits for crack widths width, especially at higher load levels. This effect was
in FRP-RC elements may be directly related to aesthetic greater in the NSC (N5#15G3, N2#25G3) beams than in the
considerations. The FRP design codes and guidelines permit HSC (H5#15G3, H2#25G3) ones. This could be explained
a larger crack width for FRP-RC elements compared to their in light of the previous studies, which confirmed the depen-
counterparts reinforced with steel. CAN/CSA S8063 and dency of the bond strength on the FRP bar diameter.24-27 For
CAN/CSA S65 specify a service-limiting flexural crack FRP bars with similar surface configuration, the larger the
width of 0.5 mm (0.02 in.) for exterior exposure (or aggres- FRP bar diameter, the lower the bond strength of the FRP
sive environmental conditions) and 0.7 mm (0.03 in.) for bar. As the smaller-diameter bars are able to exhibit higher
interior exposure. In addition, ACI 440.1R4 recommends bond strength, smaller crack spacing and smaller crack
using CAN/CSA S8063 limits for most cases. On the widths are expected.
other hand, because there is a direct relationship between
the strain in the reinforcing bars and the crack width, DEFLECTION AND CRACKING PROVISIONS
ISIS14 specifies 2000 me as a strain limit in FRP reinforcing Deflection provisions
bars to control crack width. CAN/CSA S806-023—CAN/CSA S8063 employs curva-
Figure 7 gives the moment-crack width relationships for the ture integration to determine the deflection of a concrete
tested beams. Table 6 lists the crack widths at 2000 me in the member assuming the section is fully cracked the cracked
FRP at 0.3Mn and 0.67Mn. Series I beams evidenced very high
regions of the beam. CAN/CSA S806,3 based on the closed
crack widths at service-load levels. Beyond that, the cracks
form solution, provided deflection equations for simple
were very wide with a maximum of 2.38 mm (0.09 in.). HSC
yielded smaller initial crack widths and enhanced the crack loading cases such as Eq. (7) for a simply supported member
widths at both service- and ultimate-load levels (Fig. 7(a) and subjected to two-point loading
(b)). At 2000 me, one beam (N2#13G2) showed a crack width
PL3 a I L
3
larger than 0.7 mm (0.03 in.) (0.77 mm [0.031 in.]), while at a
3
0.3Mn, the three beams exceeded this limit: 1.03, 0.78, and d= 3 - 4 - 8 1 - cr g (7)
0.83 mm (0.04, 0.03, and 0.033 in.) for N2#13G2, N3#13G1, 24 Ec I cr L L Ig L
and H2#13G1, respectively. As mentioned previously, the
strains in those three beams at 0.3Mn were very high and not ACI 440.1R-064—ACI 440.1R4 employs the effective
expected to satisfy the crack width criterion at this load level. moment of inertia Ie in deflection calculations of cracked
Thus, even with the over-reinforced section, practical design FRP-RC beams. The Ie provided by Branson9 was modi-
values for rf /rfb might be needed. The ACI 440.1R4 ratio of fied with a factor bd to account for FRP instead of steel.
rf /rfb = 1.4 might be adequate for some cases. Equation (8) provides the Ie expression of ACI 440.1R,4 where
Increasing the Ef Af from approximately 18 MN (4047 kip) Ma is the applied moment
(Series I) to approximately 60 MN (13,489 kip) (Series II)
with the same sand-coated GFRP bars greatly enhanced the Ie = (Mcr /Ma)3bdIg + [1 – (Mcr /Ma)3]Icr ≤ Ig (8a)
crack widths at all load levels. The average crack width in
Series II beams was 0.42 and 0.36 that of Series I at 2000 me where
and 0.3Mn, respectively. The beams satisfied the severe-
exposure crack width criterion (0.5 mm [0.02 in.]) at both bd = 0.2(rf /rfb) ≤ 1.0 (8b)
load levels. On the other hand, comparing Series II and
III beams reveals that the helically grooved GFRP bars ACI 440-H—ACI 440-H proposed an alternative expres-
yielded larger cracks than the sand-coated ones, even when sion for Ie that works for both FRP- and steel-RC members
the same Ef Af was provided. The average crack width of without the need for empirical correction factors.12 The
Series III beams was 1.80 and 1.72 times that of Series II expression, presented in Eq. (9), included the new factor g,
beams at 2000 me and 0.3Mn, respectively. Once again, this which accounts for stiffness variation along the member.
tends to confirm that the sand-coated GFRP bars have better
bond characteristics than the helically grooved ones. Ie = Icr /[1 – hg(Mcr /Ma)2] ≤ Ig (9a)
The effect of bar diameter on crack width can be seen in
Fig. 7(a) and (b) for NSC and HSC. Using two No. 25 GFRP h = 1 – (Icr /Ig) (9b)
Table 7—Experimental-to-predicted crack width A minimum reinforcement ratio, however, has to be main-
(wexp /wpred) using different kb values tained to generalize this phenomenon.
wexp/wpred 6. Using HSC while keeping the axial reinforcement
ACI 440.1R4; CSA S65 ISIS14
stiffness (Ef Af) constant helped enhance deflection of the
GFRP-RC beams, crack width, and ultimate load-carrying
Beam kb = 1.4 kb = 1.2 kb = 1.0 kb = 0.8 kb = 1.2 kb = 1.0 kb = 0.8
capacity. The results did not support any effect of bar diam-
N2#13G2 0.95 1.11 1.33 1.66 1.71 2.05 2.57 eter or surface configuration of the GFRP bars on the deflec-
N3#13G1 1.03 1.20 1.44 1.80 1.02 1.22 1.53 tion of the NSC and HSC beams.
H2#13G1 0.63 0.74 0.88 1.10 1.20 1.44 1.80 7. Increasing the reinforcement ratio and concrete
H3#13G2 0.66 0.77 0.92 1.16 1.57 1.88 2.36 strength resulted in a larger number of cracks and smaller
N5#15G2 0.55 0.64 0.77 0.96 0.68 0.82 1.02
crack widths. Beams reinforced with sand-coated GFRP
bars produced a larger numbers of cracks and smaller crack
N6#15G1 0.18 0.21 0.25 0.32 0.31 0.37 0.47
widths than those reinforced with helically grooved GFRP
H5#15G2 0.29 0.34 0.41 0.51 0.46 0.55 0.69 bars. This tends to confirm the better flexural bond charac-
H6#15G1 0.39 0.46 0.55 0.68 0.76 0.91 1.14 teristics of the sand-coated bars.
N5#15G3 0.59 0.69 0.83 1.03 1.20 1.44 1.80 8. Bar diameter had an effect on crack width, as evidenced
N2#25G3 0.86 1.00 1.20 1.51 1.05 1.26 1.58 by the NSC and HSC beams reinforced with GFRP bars.
H5#15G3 0.65 0.76 0.91 1.14 0.87 1.04 1.31
Employing the same axial-reinforcement stiffness (Ef Af)
with two different diameters of helically grooved GFRP bars
H2#25G3 0.73 0.85 1.02 1.28 0.63 0.76 0.95
(No. 15 and No. 25) yielded higher crack widths with the
Average 0.63 0.73 0.88 1.10 0.96 1.15 1.43 No. 25 than with the No. 15. The effect was higher in NSC
Standard than in HSC.
0.25 0.30 0.36 0.44 0.42 0.51 0.63
deviation 9. ACI 440.1R4 and ACI 440-H yielded unconserva-
COV, % 41 41 41 41 44 44 44 tive deflection values at the 0.3Mn and 0.67Mn. At 0.3Mn,
the experimental-to-predicted deflection ratio (dexp/dpred) of
ACI 440.1R4 and ACI 440-H were 1.17 ± 0.38 and 1.13 ±
expected to exhibit similar load-reinforcement-strain rela- 0.22, respectively.
tionships for NSC and HSC. 10. CAN/CSA S8063 and ISIS14 yielded conserva-
4. The HSC and HSC beams with low reinforcement tive deflection predictions. At 0.3Mn, the experimental-
ratios, rf (0.56% or less), showed very sharp increases in to-predicted deflection ratios of CAN/CSA S8063 and
reinforcement strains at cracking of over 3000 me. While ISIS14 were 0.76 ± 0.12 and 0.80 ± 0.11, respectively.
the beams were designed as over-reinforced sections, this 11. The bond coefficient (kb) value of 1.4 is very conserva-
did not prevent the large strain increase due to poor energy tive for both sand-coated and helically grooved GFRP bars
absorption at cracking. A minimum practical reinforcement in NSC and HSC. Reasonable crack-width predictions were
amount may have to be maintained. obtained from ACI 440.1R4 and CAN/CSA S65 using a kb
5. Except for Series I beams with low reinforcement ratios factor of 1.2 for the helically grooved GFRP bars and 1.0 for
rf, the curvature at 0.30Mn was 0.004/d. The GFRP beams the sand-coated bars in NSC and HSC. A kb of 0.8 provided
satisfied deflection and crack-width serviceability limits by CAN/CSA S65 yielded very unconservative predictions,
(L/240 for deflection and 0.7 mm [0.03 in.] for crack width). on average, for sand-coated and helically grooved GFRP bars
This partially confirms the Vijay and GangaRao20 findings. in NSC and HSC. This value (kb = 0.8), however, yielded