Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Engineering Structures 173 (2018) 606–619

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Evaluation of the flexural strength and serviceability of concrete beams T


reinforced with different types of GFRP bars

Amr El-Nemra,1, Ehab A. Ahmedb,2, Adel El-Saftyc, Brahim Benmokraned,
a
Department of Civil Engineering at the University of Sherbrooke, Sherbrooke, Quebec J1K 2R1, Canada
b
Department of Civil Engineering, University of Sherbrooke, Sherbrooke, Quebec J1K 2R1, Canada
c
Department of Civil Engineering, School of Engineering, College of Computing, Engineering & Construction, University of North Florida, FL, USA
d
Department of Civil Engineering and NSERC Research Chair in FRP Reinforcement for Concrete Infrastructure and Tier-1 Canada Research Chair in Advanced Composite
Materials for Civil Structures in the Department of Civil Engineering, University of Sherbrooke, Quebec J1K 2R1, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: The current available glass-fiber-reinforced polymer (GFRP) bars have a modulus of elasticity ranges from 40 to
Concrete 60 GPa in accordance with CSA S807-10 Canadian standard. The surface profile of GFRP bars, however, can be
Fiber-reinforced polymer (FRP) smooth, sand-coated, deformed, grooved, or ribbed. This study aimed at investigating the flexural behavior and
FRP bar serviceability performance of concrete beams reinforced with different types of GFRP bars. The test parameters
Beam
were: (i) modulus of elasticity (46.4–69.3 GPa); (ii) surface profile (sand-coated and helically-grooved), and (iii)
Strain
reinforcement ratio. The study included testing of 17 full-scale beams measuring 4,250 mm long × 200 mm
Crack width
Deflection wide × 400 mm deep reinforced with GFRP bars. The test results are presented and discussed in terms of de-
Serviceability flection, crack width, strain, and load-carrying capacity. The cracking behavior of the tested beams tends to
Design codes confirm that sand-coating of GFRP bars enhances the bond performance in concrete more than the helically-
grooved profile. The curvature limit of 0.005/d seems to be feasible in controlling the serviceability of GFRP-
reinforced concrete (GFRP-RC) beams. In addition, ACI 440.1R-06 and ACI 440.1R-15 underestimated the de-
flection, while ISIS M-03 and CSA S806-12 provided conservative deflection values at 0.30 of nominal moment
capacity, Mn.

1. Introduction cracking performance and consequently crack width. Thus, the per-
formance of GFRP bars with different modulus of elasticity and surface
Glass fiber-reinforced-polymer (GFRP) bars have been extensively profiles needs to be studied, considering the serviceability and ultimate
used as alternatives to steel bars for the last two decades. Recent ad- limit states. Consequently, the GFRP-RC structures may cost more than
vances in FRP technology led to the development of GFRP bars with the steel-RC structures since GFRP bars are more expensive than steel
modulus of elasticity exceeds 60 GPa which is expected to reduce the bars and satisfying the serviceability limit state may require more bars.
reinforcement amounts and yields cost-effective designs. In addition, The optimization of the structural design, however, may help in redu-
the GFRP bars have a variety of surface profiles such as smooth, de- cing the total cost of GFRP-RC structures such as in case of La
formed, sand-coated, and grooved to enhance the bond with the sur- Chancelière Parking Garage, Quebec City, Canada [3]. In this project,
rounding concrete. the 50 tons steel bars, priced at $125,000 CAD, were replaced with
Since GFRP bars have a lower modulus than steel bars, the design of 42,160 linear meters of GFRP, priced at $210,800 CAD (1.7 times the
GFRP-reinforced concrete (GFRP-RC) is often governed by the service- cost of steel). The total cost, however, was dropped by 5% when GFRP
ability limit state (deflection and cracking) rather than the ultimate was used since the asphalt layer was replaced with an anti-friction
state. Consequently, increasing the tensile properties—especially the chemical layer due to the non-corrodible nature of GFRP reinforcing
modulus of elasticity—is expected to enhance the serviceability of bars [3].
GFRP-RC members. In addition, surface profile may play a role in the Bischoff and Gross [7,8] reported that the abrupt loss of stiffness at


Corresponding author.
E-mail addresses: Amr.M.El-Nemr@USherbrooke.ca (A. El-Nemr), Ehab.Ahmed@Concordia.ca (E.A. Ahmed), adel.el-safty@unf.edu (A. El-Safty),
Brahim.Benmokrane@USherbrooke.ca (B. Benmokrane).
1
Former PhD Candidate, Department of Civil Engineering, University of Sherbrooke, Presently Assistant Professor in German University in Cairo, Egypt.
2
Former Research Assistant, Department of Civil engineering, University of Sherbrooke, Presently Instructor, Concordia University, Montreal, Quebec, Canada.

https://doi.org/10.1016/j.engstruct.2018.06.089
Received 6 December 2017; Received in revised form 22 May 2018; Accepted 22 June 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Nomenclature h1 distance from neutral axis to center of tensile reinforce-


ment (mm)
A effective tension area of concrete surrounding the flexural h2 distance from neutral axis to extreme tension fiber (mm)
tension reinforcement and bearing the same centroid as Icr transformed moment of inertia of cracked reinforced-
the reinforcement, divided by the number of bars (mm2) concrete section (mm4)
a shear span (mm) Ie effective moment of inertia (mm4)
Af area of FRP tension reinforcement (mm2) Ig gross moment of inertia of uncracked section (mm4)
b effective beam width (mm) kb bond-dependent coefficient
c neutral-axis depth (mm) L length of clear span (mm)
d distance from the extreme compression fiber to the cen- Lg length of the uncracked section (mm)
troid of tension force (mm) Ma applied moment (kN·m)
db bar diameter (mm) Mcr cracking moment (kN·m)
dc distance from extreme tension fiber to the center of the Mn nominal moment of the reinforced-concrete section (kN·m)
longitudinal bar or wire located closest thereto according P applied load (kN)
to the code or guideline (mm) s spacing between the longitudinal reinforcement bars
Ec modulus of elasticity of concrete (MPa) (mm)
Ef modulus of elasticity of longitudinal reinforcement (MPa) SD standard deviation
Es modulus of elasticity of longitudinal steel reinforcement w maximum crack width (mm)
(MPa) yt distance from centroid axis of cross-section to the extreme
f'c compressive strength of the concrete (MPa) fiber in tension (mm)
ff stress in FRP reinforcement under specified loads (MPa) δ mid-span deflection (mm)
ffu ultimate strength of FRP longitudinal reinforcement (MPa) εcu ultimate strain of concrete
fr modulus of rupture (MPa) ρf longitudinal reinforcement ratio
fs stress in bars at serviceability limit state and calculated on ρfb balanced longitudinal reinforcement ratio
the basis of a cracked section ψ curvature
ft tensile strength from cylinder-splitting test (MPa)

cracking affects the post-cracking behavior and deflection. Mousavi results concerning the bond-dependent coefficient are presented else-
et al. [21] investigated the deflection of GFRP-RC beams, claiming that where [12].
the low elastic modulus of GFRP bars accounts for the abrupt loss of
concrete stiffness. In addition, they reported that the bond-dependent
coefficient and the modulus of elasticity of the FRP bars were the main 2.1. Materials
factors affecting the behavior of the GFRP-RC beams. Arivalagn [4]
compared stainless-steel, GFRP, and conventional steel bars in simply The beams were reinforced with three commercially available GFRP
supported concrete beams. The results showed that once the concrete bars referred to as GFRP-1, GFRP-2, and GFRP-3. GFRP bars sizes No.
cracked, the GFRP-RC beams lost their stiffness at a faster rate than 13 to No. 25 (12.9–25.4 mm diameters) of each type were used. GFRP-1
those reinforced with steel. This is due to the low modulus of elasticity and GFRP-2 bars had a sand-coated surface and manufactured by
of the GFRP bars compared to that of steel. Pultrall Inc. (Thetford Mines, QC, Canada), while GFRP-3 had a heli-
Jakubovskis et al. [18] investigated the effect of distributing the cally-grooved surface and manufactured by Fiberline Composites Inc.
tensile bars on three layers in GFRP-RC beams. The results indicated (Kitchener, ON, Canada). Fig. 1 shows the GFRP reinforcing bars. The
that the deformation behavior, tension stiffening, and crack pattern and tensile properties of the GFRP bars were determined by testing re-
width were related to the arrangement of the tensile bars within the presentative specimens in accordance with ASTM D7205 [5] and the
beam section. Based on a statistical analysis of 173 flexural tests of bond performance was determined in accordance with the pullout test
GFRP-RC beams from literature, Xue et al. [24] concluded that the of ASTM D7913 [6]. Table 1 summarizes the properties while Figs. 2
reinforcement ratio of 1.5 ρb (where ρb is the balanced reinforcement and 3 show the pullout test details and the typical bond stress-slip re-
ratio) can be considered as the upper bound for beams in the transition lationships of the GFRP bars. In accordance with Canadian standard
region. On the other hand, Vijay and GangaRao [23] stated that the CSA [11], the three types of GFRP bars were classified as: (1) Grade I
reinforcement ratio should be higher than 1.4 ρb to satisfy the servi- (Ef < 50 GPa); (2) Grade II (50 GPa ≤ Ef < 60 GPa) for GFRP-1; and
ceability requirements for GFRP-RC beams. (3) Grade III (Ef ≥ 60 GPa) for GFRP-2 and GFRP-3. The beam
This paper investigates the flexural behavior and serviceability
performance of concrete beams reinforced with GFRP bars with dif-
ferent modulus of elasticity and surface profiles. A total of 17 full-scale
GFRP-RC were tested to failure in four-point bending over a clear span
of 3750 mm. The test results are reported in terms of deflection, crack
width, strains in concrete and reinforcement, flexural capacity, mode of
failure, and deformability. The design provisions are assessed and the
predicted results are compared with the measured values.

2. Experimental program

Seventeen full-scale GFRP-RC beams were tested under four-point


bending until failure. The beams were designed to fail in compression
(over-reinforced), which is the common design concept for FRP-RC
members, as recommended in design guides and standards. Some Fig. 1. GFRP reinforcing bars.

607
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Table 1
Properties of the GFRP bars.
Bar type Designated Nominal cross- Surface Tensile strength, ffu, Modulus of elasticity, Ultimate Guaranteed tensile Average bond
diameter sectional area profile (MPa) Ef, (GPa) strain, (%) strengtha, (MPa) strength, (MPa)
(mm2) (average ± SD) (average ± SD)

GFRP-1 13 129 Sand-coated 817 ± 9 48.7 ± 0.6 1.7 790 10.58


(G1) 15 199 Sand-coated 751 ± 23 48.1 ± 1.6 1.6 683 14.14
20 284 Sand-coated 728 ± 24 47.6 ± 1.7 1.5 656 13.59
22 387 Sand-coated 693 ± 23 46.4 ± 1.5 1.5 625 N/A
25 510 Sand-coated 666 ± 74 53.2 ± 2.1 1.3 444 10.11

GFRP-2 13 129 Sand-coated 1639 ± 61 67.0 ± 1.0 2.5 1456 12.20


(G2) 15 199 Sand-coated 1362 ± 33 69.3 ± 3.2 2.0 1263 14.90
20 284 Sand-coated 1082 ± 37 52.5 ± 1.7 2.1 971 12.94
25 510 Sand-coated 1132 ± 23 66.3 ± 0.9 1.7 1063 15.86

GFRP-3 15 199 Helically- 1245 ± 45 59.5 ± 1.1 2.1 1110 11.43


(G3) grooved
25 510 Helically- 906 ± 29 60.3 ± 2.9 1.5 819 10.41
grooved

a
Guaranteed strength = average − 3 × standard deviation (SD).

specimens were cast with normal-strength, ready-mixed concrete. moment zone included only two stirrups spaced at 300 mm to maintain
Table 2 lists the average concrete compressive and split-cylinder tensile the location of the longitudinal bars while minimizing the confining
strengths based on three 150 × 300 mm cylinders on the day of beam effect of the stirrups, which could affect cracking behavior. Table 2
testing for each test. provides the details of the beams. Fig. 4 shows the geometry, re-
inforcement configuration, and instrumentation details.
2.2. Test specimens
2.3. Instrumentation of test specimens
The 17 full-scale simply supported concrete beams measured
4250 mm long × 200 mm wide × 400 mm deep. The beams were re- Fig. 4 provides the details of beam instrumentation. The deflection
inforced in tension (bottom) with GFRP bars in one- and two-layer along the beam span was monitored with five linear variable differ-
configurations. The beams were divided into five series according to ential transducers (LVDTs). Crack propagation was monitored during
GFRP-bar diameter. The entire beams were reinforced in compression testing until failure, and the crack widths of the first flexural cracks
(top) with two 10 M steel bars (100 mm2, nominal dia- were monitored with one or two LVDTs. The strains of the longitudinal
meter = 11.3 mm). 10 mm steel stirrups spaced at 100 mm were in the GFRP reinforcing bars were captured with six electrical-resistance
two shear spans to avoid shear failure. In addition, the constant strain gauges, while the compressive concrete strains at the mid-span

Fig. 2. Pullout testing details: (a) Specimen details and (b) Instrumentation and testing.

608
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

corresponding loads were recorded. Fig. 5 shows the test setup during
beam testing.

3. Test results and observations

This section provides the test results of the GFRP-RC beams tested
herein. It is worth mentioning that the size effect always occurs in re-
inforced concrete beams under bending if their failure occurs in con-
crete. If the failure occurs in beams through reinforcement yielding, the
size effect does not occur. Therefore, the experimental results depend
on the beam size. Further investigation, however, is needed to quantify
this effect.

3.1. Cracking moment

The beams were observed during the test until the first crack ap-
peared and the corresponding load was recorded. The cracking load
was also verified from the change in stiffness of the moment–deflection
and moment–strain relationships. Table 3 provides the cracking mo-
ments of the tested beams, excluding beam self-weight. The cracking
moment ranged from 10.92 to 16.77 kN·m. The cracking moment was
directly proportional to the concrete’s tensile strength, which, in turn, is
a function of compressive strength: the higher the concrete strength, the
higher the cracking moment. The cracking moments were predicted
using Eq. (1). The experimental-to-predicted ratios for the cracking
moment, Mcr, are also listed in Table 3.
Mcr = (fr / yt ) Ig (1)

fr = 0.62 fc′ ACI [2] (2a)

fr = 0.6 fc′ ISIS [17], CSA [9] (2b)

fr = 0.4 fc′ CSA [10] (2c)


Table 3 shows that ACI [2] and CSA [9] yielded unconservative pre-
dictions with an average experimental-to-predicted Mcr of 0.81 ± 0.11
and 0.82 ± 0.10, respectively. On the other hand, CSA [10] yielded
conservative predictions with an average experimental-to-predicted Mcr
ratio of 1.38 ± 0.17.

3.2. Flexural capacity and mode of failure

In general, FRP-RC sections are designed as over-reinforced


(ρf > ρfb) to exhibits high deformability at the ultimate limit state. This
Fig. 3. Typical bond stress–slip relationships: (a) GFRP-1, (b) GFRP-2, and (c) design concept is common in CSA [9] and ACI [2] for FRP-RC sections.
GFRP-3. Some provisions, however, allow the use of under-reinforced designs if
additional conditions are met. The balanced reinforcement ratio is
were measured with two electrical-resistance strain gauges. In addition, generally governed by the mechanical properties of the FRP bars and
an automatic data-acquisition system connected to a computer was concrete strength as demonstrated by Eq. (3).
used to monitor loads, deflections, and strains in the concrete and re- fc′ Ef εcu
inforcement. ρfb = α1 β1
f fu Ef εcu + f fu (3)

2.4. Test setup and procedure α1 = 0.85 ACI [2] (4a)

The beams were tested under monotonic loading on four-point β1 = 0.85−0.05(fc′−27.6)/6.9 ACI [2] (4b)
bending over a clear span of 3750 mm until failure. Fig. 4 shows the
α1 = 0.85−0.0015(fc′ ) ⩾ 0.67 ISIS [17], CSA [9] (5a)
dimensions and locations of the applied loads. The load was applied at a
stroke-controlled rate of 0.6 mm/min. During testing, the beams were β1 = 0.97−0.0025(fc′ ) ⩾ 0.67 ISIS [17], CSA [9] (5b)
visually observed until the first crack appeared and the corresponding
load was recorded. Then, the test was paused and the initial crack width All beams exhibited compression failure (concrete crushing), as shown
was measured with a 50× magnifying-power hand-held microscope in Fig. 6. At failure, the concrete strain of all the tested beams was lower
and an LVDT was installed at the crack location to measure crack width than the limits provided by ACI [2] and CSA [9] (3000 and 3500 mi-
continuously as the load increased. Thereafter, the same loading rate crostrains, respectively) and supported by many studies [15,19], where
was resumed until beam failure. The formation of cracks and the cor- the concrete crushing is expected to occur at about 3000 microstrains.
responding loads were marked on the beam surface, and the This is related to the absence of closely spaced stirrups (only two

609
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Table 2
Details of the test specimens.
Series Beam a
fc' (MPa) ft (MPa) ρf (%) ρfbb (%) ρf/ρfb Af Ef (kN) Reinforcement configuration

I 3#13G1 33.50 3.60 0.56 0.43 1.31 18,347 3 No. 13 – 1 row


5#13G1 38.95 3.81 0.91 0.59 1.54 29,864 5 No. 13 – 1 row
2#13G2 33.50 3.60 0.38 0.15 2.45 17,286 2 No. 13 – 1 row

II 3#15G1 38.95 3.81 0.84 0.65 1.30 28,716 3 No. 15 – 1 row


4#15G1 38.95 3.81 1.12 0.65 1.73 38,288 4 No. 15 – 1 row
2#15G2 29.00 2.50 0.56 0.20 2.79 27,581 2 No. 15 – 1 row
2#15G3 33.83 3.11 0.56 0.21 2.69 23,681 2 No. 15 – 1 row

III 6#15G1 33.50 3.60 1.82 0.50 3.67 59,700 6 No. 15 – 2 rows
5#15G2 29.00 2.50 1.52 0.20 7.58 68,954 5 No. 15 – 2 rows
5#15G3 33.80 3.10 1.52 0.23 6.47 59,203 5 No. 15 – 2 rows

VI 2#20G1 38.95 3.81 0.81 0.69 1.61 27,037 2 No. 20 – 1 row


3#20G1 42.10 3.18 1.21 0.73 1.67 40,555 3 No. 20 – 1 row
2#22G1 38.95 3.81 1.08 0.61 1.76 35,264 2 No. 22 – 1 row
3#20G2 48.13 3.96 1.21 0.34 3.59 44,730 3 No. 20 – 1 row

V 2#25G1 48.13 3.96 1.46 0.83 1.75 54,264 2 No. 25–1 row
2#25G2 48.13 3.96 1.46 0.38 3.85 67,626 2 No. 25 – 1 row
2#25G3 33.80 3.10 1.51 0.42 3.57 61,506 2 No. 25 – 1 row

a
Beam designation: the number of GFRP bars followed by GFRP-bar diameter, with the last symbol indicating GFRP-bar type (G1: GFRP-1; G2: GFRP-2; G3: GFRP-
3).
b
Calculated according to ACI [2].

Fig. 4. Dimensions, reinforcement details, and instrumentation.

stirrups spaced at 300 mm) in the constant-moment zone of the beams.


This resulted in premature buckling of the steel bars in the top section
of the beams before yielding and the consequent disintegration of top
extreme fiber of the concrete. Accordingly, the strain gauges were no
longer able to capture the strain in the concrete at the extreme fiber.
Thus, all of the beams exhibited compression failure (concrete
crushing), as shown in Fig. 6.
The results indicated that, generally, the higher the axial-re-
inforcement stiffness (Ef Af), the higher the flexural capacity of the
tested beams. In addition, the concrete strength contributed to the load-
carrying capacity, as evidenced from by 2#25G2 and 5#15G2 with the
same Ef Af. Beam 2#25G2 showed a load-carrying capacity of
167.24 kN·m compared to 129.32 kN·m for 5#15G2; the concrete
strengths for both beams were 48.13 MPa and 29.00 MPa, respectively.
The ACI [2] and CSA [9,10] predicted load-carrying capacity were
in good agreement with the experimental results, as shown in Table 3.
Fig. 5. Test setup during a beam testing.
The differences between the ACI [2] and CSA [9,10], however, were

610
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Table 3
Experimental and predicted cracking and ultimate moments.
Beam Experimental moments ACI [2] CSA [10] CSA [9] Eurocode [14]

Mcr Mn Failure Mcr Mn Failure Mcr Mn Failure Mcr Mn Failure Mcr Mn Failure
mode a mode a mode 2 mode a mode a
kN.m kN.m Exp/pred Exp/pred Exp/pred Exp/pred Exp/pred Exp/pred Exp/pred Exp/pred

3#13G1 13.46 81.34 C.C. 0.85 0.89 C.C. 1.49 0.82 C.C. 0.88 0.82 C.C. 0.69 0.76 C.C.
5#13G1 15.26 130.6 C.C. 0.88 1.1 C.C. 1.52 1.01 C.C. 0.76 1.01 C.C. 0.73 0.78 C.C.
2#13G2 13.75 82.78 C.C. 0.87 0.94 C.C. 1.52 0.87 C.C. 0.9 0.87 C.C. 0.70 0.62 C.C.
3#15G1 12.21 101.3 C.C. 0.73 0.89 C.C. 1.26 0.82 C.C. 0.75 0.82 C.C. 0.59 0.68 C.C.
4#15G1 15.61 138.2 C.C. 0.88 1.06 C.C. 1.56 0.98 C.C. 0.94 0.98 C.C. 0.74 0.70 C.C.
2#15G2 11.22 95.93 C.C. 0.77 0.96 C.C. 1.37 0.9 C.C. 0.81 0.9 C.C. 0.82 0.56 C.C.
2#15G3 10.92 91.31 C.C. 0.69 0.9 C.C. 1.2 0.84 C.C. 0.71 0.84 C.C. 0.64 0.57 C.C.
6#15G1 11.98 118.3 C.C. 0.76 0.89 C.C. 1.32 0.84 C.C. 0.79 0.84 C.C. 0.60 0.44 C.C.
5#15G2 12.2 129.3 C.C. 0.84 0.98 C.C. 1.49 0.93 C.C. 0.88 0.93 C.C. 0.87 0.32 C.C.
5#15G3 12.61 110.6 C.C. 0.79 0.82 C.C. 1.38 0.77 C.C. 0.82 0.77 C.C. 0.73 0.29 C.C.
2#20G1 15.36 107.4 C.C. 0.89 0.96 C.C. 1.53 0.88 C.C. 0.92 0.88 C.C. 0.74 0.78 C.C.
3#20G1 16.32 140.4 C.C. 0.9 1.04 C.C. 1.55 0.95 C.C. 0.93 0.95 C.C. 0.93 0.69 C.C.
2#22G1 12.88 132.3 C.C. 1 1.08 C.C. 1.33 1 C.C. 0.8 1 C.C. 0.61 0.75 C.C.
3#20G2 12.29 171.4 C.C. 0.63 1.15 C.C. 1.07 1.04 C.C. 0.65 1.04 C.C. 0.56 0.57 C.C.
2#25G1 11.32 161.7 C.C. 0.58 1.02 C.C. 0.98 0.92 C.C. 0.6 0.92 C.C. 0.51 0.73 C.C.
2#25G2 16.77 167.2 C.C. 0.85 0.97 C.C. 1.46 0.88 C.C. 0.89 0.88 C.C. 0.75 0.46 C.C.
2#25G3 13.2 115.9 C.C. 0.83 0.81 C.C. 1.45 0.76 C.C. 0.86 0.76 C.C. 0.76 0.39 C.C.

Average 0.81 0.97 – 1.38 0.89 – 0.82 0.89 – 0.70 0.59 –


Standard deviation 0.11 0.1 – 0.17 0.08 – 0.1 0.08 – 0.11 0.16 –
Coefficient of variation (%) 13% 10% – 12% 9% – 12% 9% – 16% 27% –

Note – Measured and predicted moments exclude beam self-weight.


a
C.C.: Concrete crushing.

related to the β1 factor and the assumed strain at the ultimate, which is nominal moment capacity of the beams. It is worth mentioning that the
0.003 for ACI [2] and 0.0035 for CSA [9,10]. Table 3 also lists the 30% of the nominal flexural capacity (0.30Mn) was selected as Mota
Eurocode [14] predictions through replacing the steel properties with et al. [20], Kassem et al. [19], and El-Nemr et al. [13] reported that
GFRP ones. 0.30Mn is a reasonable value for the service load of FRP-RC beams and
The neutral axis-to-depth ratios (c/d) were calculated from the one-way slabs. The neutral-axis depth increased as the reinforcement
measured strains and reported in Table 4. The c/d values reveal slight ratio increased, since the equilibrium of forces requires a larger com-
changes in the c/d ratios between 0.30Mn and Mn, where Mn is the pression block for the greater tensile forces. The theoretical neutral

Fig. 6. Typical compression failure of beams reinforced with types G2 and G3 GFRP bars.

611
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Table 4
Strains, neutral axis-to-depth ratio, and curvature of test specimens.
Beam Strain in concrete1 (microstrains) Strain in FRP1 (microstrains) c/d Curvature, ψ, 1/d

1 1
At 2000 0.30Mn Mn 0.30Mn Mn At 2000 0.30Mn Mn Theo. At 20001 0.30Mn Mn

3#13G1 −203 −314 −1561 4378 13,726 0.09 0.07 0.10 0.13 0.002 0.005 0.015
5#13G1 −421 −714 −1933 3994 15,095 0.17 0.15 0.15 0.15 0.002 0.005 0.011
2#13G2 −173 −690 −2541 5349 16,359 0.05 0.11 0.13 0.12 0.003 0.006 0.019
3#15G1 −388 −651 −2341 4434 13,345 0.16 0.13 0.15 0.15 0.002 0.005 0.008
4#15G1 −409 −510 −1816 2405 13,489 0.14 0.14 0.13 0.17 0.002 0.003 0.014
2#15G2 −586 −346 −1454 4189 14,136 0.08 0.08 0.09 0.16 0.008 0.005 0.016
2#15G3 −294 −516 −2129 3448 10,277 0.13 0.13 0.17 0.14 0.002 0.004 0.012
6#15G1 −488 −562 −1976 2367 7693 0.19 0.19 0.20 0.22 0.003 0.003 0.010
5#15G2 −729 −745 −2959 2053 7550 0.28 0.28 0.29 0.24 0.003 0.003 0.010
5#15G3 −521 −412 −1839 1571 6430 0.21 0.21 0.22 0.22 0.003 0.002 0.008
2#20G1 −282 −433 −2090 3860 13,372 0.11 0.09 0.13 0.15 0.002 0.004 0.015
3#20G1 −439 −657 −3087 3107 6794 0.18 0.18 0.20 0.18 0.002 0.004 0.015
2#22G1 −550 −605 −2646 4262 13,651 0.12 0.12 0.16 0.17 0.004 0.005 0.016
3#20G2 −380 −657 −2648 3755 11,823 0.16 0.15 0.18 0.18 0.002 0.004 0.014
2#25G1 −371 −542 −2529 2991 10,028 0.18 0.15 0.20 0.19 0.002 0.004 0.013
2#25G2 −473 −508 −2045 2127 7573 0.16 0.16 0.18 0.21 0.002 0.003 0.009
2#25G3 −575 −459 −1627 1666 6429 0.22 0.22 0.20 0.22 0.003 0.002 0.008

axis-to-depth ratios (c/d) were also calculated based on strain com- This range is in agreement with that of 0.008/d to 0.015/d reported by
patibility using a cracked-section analysis, and the values were in El-Nemr et al. [13], 0.0089/d to 0.012/d reported by Gulbrandsen [16],
agreement with the measured ones. and 0.009/d to 0.014/d reported by Kassem et al. [19] for GFRP-RC
Moreover, the curvature (Ψ) was also calculated at 0.30Mn and Mn beams.
(see Table 4. The calculated Ψ values at Mn were 3–4 times the values
calculated at 0.30Mn, which confirms that the curvature is proportional 3.3. Strain in reinforcement and concrete
to the applied flexural moment. The curvature (Ψ) values at Mn ranged
from 0.008/d to 0.019/d for the compression-controlled failure mode Fig. 7 shows the mid-span strains versus the applied moment, while
(beams tested herein, since they showed concrete-crushing failure). Table 4 provides the strains at 0.30Mn and failure. The concrete strains

Fig. 7. Comparison of concrete and reinforcement strains.

612
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

corresponding to 2000 microstrains in the GFRP bars are also pre- findings of Vijay and GangaRao [23] for GFRP bars with different
sented. All GFRP-RC beams failed in compression by concrete crushing, surface profiles.
as they were designed with ρf > ρfb. The concrete crushing failure in
some beams was triggered by the buckling of the steel bars on the
compression side of the beams. This was evidenced by the small peak 3.4. Deflection behavior
values for the concrete strains at failure. Thus, the concrete strain va-
lues at ultimate (failure) may have been affected by the buckling of Fig. 8 provides moment versus mid-span deflection relationships for
compression steel bars. the tested beams. Each curve represents the average deflection obtained
The moment–strain in GFRP bars relationship was a typical bilinear from two LVDTs mounted at beam mid-span. The moment–deflection
(Fig. 7). Increasing the reinforcement ratio ρf decreased the strain in the relationships revealed that ρf is directly impacting the beam stiffness
GFRP bars at the same load level. Beams with low ρf (Series I and II) and moment–deflection behavior where the highest deflection was
experienced a sharp reinforcement-strain increase at cracking that ex- corresponding to the lowest ρf. Table 5 summarizes the deflection of the
ceeded 3000 microstrains. Although beams 2#13G2 and 3#13G1 were tested beams at 0.30Mn.
designed with ρf > ρfb (over-reinforced), they evidenced a large in- The relationships in Fig. 8 indicate that, regardless of the small
crease in crack width due to poor energy absorption at cracking. differences in concrete strength between the beams, increasing the Ef Af
Maintaining minimum practical reinforcement may be important en- contributed to lower beam deflection. Moreover, the beams with the
sure that the section behaves reasonably after cracking. ACI [2] spe- same Ef Af and reinforced with the same FRP bar type and surface
cifies a minimum ρf /ρfb of 1.4; however, this limit did not prevent beam profile showed the same moment–deflection relationships, as in the
2#13G2 from exhibiting a sharp strain increase at cracking. case of 5#15G3 and 2#25G3 and in the case of 5#15G2 and 2#25G2.
Furthermore, Fig. 7 indicates that beams 2#25G3 and 5#15G3 had Thus, it could be concluded that using the equivalent Ef Af of GFRP bars
the same moment–strain relationship. Beams reinforced with GFRP bars with the same bar type and surface profile would not affect the mo-
of similar surface profile and similar Ef Af are expected to have the same ment–deflection properties. Changing the surface profile, however, may
moment–strain relationships. Therefore, 2#25G2 and 5#15G2 should affect the behavior as it will impact the bond characteristics and
have behaved similarly, but 5#15G2 had lower concrete strength than cracking performance.
2#25G2 (29.0 MPa and 48.13 MPa, respectively) which resulted in The deflection values were plotted against ρf/ρfb and curvature, as
higher stains in 5#15G2. shown in Figs. 9(a) and 7(b), respectively. Fig. 9(a) shows that the
The curvature of the tested beams was calculated as a function of 1/ deflection of the beam specimens at 0.30Mn was less than 15.63 mm
d (see Table 4. Vijay and GangaRao [23] concluded that, at a curvature (assumed deflection limit at service, L/240 = 15.63 mm[9], except
limit of 0.005/d, the strains in FRP are typically as high as 4500 mi- 3#15G1, 2#15G2, 2#20G1, and 2#22G1, which exhibited larger de-
crostrains. The test results revealed that, at a curvature of 0.005/d, the flections. Vijay and GangaRao [22] concluded that a curvature limit of
strains in the GFRP bars were 4378, 3994, 4189, 4434, and 4262 mi- 0.005/d may yield GFRP-RC beams satisfying deflection and cracking
crostrains for 3#13G1, 5#13G1, 3#15G1, 2#15G2, and 2#22G1 with serviceability criteria. The deflection–curvature relationships
ρf/ρfb of 1.31, 1.54, 1.3, 2.79, and 1.76, respectively. This confirms the (Fig. 9(b)) indicate that most of the GFRP-RC beams with a curvature of
less than 0.005/d yielded deflection values lower than 15.63 mm (L/

Fig. 8. Deflection vs. applied moment for beams reinforced with different GFRP grades: (a); (b) GFRP-1, (c) GFRP-2, and (d) GFRP-3.

613
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Table 5
Experimental-to-predicted deflection ratios (δexp/δpred).
Beam Measured deflection (mm) δexp/δpred δexp/δpred δexp/δpred δexp/δpred

ACI [1] ISIS [17] CSA [9] ACI [2]

0.30Mn 0.67Mn 0.30Mn 0.67Mn 0.30Mn 0.67Mn 0.30Mn 0.67Mn 0.30Mn 0.67Mn

3#13G1 15.38 41.00 1.17 0.97 0.82 0.87 0.76 0.85 1.29 1.03
5#13G1 15.45 39.00 1.12 1.05 0.90 0.95 0.85 0.93 1.26 1.10
2#13G2 15.07 48.00 1.65 1.17 0.75 0.99 0.70 0.96 1.22 1.16
3#15G1 18.00 48.00 1.37 1.31 1.09 1.19 1.03 1.18 1.52 1.38
4#15G1 15.45 39.00 1.20 1.17 1.02 1.08 0.97 1.07 1.36 1.22
2#15G2 16.00 41.00 1.53 1.22 1.05 1.11 0.99 1.09 1.42 1.25
2#15G3 14.00 37.00 1.33 0.99 0.80 0.87 0.75 0.86 1.18 1.01
6#15G1 9.43 27.00 0.95 1.05 0.79 0.97 0.76 0.97 0.97 1.07
5#15G2 10.98 29.00 1.29 1.28 1.06 1.19 1.03 1.18 1.28 1.30
5#15G3 7.92 24.00 0.87 0.92 0.66 0.84 0.63 0.83 0.82 0.94
2#20G1 17.99 45.00 1.22 1.16 1.02 1.05 0.96 1.03 1.47 1.23
3#20G1 15.00 37.00 1.13 1.11 0.93 1.00 0.89 0.99 1.29 1.16
2#22G1 19.00 46.00 1.50 1.38 1.23 1.26 1.16 1.25 1.66 1.44
3#20G2 15.00 38.00 1.42 1.16 0.94 1.00 0.93 1.01 1.32 1.18
2#25G1 15.00 38.00 1.21 1.24 0.93 1.00 1.43 1.29 1.38 1.29
2#25G2 10.00 27.00 0.99 1.00 0.77 0.89 0.75 0.88 1.00 1.02
2#25G3 13.15 37.00 1.10 1.03 0.79 0.94 0.76 0.93 1.03 1.05

Average 1.24 1.13 0.91 1.01 0.90 1.02 1.26 1.17


Standard deviation 0.21 0.13 0.15 0.12 0.20 0.14 0.21 0.14
Coefficient of variation (%) 17% 12% 16% 12% 22% 14% 17% 12%

240), which is in agreement with Vijay and GangaRao [22]. the helically-grooved bars than in those reinforced with the sand-coated
bars. This indicates that the sand-coated GFRP bars have better bond
characteristics than the helically-grooved GFRP bars.
3.5. Crack pattern and crack width
Fig. 11 gives the moment–crack-width relationships for the tested
beams, while Table 6 lists the crack widths corresponding to 2000
The beams were initially uncracked except 6#15G1, which showed
microstrains in the FRP bars and at 0.30Mn. Generally, increasing the Ef
a hair crack before testing. At cracking moment Mcr, cracks propagated
Af decreased the crack widths at all load levels. Fig. 12 also confirms
vertically and perpendicularly to the maximum tensile stress induced at
that beams with the same Ef Af of GFRP bars of the same type and
the extreme fiber of the beam. The cracks in the constant-moment zone
surface profile are expected to exhibit similar moment-to-crack width
grew vertically, although the cracks in the shear span were inclined
relationship, as in the case of beams 5#15G3 and 2#25G3 and beams
toward the central zone due to shear stresses in this region. After
5#15G2 and 2#25G2.
0.67Mn, the crack pattern stabilized; no new cracks appeared and the
The FRP design provisions permit a larger crack width for FRP-RC
existing cracks widened till failure.
elements compared to their counterparts reinforced with steel. CSA
The crack patterns of tested beams at 0.30Mn and 0.67Mn, shown in
[9,10] specifies a service-limiting flexural-crack width of 0.50 mm for
Fig. 10, reveal that increasing ρf, increased the number of cracks and,
exterior exposure (or aggressive environmental conditions) and
consequently, reduced the average crack spacing. The patterns show
0.70 mm for interior exposure. ACI [2] recommends using CSA [10]
that increasing the bar diameter from No. 15 to No. 25 (15.9–25.4 mm
limits for most cases. On other hand, since there is a direct relationship
diameter) in the case of the sand-coated (G2) and helically-grooved
between the strain in the reinforcing bars and crack width, ISIS [17]
(G3) bars resulted in fewer cracks. In addition, increasing the Ef Af
specifies 2000 microstrains as a strain limit in FRP bars to control crack
yielded smaller initial crack widths, except in the case of 6#15G1,
width. Considering the measured crack-widths reported in Table 6, it
which seemed to be affected by preexisting hair cracks. Furthermore,
can be seen that, at 0.30Mn, half the beams showed crack widths less
the comparison also revealed fewer cracks in the beams reinforced with

Fig. 9. Deflection versus (a) ρf/ρfb and (b) curvature Ψ 1/d.

614
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Fig. 10. Crack patterns of beam specimens in Series I, III, and V at 0.30Mn and 0.67Mn.

than 0.7 mm, while the other half was close to 0.7 mm, excluding beams 4. Discussion
2#13G2 and 2#15G2. On the other hand, at 2000 microstrains in the
GFRP bars, the beams showed crack widths less than 0.5 mm, except 4.1. Influence of number of bars
beams 2#13G2 and 2#15G2. This is in agreement with ISIS [17].
Furthermore, to investigate the effect of limiting the curvature of GFRP- Generally, increasing the number of bars (of the same mechanical
RC beams to 0.005/d to satisfy the cracking serviceability criterion properties) reduces the strains at the same load and the increases the
[23], the relationship between the crack width and curvature was flexural capacity due to the increase in the Ef Af. In addition, when the
plotted as shown in Fig. 12. The crack width–curvature relationship mechanical properties were different, the beams with similar Ef Af were
indicates that limiting the curvature of the FRP-RC beams may be used considered: (2#13G2; 3#13G1 with Ef Af of 17.286; 18.347 MN, re-
to satisfy the serviceability requirements. More investigations, how- spectively) and (2#15G2; 3#15G1, with Ef Af of 27.581; 28.716 MN,
ever, are needed to verify the limit. respectively). The strains at 0.30Mn decreased from 5349 to 4378 mi-
crostrains for 2#13G2 and 3#13G1, respectively. Beams 2#15G2 and
3#15G1, however, did not show the same trend because of the lower
cracking load of 3#15G1 compared with that of 2#15G2, which

615
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

Fig. 11. Crack width vs. applied moment for beams reinforced with different GFRP bars: (a) and (b) for GFRP-1, (c) for GFRP-2, and (d) for GFRP-3.

Table 6
Experimental-to-predicted crack widths (wexp/wpred).
Beam Measured crack width, wexp (mm) wexp/wpred

Initial 2000 µɛ 0.30Mn ACI [1] ISIS [17] CSA [10]

2000 µɛ 0.30Mn 2000 µɛ 0.30Mn 2000 µɛ 0.30Mn

3#13G1 0.10 0.40 0.78 1.11 1.03 1.02 0.88 1.95 1.68
5#13G1 0.05 0.23 0.46 0.76 0.86 0.70 0.75 1.33 1.39
2#13G2 0.20 0.77 1.03 1.62 0.95 1.71 0.94 2.81 1.58
3#15G1 0.10 0.37 0.85 1.04 1.39 0.9 1.13 1.91 2.26
4#15G1 0.08 0.38 0.41 1.17 0.85 1.03 0.69 2.05 1.38
2#15G2 0.22 0.62 1.18 1.35 1.54 1.40 1.52 2.23 2.56
2#15G3 0.16 0.31 0.63 0.66 0.71 0.69 0.69 0.92 0.92
6#15G1 0.05 0.14 0.15 0.36 0.18 0.31 0.15 0.64 0.31
5#15G2 0.09 0.30 0.31 0.81 0.55 0.68 0.44 1.40 0.92
5#15G3 0.12 0.53 0.40 1.40 0.59 1.20 0.48 2.03 0.54
2#20G1 0.10 0.33 0.73 0.71 0.83 0.70 0.76 1.25 1.33
3#20G1 0.15 0.32 0.51 0.87 0.94 0.75 0.73 1.54 1.48
2#22G1 0.09 0.47 0.51 0.98 0.70 0.96 0.63 1.73 1.13
3#20G2 0.11 0.38 0.66 1.08 1.22 0.93 0.95 1.83 1.92
2#25G1 0.10 0.36 0.55 0.76 0.87 0.72 0.66 1.31 1.37
2#25G2 0.06 0.33 0.36 0.70 0.65 0.66 0.56 1.21 1.02
2#25G3 0.13 0.53 0.45 1.09 0.86 1.05 0.78 1.53 1.14

Average 0.97 0.87 0.91 0.75 1.63 1.35


Standard deviation 0.31 0.32 0.32 0.30 0.52 0.57
Coefficient of variation (%) 32% 37% 36% 40% 32% 42%

616
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

yields better cracking behavior and yields small crack widths. In ad-
dition, as evidenced in Table 3, the beams with similar Ef Af showed the
same load-carrying capacity as beams 4#15G1; 3#20G1 and 5#15G3;
2#25G3.
It is worth mentioning that concrete strength may affect mo-
ment–strain relationships, even in beams with similar Ef Af, as cracking
moment (which is directly proportional to concrete strength) controls
the point at which a significant increase in reinforcement strain starts.
In addition, the surface profile may also affect the measured strains and
crack widths, as indicated in Fig. 11, where beam 6#15G1 with sand-
coated bars evidenced crack widths significantly narrower than those in
beam 5#15G3 with helically-grooved bars, despite the similar Ef Af.
This tends to confirm that sand-coated bars behave better than de-
formed bars in controlling strains and cracks due to their uniform
surface and the absence of discontinuity points, as in the case of de-
formed bars. This can also be confirmed from Fig. 10, in which the
distributed cracks in 6#15G1 can be observed in comparison with beam
Fig. 12. Crack width versus curvature Ψ 1/d for tested beams.
5#15G3.

affected the strain values. Furthermore, the crack width at 0.30Mn de-
4.3. Influence of modulus of elasticity
creased from 1.03 to 0.78 mm for 2#13G2 and 3#13G1, and from 1.18
to 0.85 mm for 2#15G2 and 3#15G1 respectively. Thus, it can be
Since the Ef Af is affecting the general behavior of the GFRP-RC
concluded that the use of more number of reinforcing bars, even when
beams, increasing the bar modulus, while maintaining the surface
the modulus of elasticity is smaller, can enhance the strain and cracking
profile is expected to enhance performance, as can be observed in
performances. On the other hand, deflection of these beams was not
beams 2#25G1 and 2#25G2. Increasing the modulus of elasticity from
significantly affected since they had the same Ef Af (2#15G2; 3#15G1
53.2 to 66.3 GPa for the bars with the same surface profile (sand-coated
and 2#13G2; 3#13G1).
in this case) increased the load-carrying capacity and enhanced the
strain, cracking, and deflection behavior. Similarly, beams 3#20G1 and
4.2. Influence of axial stiffness (Ef Af) and surface profile 3#20G2 showed the same behavior when the modulus increased from
47.6 to 52.5 GPa. In addition, since the design of GFRP-RC sections is
Since the beams tested herein were provided with different re- normally controlled by serviceability limit state rather than ultimate
inforcement ratios of GFRP reinforcing bars with different mechanical limit state, increasing the modulus of elasticity is expected to reduce the
properties and surface profiles, axial-reinforcement stiffness (Ef Af) was amount of GFRP bars needed to satisfy the design.
used for comparison. The test results show that the beams with the
same Ef Af of GFRP bars with the same surface profile exhibited similar 4.4. Comparison with predicted results
behavior. This can be seen in Figs. 5, 6, and 9 looking at beams 4#15G1
and 3#20G1. Both beams 4#15G1 and 3#20G1 showed the same The predicted deflections, using the equations in Table 7, were
moment-strain and moment-deflection relationships. Similarly, beams compared against the measured deflections at 0.30Mn and 0.67Mn.
5#15G3 and 2#25G3 showed similar behavior. The crack widths in the Table 5 provides the experimental-to-predicted deflection ratios (δexp/
beams reinforced with smaller diameter GFRP bars, however, were δpred). The comparison indicates that ACI [1] and ACI [2] equations
narrower than those in the beams with larger-diameter bars. Thus, underestimated the deflection at 0.30Mn with overall average δexp/δpred
using small-diameter GFRP bars, assuming similar surface profiles, of 1.24 ± 0.21 and 1.26 ± 0.21, respectively. Both CSA [9] and ISIS

Table 7
Deflection and crack width provisions.
Deflection provisions

ACI Committee 440 ISIS [17] CSA [9]

ACI [1] Ig Icr


PL3 a 3 Lg 3
Mcr 3 Mcr 3⎤
Ie =
M
Icr + ⎛1 − 0.5( cr )2⎞ (Ig − Icr )
(9) δmax =
24Ec Icr
⎡3
⎢ ( ) 4 ( ) −8η ( ) ⎤⎥⎦
a
L L L
(10a)
Ie = ( )
Ma
βd Ig + ⎡1−


( )⎦
Ma ⎥ cr
I ⩽ Ig (6a)

⎝ Ma ⎠

bd3 3

Icr = k + nf Af d 2 (1−k )2 (10b)
3
βd = 0.2(ρf / ρfb ) ⩽ 1.0 (6b) I
η= 1− cr (10c)
Ig
ACI [2]
Icr
Ie = 2 ⩽ Ig (7a)
M ⎡ I ⎤
1 − γ ⎛ cr ⎞ ⎢1 − cr ⎥
⎜ ⎟

⎝ Ma ⎠ ⎣ Ig ⎦

γ = 1.72−0.72 ( )
Mcr
Ma
(7b)
ACI [1], ACI [2] for two-point load:
PL
δmax = [3L2−4a2] (8)
24Ec Ie

Crack width provision


ACI [1] ISIS [17] CSA [10]

ff h ff h ff h
s 2
w=2 2
Ef h1
kb dc2 + ()
2
(11) w = 2.2kb 23
Ef h1
dc A (12) w=2 2
Ef h1
kb dc2 + (s /2)2 (13)

617
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

[17] equations, however, slightly underestimated the deflection for all absence of discontinuity points.
of the beams at 0.67Mn with average δexp/δpred of 1.02 ± 0.14 and 5. At a strain of 2000 microstrains in the GFRP bars, most of the tested
1.01 ± 0.12, respectively. On the other hand, CSA [9] and ISIS [17] beams showed crack widths under 0.5 mm, which is in agreement
equations yielded reasonable yet conservative deflection values for all with ISIS recommendations (2007). At 0.30Mn, however, the strains
of the beams at 0.30Mn with average δexp/δpred of 0.90 ± 0.20 and in the GFRP bars in almost all the beams exceeded 2000 micro-
0.91 ± 0.15, respectively. ISIS [17] equation, however, had the lowest strains and the corresponding crack widths exceeded 0.50 mm.
coefficient of variation (COV) amongst the equations used (16%). 6. The curvature of the GFRP-RC members at service load can be
Table 6 compares the measured crack widths, using equations in considered as a limit to control crack width and deflection. The
Table 7, to the predicted values. The comparison was conducted at curvature limit of 0.005/d provided by Vijay and GangaRao [23]
three different load levels: 2000 microstrains in the GFRP reinforcing seems to be feasible for most of the tested GFR-RC beams. More
bars, 0.30Mn, and 0.67Mn. The predictions were made with a kb value of verification and refinement, however, is needed to confirm this
1.4 for ACI [1] and 1.2 for ISIS [17]. For CSA [10], the predictions were limit.
made with a kb value of 0.8 for the sand-coated GFRP bars and 1.0 for 7. At 2000 microstrains in the FRP, both ACI [1] and ISIS [17] yielded
the helically-grooved GFRP bars. The average wexp/wpred at 0.30Mn good crack-width predictions, on average. CSA [10], however, un-
calculated according to ACI [1], CSA [10], and ISIS [17] were derestimated crack widths. At 0.30Mn, ACI [1] and CSA [10] yielded
0.87 ± 0.32, 1.35 ± 0.57 and 0.75 ± 0.30, respectively. The degree non-conservative crack widths, while ISIS [17] provided values in
of conservativeness was higher for the GFRP-RC beams with the mul- good agreement with the test results. The conservativeness for all
tilayer GFRP bar configuration. It should be mentioned that the ex- the predictions was higher for the GFRP-RC beams with multilayer
perimental determination of the bond-dependent coefficient (kb) may GFRP bars.
enhance the crack-width predictions. 8. ACI [1] and ACI [2] underestimated beam deflection at 0.30Mn,
At 2000 microstrains, both ACI [1] and ISIS [17] yielded good with average δexp/δpred of 1.24 ± 0.21 and 1.26 ± 0.21, respec-
predictions for all of the beams with average wexp/wpred of 0.97 ± 0.31 tively. On the other hand, ISIS [17] and CSA [9] provided con-
and 0.91 ± 0.32, respectively. CSA [10], however, underestimated the servative deflection values with average δexp/δpred of 0.91 ± 0.15
crack widths with an average wexp/wpred of 1.63 ± 0.52. Furthermore, and 0.90 ± 0.20 at 0.30Mn, respectively.
ACI [1] returned the lowest COV at 2000 microstrains, 0.30Mn, and
0.67Mn. Acknowledgements
As shown in Table 6, three beams had crack widths exceeding
0.50 mm. Thus, maintaining the strain level in the GFRP reinforcing The authors wish to acknowledge the financial support of the
bars at 2000 microstrains at service load may keep the crack width Natural Sciences and Engineering Research Council of Canada (NSERC),
below 0.50 mm, as recommended by ISIS [17]. the NSERC Research Chair in Innovative FRP Reinforcement for
Concrete Structures, the Fonds de la recherche du Quebec en nature et
5. Summary and conclusions technologies (FRQ-NT), and the Ministry of Transportation of Quebec.
The authors would like to thank Pultrall Inc. (Thetford Mines, QC,
This paper reports on the flexural behavior of concrete beams re- Canada) and Fiberline Composite Canada Inc. (Kitchener, ON, Canada)
inforced with different types of GFRP bars. A total of 17 full-scale for donating the GFRP materials, and the technical staff of the structural
beams (measuring 4250 mm long × 200 mm wide × 400 mm deep), & materials lab in the Dept. of Civil Engineering at the University of
reinforced with sand-coated and helically-grooved GFRP bars, were Sherbrooke.
constructed and tested under four-point bending until failure. Based on
the results and discussions presented herein, the following conclusions References
have been drawn:
[1] ACI Committee 440. Guide for the design and construction of concrete reinforced
1. The tested GFRP-RC beams failed due to concrete crushing as they with FRP bars (ACI 440.1R-03). Farmington Hills, MI: American Concrete Institute;
2003. p. 44.
were designed as over-reinforced. The beams showed distributed [2] ACI Committee 440. Guide for the design and construction of concrete reinforced
flexural cracks with no signs of slippage of the tensile reinforcement. with FRP bars (ACI 440.1R-15). Farmington Hills, MI: American Concrete Institute;
The beams showed typical bilinear behavior for strain and deflec- 2015. p. 85.
[3] Ahmed EA, Benmokrane B, Sansfacon M. Case study: design, construction, and
tion until failure. The pre-cracking response of all the beams was not performance of the La Chancelière Parking Garage’s concrete flat slabs reinforced
affected by the reinforcement ratio, since it is governed by the gross with GFRP bars. ASCE J Compos Constr 2017;21(1):05016001. 15 p.
concrete section. [4] Arivalagan S. Engineering performance of concrete beams reinforced with GFRP
bars and stainless steel. Struct Eng, Glob J Inc. 2012;12(1):1–6.
2. Axial-reinforcement stiffness (Ef Af) governs the flexural behavior of
[5] ASTM D7205. Tensile properties of fiber reinforced polymer matrix composite bars.
FRP-RC members: the higher the Ef Af, the better the flexural per- Conshohocken, USA: American Society for Testing and Materials; 2011. p. 12.
formance (higher load-carrying capacity, lower deflection, and [6] ASTM D7913. Standard test method for bond strength of fiber-reinforced polymer
matrix composite bars to concrete by pullout testing. Conshohocken, USA: American
narrower crack widths). Beams with the same Ef Af reinforced with
Society for Testing and Materials; 2014. p. 9.
the same type of GFRP bars are expected to exhibit similar behavior. [7] Bischoff PH, Gross S. Design approach for calculating deflection of FRP reinforced
The difference in surface profiles of the GFRP bars in beams with the concrete. J Compos Constr, ASCE 2011;15(4):490–9.
same Ef Af, however, affects the strains and crack widths, but it does [8] Bischoff PH, Gross S. Equivalent moment of inertia based on integration of curva-
ture. J Compos Constr, ASCE 2011;15(3):263–73.
not significantly affect deflection and load-carrying capacity. [9] Canadian Standard Association (CSA). Design and construction of building com-
3. When the mechanical properties and surface profile of GFRP bars ponents with fibre reinforced polymers. Rexdale, ON: CAN/CSA S806-12; 2012. p.
are maintained, using smaller diameters enhances the strain and 187.
[10] Canadian Standard Association (CSA). Canadian highway bridge design code.
cracking behavior. Smaller GFRP bars exhibit smaller stains and Rexdale, ON, Canada: CAN/CSA S6-14; 2014. p. 733.
narrower crack widths than larger bars due to limiting bar spacing, [11] Canadian Standards Association (CSA). Specification for fibre-reinforced polymers.
which controls crack propagation. Our results, however, did not Rexdale, Ontario, Canada: CAN/CSA S807-10; 2010. p. 27.
[12] El-Nemr A, Ahmed E, Barris C, Benmokrane B. Bond-dependent coefficient of glass-
confirm a clear trend between GFRP-bar diameter and the service- and carbon-FRP bars in normal- and high-strength concretes. Constr Build Mater
ability behavior of GFRP-RC beams. 2016;113:77–89.
4. The cracking behavior of the tested beams tends to confirm that [13] El-Nemr A, Ahmed E, Benmokrane B. Flexural behavior and serviceability of
normal- and high-strength concrete beams reinforced with glass fiber reinforced
sand-coated GFRP bars provide better bond performance than he-
polymer bars. ACI Struct J 2013;110(6):1077–88.
lically-grooved GFRP bars because of the uniform surface and the

618
A. El-Nemr et al. Engineering Structures 173 (2018) 606–619

[14] Eurocode 2: EN1992-1-1:2004. Design of concrete structures. European Committee Constr, ASCE 2011;15(5):682–95.
for Standardization. [20] Mota C, Alminar S, Svecova D. Critical review of deflection formulas for FRP-RC
[15] Gangarao VS, Vijay PV Design of Concrete Members Reinforced with GFRP bars. In: members. J Compos Constr, ASCE 2006;3(10):183–94.
Proceedings of the third international symposium on non-metallic (FRP) re- [21] Mousavi SR, Esfahani MR. Effective moment of inertia prediction of FRP-reinforced
inforcement for concrete structures, Sapporo, Japan. vol. 1; 1997. p. 143–149. concrete beams based on experimental results. J Compos Constr, ASCE
[16] Gulbrandsen P. Reliability analysis of the flexural capacity of fiber reinforced 2012;16(5):490–8.
polymer bars in concrete beams [M.Sc. Thesis]. USA: University of Minnesota; [22] Vijay PV, GangaRao HVS. Unified limit state approach using deformability factors
2005. p. 80. in concrete beams reinforced with GFRP bars. In: Proceedings of the Fourth
[17] ISIS Manual No. 3. Reinforced concrete structures with fibre-reinforced polymers. Materials Engineering Conference, ASCE, Washington, D.C., Nov. 10–14; 1996. p.
ISIS Canada Research Network. Winnipeg, MB: University of Manitoba; 2007. p. 657–65.
151. [23] Vijay PV, GangaRao HVS. Bending behavior and deformability of glass fiber-re-
[18] Jakubovskis R, Kaklauskas G, Gribniak V, Weber A, Juknys M. Serviceability ana- inforced polymer reinforced concrete members. ACI Struct J 2001;98(6):834–42.
lysis of concrete beams with different arrangement of GFRP bars in the tensile zone. [24] Xue W, Peng F, Zheng Q. Design equations for flexural capacity of concrete beams
J Compos Constr, ASCE 2014;18(5):1–10. reinforced with glass fiber-reinforced polymer bars. J Compos Constr, ASCE
[19] Kassem C, Farghaly AS, Benmokrane B. Evaluation of flexural behavior and servi- 2015;18(5):1–10.
ceability performance of concrete beams reinforced with FRP bars. J Compos

619

You might also like