You are on page 1of 16

Engineering Structures 176 (2018) 785–800

Contents lists available at ScienceDirect

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

Reliability analysis of shear strength provisions for FRP-reinforced concrete T


beams
Kourosh Nasrollahzadeh , Reza Aghamohammadi

Department of Civil Engineering, K. N. Toosi University of Technology, P.O. Box 15875-4416, Tehran, Iran

ARTICLE INFO ABSTRACT

Keywords: This paper is aimed at conducting a reliability based assessment on shear strength provisions of the existing
Fiber Reinforced Polymer (FRP) bar design codes for fiber-reinforced polymer reinforced concrete (FRP-RC) beams. Herein, shear design equations of
Concrete beam the latest versions of ACI 440.1R and CSA S806, which are two of the most widely used codes for FRP-RC beams,
Shear strength are studied. The shear strength formula of ACI 440.1R is different from that of CSA S806 in that how they
Reliability
incorporate various parameters. It is, therefore, of interest to evaluate the reliability levels offered by these
FRP-RC beam
design equations. In doing so, the uncertainty in mechanical and geometrical variables, the error inherent in
shear resistance models, and the probabilistic characteristics of various types of loads including dead, live, snow,
wind, and earthquake under different load combinations are considered. The shear model errors have been
assessed using a large database of shear-failed test results, which was assembled from literature and contained
288 FRP-RC beams without stirrup and 109 beams with FRP stirrups. It was revealed that the safety level for
shear capacity of beams without stirrup is different from that for beams with stirrup in ACI, and to a lesser
extent, in CSA. Moreover, the reliability indexes of shear strength of FRP-RC beams in ACI differ from those in
CSA. In order to meet the target reliability index, the modified shear design formulae and the corresponding
safety factors are proposed. Besides, the shear compression capacity, which proves to play a key role in the
reliability index of shear strength of beams with FRP stirrups, is revised for different reliability levels.

1. Introduction other design codes. To improve accuracy of the shear capacity relations
for FRP-RC beams, some researchers have used soft computing para-
Application of fiber reinforced polymer (FRP) bars instead of steel digms such as neural network, genetic algorithm, fuzzy inference
bars for internal reinforcement of concrete beams provides an attractive system, etc. so as to predict the existing test data [3,5,9–11]. It is no-
solution to overcome corrosion problems in buildings and bridges. ticed that in the context of such supervised learning approaches, there
Among different design aspects of FRP-reinforced concrete (FRP-RC) are plenty of possibilities to offer new equations with different formats
beams, shear capacity evaluation has been a subject of research in re- so as to improve the model uncertainty, which is hereafter called as
cent years [1–4]. Researchers have noted that current shear provisions professional factor. However, very few researches have dealt with re-
for FRP-RC beams, that are essentially modified versions of shear liability analysis by accounting for resistance reduction factors and
equations for steel-RC beams, differ much from each other on how they different load combinations used in the design codes. Some applications
incorporate various parameters. Particularly, North American guide- of reliability analysis for FRP-retrofitted beams have been reported by
lines namely ACI 440.1R and CSA S806, hereafter abbreviated to ACI Pham and Al-Mahaidi [12] and Zeng et al. [13]. The other works on
and CSA, respectively, propose totally different equations for shear reliability of design for FRP-RC beams are related to flexural capacity as
strength contributed by concrete and FRP stirrups. Assessment of the reported by He and Qiu [14], Ribeiro and Diniz [15], and Shield et al.
shear provisions of ACI against available experimental data demon- [16].
strated that the existing equations for shear capacity of FRP-RC beams As for safety assessment of shear provisions of FRP-RC beams, only
are either highly conservative or even inadequate in some cases [5–7]. two works by Zadeh and Nanni [6] and Shahnewaz et al. [17] have
Razaqpur and Spadea [8] have provided comprehensive details on employed reliability analysis by taking into account the probabilistic
underlying concepts of CSA shear provisions and evaluated the shear nature of load and resistance. Zadeh and Nanni [6] investigated both
strength equations of CSA against the available experimental data and flexural and shear strength provisions of ACI 440.1R-06 [18] through a


Corresponding author.
E-mail address: nasrollahzadeh@kntu.ac.ir (K. Nasrollahzadeh).

https://doi.org/10.1016/j.engstruct.2018.09.016
Received 12 April 2018; Received in revised form 7 September 2018; Accepted 9 September 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

Nomenclature MF factored moment caused by loads (kN·m)


nf ratio of modulus of elasticity of FRP bars to modulus of
Af area of longitudinal FRP reinforcement (mm2) elasticity of concrete
Afv amount of FRP shear reinforcement within spacing s P professional factor
(mm2) Pequ professional factor for equivalent normal distribution;
Afv,min minimum amount of FRP shear reinforcement within S snow load (kN)
spacing s (mm2) s stirrup spacing (mm)
a/d ratio of shear span to effective depth Vc unfactored shear strength provided by concrete (kN)
bw width of the web (mm) VD unfactored load caused by dead load (kN)
D dead load (kN) VF factored shear force caused by loads (kN)
d distance from extreme compression fiber to centroid of Vf unfactored shear resistance provided by FRP stirrup (kN)
tension reinforcement (mm) VL unfactored load caused by live load (kN)
dv effective shear depth (0.9d) (mm) Vexp experimental shear strength (kN)
E earthquake load (kN) Vload unfactored shear force (kN)
Ec modulus of elasticity of concrete (GPa) Vpred predicted shear strength (kN)
Ef modulus of elasticity of longitudinal FRP bar (GPa) Vtotal total factored shear resistance provided by concrete and
Es modulus of elasticity of steel (GPa) stirrup (kN)
Efv modulus of elasticity of FRP stirrup (GPa) Vsc,ACI ACI shear-compression limit
f’c compressive strength of concrete (MPa) Vsc,CSA CSA shear-compression limit
ffb strength of bent portion of FRP bar (MPa) W wind load (kN)
ffu tensile strength of longitudinal FRP bar (MPa) β reliability index
ffv effective tensile strength of FRP stirrup (MPa) βT target reliability index
ffu,v tensile strength of FRP stirrup (MPa) δ coefficient of variation
G limit state function (LSF) θ angle of the diagonal compressive stress (degree)
k ratio of depth of neutral axis to reinforcement depth λ bias (mean to nominal value)
kc coefficient of calibration for modified CSA provisions μ mean
km coefficient taking into account the effect of moment at ρf longitudinal reinforcement ratio (Af/bw d)
section on shear strength ρfv shear reinforcement ratio (Afv/bw s)
kr coefficient taking into account the effect of reinforcement ϕ strength reduction factor in ACI
rigidity on shear strength ϕc partial reduction factor for concrete in CSA
ks coefficient taking into account the effect of member size ϕf partial reduction factor for FRP reinforcement in CSA
on shear strength σ standard deviation
L live load (kN)

proposed approach called as comparative reliability, that is based on beams without stirrups and 46 specimens with FRP stirrups. As a result,
the first order second moment (FOSM) method with lognormal prob- the finally proposed equations were totally different from those ap-
abilistic distribution for resistance and load parameters, and is aimed at peared in the current codes. In the second part of the paper, the relia-
maintaining the target reliability index βT for FRP-RC beams the same bility analysis was included by employing the reliability index to cali-
as that for steel-RC beams. As for shear capacity, the professional factor brate the safety level of the modified equations by recommending
was determined using a test database including 196 FRP-RC beams resistance reduction factors to meet the target reliability index of 3.5.
without stirrups as well as 118 beams with FRP stirrups, of which 27 The only load combination considered was dead and live loads. The
beams had longitudinal steel bars. Zadeh and Nanni [6] concluded that reliability analysis was based on FOSM with the normal probabilistic
a resistance reduction factor of ϕ = 0.84 for βT = 3.5, and ϕ = 0.76 for distribution for parameters of resistance and load. Since the focus of
βT = 4.5 would meet the desirable reliability for the shear strength of their paper was on the optimized versions of design provisions, the
beams without stirrup. For FRP-RC beams with stirrups, it was sug- reliability indixes were determined for only the modified equations, and
gested to limit the stirrup capacity to three times the concrete strength, thus, there was no need to provide reliability indexes for the original
and thus ϕ = 0.77 for βT = 3.5 and ϕ = 0.7 for βT = 4.5 were derived. equations existing in the current codes.
Furthermore, the first order reliability method (FORM) was conducted In the present study, the shear strength provisions of the latest
to validate the proposed factor of strength reduction against two load versions of both ACI 440-2015 [20] and CSA S806-2012 [21] are
combinations containing live load in addition to wind load. assessed and compared using reliability methods including FORM and
Another attempt towards reliability analysis of shear provisions of FOSM under variety of load combinations containing dead, live,
FRP-RC beams was carried out by Shahnewaz et al. [17], whose paper wind, snow and earthquake loads. The professional factor for each
contains two parts: First, the shear design equations existing in the provision is determined by a large database of shear-failed test results
current codes (i.e., ACI 440-2006 [18] and CSA S806-2002 [19]) were on 288 FRP-RC beams without stirrups and 109 beams with FRP bars
totally altered by changing both the constant coefficients and the power as longitudinal and transverse reinforcement. This test database,
values of different terms in the current equations. Besides, Shahnewaz which is considered to be the largest one available so far, was as-
et al. [17] suggested that the concrete compressive strength should be sembled by an extensive survey of open literature. Safety assessment
incorporated in the equation of FRP stirrup shear capacity to enhance for beams with FRP stirrups is conducted by taking into account
the accuracy although it was initially absent in the original code different modes of shear failure, namely shear tension and shear
equations. The objective was to improve the professional factor (P) in compression, which are hereafter abbreviated to ST and SC, respec-
order to reach an average of unity and the least coefficient of variation tively. The reliability indexes for the current shear provisions are
through an optimization procedure namely genetic algorithm (GA) by calculated and modifications needed to capture the target reliability
using the available experimental database divided equally into training index are recommended.
and validation sets. The experimental database included 116 FRP-RC

786
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

2. Research significance 3. Details of reliability assessment procedure

There are several aspects which contribute to the novelty of the In order that the shear strength provisions of ACI [20] and CSA
current paper and introduce definite differences with the similar works [21], which are presented in Table 1, to be evaluated through a relia-
in the literature such as [17]. As a part of this research is to determine bility analysis, the following details are addressed.
the reliability indexes for the shear design provisions, as they exist in
the latest codes of ACI and CSA, and before any modification is made. 3.1. Statistical characteristics of random variables
Then, the existing design equations are modified so as to meet the
target reliability index. In doing so, the present paper is aimed at The variables which are involved in the code provisions for eva-
maintaining the format (i.e., the list of the variables involved and their luation of the shear strength of FRP-RC beams are treated as random
power values) of the existing design equations, and then, any mod- parameters that incorporate uncertainty within both resistance as well
ification is made exclusively to the constant coefficients behind each as load. The variables along with their statistical descriptors are listed
equation. The rationale behind this approach is that any change in in Table 2. The resistance related variables themselves are either fab-
powers of the equation parameters can be adopted only if the me- rication based or material based uncertainties. The former include the
chanics behind each design equation in ACI or CSA is justified in-depth beam width, bw, the effective depth of beam, d, the spacing of FRP
because the design equations in international codes, though being semi- stirrups, s, the cross-sectional area of longitudinal and transverse FRP
empirical to some extent, are essentially based on the mechanics of the reinforcement, Af or Afv. The latter contain the compressive strength of
problem in hand, and thus, altering their parameters and powers needs concrete, f’c, the tensile strength of longitudinal or transverse FRP bars,
to be done cautiously in order not that the mechanics behind the ffu or ffu,v , the elastic modulus of longitudinal or transverse FRP bars, Ef
equations is lost. Capturing underlying mechanics is, however, far be- or Efv. Due to high correlation between compressive strength of con-
yond the minimization of errors between experiment and prediction, crete and elastic modulus of that, Ec, the equation provided by the ACI
and instead, it requires development of accurate mechanical models as design guideline (Ec = 4700 f 'c ) is used to calculate Ec [30]. The sta-
substitutes for the current code equations, which is indeed beyond the tistical characteristics of the random variables are adopted from several
scope of this paper. references in the literature, as appeared in Table 2.
Furthermore, other constraints in each code including the need to Another source of uncertainty is the inherent error associated with
have one single reduction factor for both beams without and with predictive models, what is called as professional factor or model un-
stirrups in ACI, or the requirement to maintain the predefined values of certainty. The professional factor, P, is calculated as the ratio of the
partial safety factors of concrete and FRP in CSA are also taken into experimentally obtained resistance to the nominal resistance computed
account in the current paper. This feature is important since it reflects by the predictive equation (i.e., the provisions of design code). To
the substantial difference between ACI and CSA as the latter employs conclude the statistical parameters for the professional factor, a large
material strength reduction factors. As for the reliability analysis of the database of test results on FRP-RC beams failed in shear mode is col-
shear strength equations of CSA, although the version of 2002 has been lected through an extensive survey of the open literature. In total, 288
studied in the past [17], the version of 2012, which has major changes specimens for FRP-RC beams without stirrups [4,8,26–28,31–62] and
compared to the previous version, has not yet been subject of the re- 109 samples for FRP-RC beams with FRP stirrups
liability assessment. [8,36,37,39,48,50,63–70] are compiled. All the tested beams were
The current paper takes advantage of two reliability techniques prismatic, simply supported and three- or four-point loaded with the
including FORM and FOSM. Besides, FOSM is applied in two manners, shear span-to-depth ratio of greater than 2.5. The concrete beams with
one is based on normal probabilistic distributions for load and re- FRP stirrups but with steel longitudinal reinforcement were omitted
sistance models and the other is based on lognormal distributions for from the database. Table 3 summarizes the statistics of the experi-
those. The former is abbreviated as FOSMN and the latter is FOSML. It is mental database. Careful attention is paid to differentiate the failure
noted that the professional factor P is better defined by a lognormal modes of FRP-RC beams with stirrups into either shear-tension or shear-
distribution due to its non-negative nature rather than a normal type. compression mode. In the literature, 32 specimens were reported to fail
Meantime, it is recalled that FORM is independent from the format of in shear tension while 19 beams failed in shear compression. For the
the limit state function. Another novelty of the present work is related other tested beams (i.e., 58 out of 109 specimens), the failure mode was
to the emphasis made on differentiation between two modes of shear not reported in the original papers. Table 4 presents the statistical
failure which are shear tension versus shear compression (for short, ST properties of the available experimental data with reported failure
and SC, respectively). Herein, this is included directly in the reliability modes.
analysis through the SC limit recommended by the existing design Table 5 lists the factored load combinations used in this study for
codes despite it was mostly disregarded by the others [6,17]. each of the design guidelines of ACI and CSA. A factored load

Table 1
Shear strength provisions of ACI 440.1R and CSA S806 design codes.
ACI 440.1R-15 [20] CSA S806-12 [21]

1
Vc = 0.4 f c' b w (kd) Vc = 0.11 f c' b w dv 0.05km kr (f c' ) 3 b w dv 0.22 fc' bw dv
Ef VF d 1 750
k= 2 f nf + ( f nf )2 f nf , nf = , Ec = 4700 f c' dv = 0.9d , km = 1.0 , kr = 1 + (Ef f )3 , ks = 1.0
Ec MF 450 + d
Afv Afv
Vf = f d Vf = f d cot( )
s fv s fv v
f fv = 0.004Efv f fb f fv = min(0.4f fu, v , 0.005Efv, 1200)

VSC , ACI =
2
fc' b w d VSC , CSA = 0.22f c' b w dv
3
Vtotal = Vc + (Vf VSC , ACI ) Vtotal = (Vc + Vf ) VSC ,CSA
= 0.75 c = 0.65 , f = 0.75
0.35
Afv, min = b s 0.07 f c'
f fv w Afv, min = bw s
f fv

787
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

Table 3

Mirza and MacGregor [22]

Nowak and Szerszen [24]


Statistical properties of experimental database for FRP-RC beams.

Nowak and Collins [29]


Michaluk et al. [26]
Parameter Without Stirrup (288) With Stirrup (109)

El-Sayed et al. [27]

El-Sayed et al. [28]


Zadeh et al. [23]
Min. Max. Median Min. Max. Median
References

bw (mm) 89.000 1000.000 200.000 150.000 457.000 200.000


d (mm) 73.000 889.000 218.173 170.000 937.000 252.809
s (mm) – – – 24.000 406.000 125.000

ρf (%) 0.105 3.980 0.930 0.510 3.980 1.900


100–1000
150–1000

ρfv (%) – – – 0.040 1.500 0.350


ρfv bw da
40–500

2.5–6.5
ρf bw da

a/d 2.500 12.500 4.000 2.500 5.340 3.200


20–60
Range

f'c (MPa) 19.200 93.100 36.300 20.000 84.200 34.300







Ef (GPa) 32.000 148.000 48.200 29.400 140.000 60.797
Efv (GPa) – – 30.000 144.000 60.500

Extreme Value Type II


Extreme Value Type I

Extreme Value Type I


Vexp (kN) 8.800 263.000 33.271 20.450 590.000 148.000
(Normal) [18]
Deterministic

Normal [25]
Distribution

combination includes dead load as a permanent load plus a principal


Lognormal
Lognormal

Lognormal
load (e.g. live, snow, wind or earthquake) that dominates the load
Normal
Normal
Normal

Normal

Normal

combination. Should other companion loads exist in a load combina-


tion, such additional loads are set equal to zero herein for the sake of
simplicity and easier illustration of results in two-dimensional curves.
COV, δ

0.003
0.003

0.101

0.100
0.180
0.260
0.370
0.560

The nominal value for each load type is determined by assuming a value


for the ratio between the two loads in a load combination. Given the
nominal value for each load type, its mean and standard deviation can
(1.250)

(1.000)

(5.000)
12.700
13.462

19.400

19.000

31.000
4.826
SD, σ

be determined by use of information provided in Table 2. It is also








noted from Table 2 that the probabilistic distribution function may be


Normal, Lognormal, Extreme value type 1 (or Gumbel), or Extreme
3.0649

value type 2, depending on the load type.


1 (f ' )+
c

3.2. Design space


1.3543 × 10

The design space, which contains the input parameters for hy-
pothetical beams, is intended to take into account all possible design
scenarios [73]. In the present work, the design space is generated by
3 (f ' ) 2

taking at least three values including minimum, median, and maximum


c

for each of the input parameters: bw, d, s, a/d, f’c, ρf, and ρfv, of which
3.1743 × 10

the ranges are given in Table 2. Similarly, for each load combination,
three values are assumed for the ratio of the principal load type to the
total load. For instance, for the basic combination of dead and live
5 (f ' )3+

loads, the ratio of live load to the total load (i.e., L/D + L) is most likely
c

to fall in the range of 0.3–0.7, as suggested by Szerszen and Nowak


2.413 × 10

[74]. Thereby, at least three values of 0.3, 0.5 and 0.7 are selected. As
Bias, λ

for ffu and ffv (or Ef and Efv), the three values mentioned in Table 2 are
1.050
1.000
0.820
0.780
0.660

considered. Finally, by accounting for all possible combinations of the


1
1


given values for the input parameters, the design space is generated.
bw + 2.286

The hypothetical beams not satisfying the design code requirements


(128.000)
1536.000
d - 4.826

(41.300)

(42.000)
692.000

754.000
Mean, μ

(e.g. minimum amount of longitudinal or transverse reinforcement) are


however omitted from the design space. The number of remaining







s

beams for each of the resistance equations is assured to be at least 3000


Nominal Value

beams, which are later employed to calculate the reliability index, β.


Probabilistic characteristics of different variables.

3.3. Limit state function


μ − 3σ
a/d
Afv

(μ)
bw

W
Af

f'c

E
d

S
L
s

The limit sate function G is herein expressed as the difference be-


are 0.003–3 (%).
Earthquake (kN)

tween the resistance and load by the following equation:


(Ef or Efv) (GPa)
ffu or ffuv (MPa)
Name (Unit)

(1)
Snow (kN)

G = PVpre VLoad
Wind (kN)
Dead (kN)
Afv (mm2)

Live (kN)
Af (mm2)

f'c (MPa)
bw (mm)
d (mm)
s (mm)

where Vpre is the equation used to predict the shear strength of FRP-RC
Maximum 50 years load.
a/d

beam, P is the professional factor or model uncertainty that is related to


fv

the uncertainty existing in Vpre, and Vload is the shear due to the total
Source of uncertainty

and

unfactored load acting on the beam. In this study, various equations are
examined for Vpred according to the different states. For instance, in the
Range off

case of shear strength provision of ACI for beams without stirrup, Vpre in
Fabrication

Eq. (1) is replaced by the equation of Vc (see Table 1). Also, VLoad under
Material
Table 2

Loadb

combination of dead and live loads is equal to VD + VL, where VD and


b
a

VL are shear due to unfactored dead and live load, respectively. The

788
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

Table 4
Statistics of experimental database for beams failed in shear-tension and shear-compression modes.
Parameter With Stirrup – Shear Tension (32) With Stirrup – Shear Compression (19)

Min. Max. Median Min. Max. Median

bw (mm) 150.000 450.000 250.000 150.000 250.000 250.000


d (mm) 230.000 937.000 252.809 250.000 252.809 252.809
s (mm) 24.000 400.000 80.000 24.000 90.000 35.000
ρf (%) 0.510 2.300 1.900 1.520 3.024 1.900
ρfv (%) 0.094 1.500 0.500 0.419 1.500 1.000
a/d 2.688 5.340 3.243 2.967 5.340 4.153
f'c (MPa) 24.100 46.000 33.439 22.554 39.518 32.850
Ef (GPa) 29.400 60.797 56.286 56.286 105.000 56.286
Efv (GPa) 31.400 112.769 46.280 39.000 112.769 60.797
ρfv Efv / Es (%) 0.0191 0.846 0.186 0.082 0.846 0.331
Vexp (kN) 48.975 500.000 202.494 73.300 292.219 206.907

Table 5 (iHLRF), which was proposed by Der Kiureghian [75] as an improved


Various load combinations in ACI 440.1R and CSA S806 codes. version of HLRF [76], is employed in the present study. As for the FOSM
# ACI 440.1R-15a CSA S806-12b approach, Taylor approximation is sufficient to estimate the mean and
standard deviation without the need to choose a particular type of
1 1.2D + 1.6L 1.25D + 1.5L probability distribution. This approach has been taken in the current
2 1.2D + 1.6S 1.25D + 1.5S paper too, meaning that the statistical descriptors are determined by
3 1.2D + 1.6 W 1.25D + 1.4 W
4 1.2D + 1.0E 1.0D + 1.0E
Taylor approximation while it does not matter if the distribution is
Normal or Lognormal. Then, in the FOSM method, β is determined by
a
ACI 318-14 [71]. dividing the mean μ of limit state function, i.e., G in Eq. (1), to its
b
CSA A23.3-04 (Reaffirm 2010) [72]. standard deviation σ. However, this method of calculating β is exact
only if the load and resistance (denoted by Q and R, respectively) are
nominal values of VD and VL for a hypothetical beam with an assumed normal distributions. In the case of lognormal distributions for R and Q
load ratio under the corresponding factored load combination are de- (in Eq. (1), there is an error in determining β by dividing μ to σ. On the
termined taking into account the considered resistance reduction factor other hand, R (and P) and Q are better matched with a lognormal rather
as well as the value of Vpred evaluated using the nominal values for than normal distributions due to the fact that their values are always
fabrication and material properties of the hypothetical beam. Once positive. To have an accurate estimate of β in case of lognormal, the
nominal values for all parameters are determined, the mean, standard limit state function is formed by employing logarithm of load and re-
deviation and probabilistic distribution function are readily derived sistance (i.e., lnR and lnQ) instead of R and Q themselves. As a result,
from Table 2. As for professional factor P, the statistical characteristics the limit state function becomes normal although R and Q are log-
are obtained using experimental database, as explained earlier. With normal. Consequently, β is determined based on lnQ and lnR, in which
this information, the reliability index β for the hypothetical beam can the formula for β becomes different from that of the normal distribution
now be calculated following the FORM or FOSM algorithm. Finally, the case. Thereby, the differentiation of distributions appears in the stage of
average value for reliability indexes determined for all beams in the determination of β whereas it has no influence on Taylor approximation
design space is reported. of μ and σ. To address this issue, the FOSM results in this study are
presented for two cases for the sake of comparison, as follows: Normal
3.4. Methods of reliability analysis distribution function (denoted as FOSMN) versus Lognormal distribu-
tion function (denoted as FOSML).
FORM (first order reliability method) and FOSM (first order second Generally, application of FOSM is simpler than FORM but it is less
moment) are the well-known techniques to calculate the reliability accurate in case of some nonlinear functions for limit state. This is
index, β. Among different algorithms available for implementing because FOSM linearizes the function G in the mean point. Thus, by
FORM, the method of improved Hassofer-Lind-Rackwitz-Fiessler altering the form of limit state function for a particular problem, the

Table 6
Calculated reliability indexes for the current provisions of shear strength in ACI 440.1R and CSA S806.
State Vpre Description Number of Vexp Pequ β
P=
Number specimens Vpre
FORM FOSML FOSMN

μ δ μ δ μ δ μ δ μ δ

1 Vc Vc and Vf are based on the current ACI 288 1.786 0.207 1.669 0.153 4.988 0.011 5.153 0.011 4.210 0.013
2 Vc + (Vf Vsc, ACI ) code. 109 1.564 0.311 1.359 0.203 3.106 0.026 3.087 0.043 2.667 0.052
3 Vc + Vf 32 1.351 0.352 1.135 0.219 2.487 0.031 2.115 0.042 1.883 0.064
4 Vc + Vsc, ACI 19 0.825 0.181 0.783 0.139 2.181 0.042 2.157 0.071 2.093 0.083

5 Vc Vc and Vf are based on the current CSA 288 1.093 0.188 1.032 0.143 3.832 0.021 4.869 0.045 3.439 0.024
6 (Vc + Vf ) Vsc, CSA code. 109 1.390 0.270 1.246 0.185 3.373 0.023 4.818 0.042 2.663 0.053
7 Vc + Vf 32 1.401 0.272 1.254 0.186 3.108 0.023 5.637 0.041 2.473 0.052
8 Vsc, CSA 19 0.600 0.215 0.558 0.158 1.394 0.098 1.355 0.112 1.242 0.117

Note: For ACI, the above results are presented under load combination of 1.2D + 1.6L and ϕ = 0.75; and for CSA, the calculations are for 1.25D + 1.5L and ϕc = 0.65
and ϕf = 0.75.

789
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

reliability index β may change consequently in FOSM method whereas β investigation on this issue is left to the future works. It is emphasized
by FORM does not have this problem. Despite the mentioned problem, that selection of target reliability index is somehow based on en-
FOSM has been used by several researchers. Thus for the sake of gineering judgment, and thus, there exists no general agreement on the
comparison, FOSM is included in this paper in addition to FORM al- value of that. To this end and for the sake of comparison, different
though calibration of the code equations is made on the basis of FORM target reliability indexes (i.e., 3.5, 4.0 and 4.5) are addressed in the
only. current paper.

4. Selection of target reliability index 5. Results and discussion

The acceptable probability of failure is expressed by a target relia- 5.1. Evaluation of safety level in the current design codes: General
bility index, βT. Generally, determination of the proper value for the considerations
target reliability index requires sophisticated justifications including
assessment on consequence of failure and incremental cost of safety The shear strength equations provided by ACI 440.1R [20] and CSA
[74], which are beyond the scope of the present work. Besides, judg- S806 [21] along with the corresponding calculated reliability indexes β
ment on the suitable βT needs considerable statistical data, which are are listed in Table 6. The equations in Table 6 include three parts: shear
not available in the case of FRP-RC beams. This paper has no intention strength by concrete itself Vc, shear strength due to FRP stirrups Vf, and
to set any specific value for the target reliability index on shear design SC limit Vsc, each presented for both design codes of ACI and CSA (also
of FRP-RC beams. But instead, the paper presents modified equations see Table 1). The experimental data used to determine the professional
for all three different levels of βT , namely 3.5, 4 and 4.5 so that given factor P are in two groups: 288 FRP-RC beams without stirrups and 109
any target reliability index, the proper equation and the corresponding specimens with FRP stirrups. Besides, the test data for FRP-RC beams
safety factor can be extracted from the figures and tables provided in with FRP stirrups are divided into three categories: beams failed in ST
the current paper. It is noted that the range of βT = 3.5–4.5 covers the mode (32 specimens), beams failed in SC mode (19 specimens), and
usual target reliabilities used by other codes. For example, the target beams with no reported failure mode (58 specimens).
reliability indexes of 3.8 and 4.3 are recommended by Eurocode and For each of the design code provisions (denoted by a state number in
ISO 2394, respectively. Meantime, the reference period of 50 years is Table 6), and for a test specimen, the professional factor P is determined
considered for evaluating target reliability indexes. On the other hand, as the ratio of Vexp to Vpred, where Vexp is the shear strength obtained
the shear design equations for FRP-RC beams without and with stirrups from experiment, and Vpred is the shear strength predicted by the code
existing in the current codes of ACI and CSA yield various values of equation under study for the considered test specimen. Then, the mean
reliability index ranging from 3.1 up to about 5 (see Table 6), making μ and coefficient of variation δ of professional factor P are calculated
difficult to figure out which target reliability was actually sought by the over the test database relevant to each code equation, as presented in
codes. Table 6. Accordingly, the determined μ and δ are suitable to be used in
In the literature, the target reliability index is judged based on the FOSML where the load and the resistance, which also contains P ac-
existing βT for steel-RC beams while taking into account the differences cording to Eq. (1), are assumed to be lognormal. This is because the
between steel and FRP in terms of material behavior, the amount of professional factor P is better presented with a lognormal distribution
ductility provided by failure modes in the case of FRP-RC beams, and function rather than a normal one since it is essentially a positive
the conservatism required to compensate the insufficient database. In variable. When dealing with FOSMN, which assumes normal distribu-
this regard, it was recommended to increase βT by the amount of 0.5 for tion function for all variables, the use of the same values of μ and δ may
members failing suddenly and losing their capacity [30]. In [74], introduce some errors in calculation of β as those are essentially proper
βT = 3.5 was taken for tension mode of flexural failure in steel-RC for P with a lognormal distribution. To avoid this error, as suggested by
beams. As for compression mode of flexural failure in steel-RC beams, some researchers such as Nowak and Collins [29], the concept of an
which is brittle compared with the tension mode, βT is expected to be a “equivalent normal distribution” is employed herein. This means that P
higher value. It is natural to think that compression-flexural mode in is taken lognormal, as it is, with the experimentally determined μ and δ.
FRP-RC beams is similar to compression-flexural mode in steel-RC Then, new values for μ and δ corresponding to an unknown normal
beams because in both states, the concrete crushing governs the beha- distribution are sought in order that the probability density function
vior. In the case of tension-flexural failure of FRP-RC beams, it is noted (PDF) at 5% probability and cumulative density function (CDF) up to
that the behavior is brittle contrary to the tensile-flexural failure of 5% probability of the normal distribution become equal to those of the
steel-RC beams because in the former, rupture of FRP bar occurs in a known lognormal distribution of P with experimentally determined μ
brittle manner whereas in the latter, steel bar yields in a ductile state. and δ. Then, this so-called equivalent normal distribution is employed
Tensile rupture of FRP bars is considered to be more brittle than even in determining β in FOSMN. One more note is that 5% probability is
concrete crushing [20]. However, experimental evidences showed that referred to by and Mirza and MacGregor [22], and is justified on the
FRP-RC beams would develop large displacements prior to tensile- basis that for the resistance models including P, the lower tale (say 5%)
flexural failure [6,16]. Keeping this in mind, shear failure of FRP-RC is of primary concern in evaluating β. In this way, the results of β by
beams is addressed as follows. Firstly, for steel-RC beams, βT for shear FOSMN can become more accurate. The statistical descriptors including
failure should be increased as compared to flexural failure to account μ and δ of the equivalent normal distribution of P (denoted by Pequ) are
for the brittle nature of shear mode. Following the same analogy em- given in Table 6. This information is used solely to determine the re-
ployed in steel-RC beams, the target reliability index for shear failure of liability index in the method of FOSMN. As explained earlier, in this
FRP-RC beams should be larger than that for flexural failure. It is also study, the reliability index β is calculated by three different methods:
noticed that ST failure of FRP-RC beams (i.e., rupture of FRP stirrups) is FORM, FOSMN, and FOSML. If the method is not mentioned explicitly,
more brittle than ST failure of steel-RC beams (i.e., yielding of steel the reliability index refers to the method of FORM, that is more accu-
stirrups). As for SC failure of FRP-RC beams, where concrete crushes rate than FOSM.
before tensile rupture of FRP stirrups, it is recommended to provide In this paper, the reliability analysis is conducted on several states
some conservatism against the limited number of data reported for this corresponding to different shear design equations. For each state, the
particular mode of failure. It is worthy to note that for different modes design space containing 3000 hypothetical beams, as explained earlier,
of shear failure such as shear-compression and shear-tension considered is considered to cover possible ranges of various parameters, and then,
in this study, one common value for target reliability is recommended. the reliability index is determined for each hypothetical beam based on
Nonetheless, no target reliability is suggested in the current paper, and the mentioned limit state, and for each method of reliability analysis.

790
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

Then, the mean and coefficient of variation CoV for the values of β 5.3. Results of reliability analysis for shear strength provisions of ACI 440
corresponding to 3000 beams are calculated and reported as the sta-
tistical descriptors of β for that particular state as presented in Table 6. Table 6 presents the reliability indexes for the shear strength
The mean is taken as a representative of β values for the hypothetical equations of ACI under the basic load combination of 1.2D + 1.6L. The
beams provided that the dispersion of beta (i.e., CoV) is low enough. In ACI strength reduction factor of ϕ = 0.75 is applied in the reliability
order to evaluate the reliability for the cases of ST and SC modes, the analysis of this section in order to assess the level of safety offered by
hypothetical beams in the design space are divided accordingly into ST- the existing design guideline. State numbers 1 and 2 in Table 6 refer to
failed and SC-failed beams based on the shear compression limit under shear strength of FRP-RC beams without stirrup and with FRP stirrup,
consideration. Then, β for each failure mode is calculated by averaging respectively. The professional factors for these cases are evaluated
over the hypothetical beams with the same failure mode. based on the largest experimental database available in the literature,
It is noted that the reliability index is obviously affected by the which contains 288 specimens without stirrup and 109 specimens with
distribution types of the resistance and load models as appeared in stirrup. Based on FORM method, the average reliability indexes of
Table 6, where the reliability indexes calculated by FOSMN and FOSML, about 5 and 3.1 are determined for states 1 and 2, respectively. Re-
that are based on assuming normal and lognormal distributions re- garding the target reliability index of either 3.5 or even 4.5 for shear in
spectively for both load and resistance models, are different from each FRP-RC beams, it is concluded that the current provisions of ACI are
other. However, the ultimate goal is to satisfy a target value for the conservative for FRP-RC beams without stirrup, whereas are unsafe in
reliability index since the expectation on the failure probability is of the case of beams with FRP stirrups. It should be explained that from a
primary concern regardless of the assumptions made on the distribution statistical point of view, when the mean value of professional factor P
types. For instance, for evaluation of reliability index, Eurocode usually for Vc becomes farther from that of Vf, which is the case herein due to
assumes lognormal or Weibull distributions for material and structural conservatism associated to Vc as per ACI provision, the coefficient of
resistance parameters and model uncertainties while uses normal or variation δ of P for (Vc + Vf) may yield a larger value than that of either
extreme value distributions for variable actions. However, according to Vc or Vf alone. This in turn reduces the reliability index as a con-
Eurocode, the recommended minimum values for reliability index at sequence of a high value of δ (see Table 6, state 2). To shed more light,
ultimate limit states and for reference period of 50 years are 3.8 and 4.3 for the FRP-RC beams failed in shear tension mode, the reliability of
for reliability classes RC2 and RC3, respectively. Once again, it is em- shear strength provision of ACI is assessed by use of 32 test data with ST
phasized herein that the current paper is not aimed at proposing a mode (see state 3 in Table 6). In this case, the reliability index of about
particular target reliability index for the problem under study. 2.5 is obtained. This low value of reliability index is partly attributed to
a relatively large amount of δ = 35% for professional factor of (Vc + Vf)
in state 3, that itself is because of underestimation of concrete strength
5.2. Effects of number of test data on evaluation of professional factor or Vc by ACI, as explained earlier. On the other hand, it is noted that the
model uncertainty term of shear strength provided by FRP stirrups as appeared in ACI
(state 2 in Table 6) is limited by the shear compression strength, VSC,ACI.
It is noted that the database which is employed herein is the largest In order to assess the reliability of the SC limit suggested by ACI, state 4
ever used on the problem of shear design of FRPRC beams. And un- (in Table 6) is examined using 19 test data failed in shear compression
avoidably, the number of test data for beams with stirrups is different mode. The mean of professional factor P for this state has a low value of
from that of beams without stirrup, which are herein 288 versus 109 0.825 despite conservatism existing in Vc, showing that the ACI shear
specimens, respectively. The determination of μ and δ for P by finite compression limit is higher than experimental evidences for FRP-RC
number of test data has been already employed by others such as: 34 beams. Considering that the SC limit currently employed in ACI 440 is
data for tension flexural failure versus 91 for compression flexural the same as that in ACI 318 [77], the present results indicate that the
failure of FRP-RC beams [14] or 196 data for beams with stirrups versus ACI SC limit is in need of revision when used for FRP-RC beams. All in
86 test data for beams without stirrups for shear strength of FRP-RC all, the current shear provisions of ACI for estimation of total strength of
beams [6]. FRP-RC beams with FRP stirrups are not safe enough in both SC and ST
In order to examine that how uncertainty from the finite number of modes.
test data may affect the value of the professional factor in each state of According to recommendations by Joint Committee on Structural
Table 6, it is reminded that the mean of a random variable (herein P) is Safety (JCSS) [78], the resistance model uncertainty (i.e., professional
itself treated as a random variable which has its own mean and stan- factor) is modeled as lognormal with a mean of 1.2 and CoV of 0.15 for
dard deviation s, in which the mean of mean is equal to the mean itself concrete members under bending [13]. However, the JCSS is for steel-
whereas s of mean is = n , where n is the number of test data and σ is RC and not dealing with FRP-RC, where steel bars are replaced by FRP
the standard deviation of P. Now, assuming a normal distribution for rods. In the current paper, μ and CoV of P are determined by a large
the mean of P, the probability for mean of P to fall less than 90% of the database of experiments on shear-failed FRP-RC beams. Because the
mean can be readily calculated. This so-called 10% margin is con- statistics of P is calculated in this study on the basis of experiments,
sidered as an acceptable tolerance herein. It is clear that by decreasing there is no longer any need to follow the rough estimates suggested by
n, the amount of s increases and thus the mentioned probability in- the other codes. However, for the sake of comparison with JCSS, it is
creases too. Consequently, decreasing the real value of P from its cal- seen that CoV of P obtained for all states of shear design in this study is
culated mean reduces the reliability index. Taking this approach into larger than CoV for flexural design of steel-RC beams according to JCSS
account, s and the 10% probability are calculated for all states con- (that is 0.15), as expected regarding a higher uncertainty required for
sidered in this study. The results show that the worst case happens for shear failure versus that for flexural failure. It is noted that for dead and
the state 3 with 32 test data (see Table 6) which yields s = 0.084 and live loads, JCSS takes the permanent load as a normal distribution,
the 10%-probability of 5.4%. The 10% probability for state 2 (with 109 whose mean is the same as the specified value with CoV of 0.05,
test data) is about 0.04% and that for state 1 (with 288 test data) is whereas the variable load is taken as a Gumbel distribution (i.e., Gen-
quite negligible (i.e., almost zero). This means that the probability of eralized Extreme Value distribution Type 1) with a mean of 0.6 times
deviation from the determined mean value of the professional factor the specified value and CoV of 0.35.
due to the finite number of test data is negligible for all considered
states in this study.

791
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

5.4. Results of reliability analysis for shear strength provisions of CSA and with βT = 1.39 and 2.18, respectively.
comparison with ACI
5.5. Effect of shear-compression limit on reliability of shear strength
Reliability indexes for shear strength provisions of CSA are de- provisions for FRP-RC beams with stirrup
termined to be equal to 3.83 and 3.37 for FRP-RC beams without and
with stirrups, respectively (see states 5 and 6 in Table 6). The calcu- Based on experimental results, change in failure mode of FRP-RC
lations in Table 6 correspond to the basic load combination of beams from ST into SC is dependent on the shear reinforcement index
1.25D + 1.5L with consideration of the partial safety factors of (i.e., ρfv Efv), as stated in ACI 440 [18]. It is well known that by in-
ϕc = 0.65 for concrete and ϕf = 0.75 for FRP as recommended by CSA. creasing this ratio, the shear strength contribution by FRP stirrups
It is noted that the mean of P for the CSA equation of shear strength theoretically increases but should be limited by the SC strength since
contribution by concrete is about 1.1, which is more accurate than that the failure mode is controlled by crushing of the concrete web rather
by ACI (i.e., 1.78). Besides, the difference existing in the reliability level than rupture of the FRP stirrups [20]. To examine the effect of SC limit
offered by CSA for shear capacity of concrete versus that of concrete on the reliability of shear strength provisions for FRP-RC beams with
plus stirrup indicates that there is no consistent level of safety between FRP stirrups, two cases are studied, as follows. The first case refers to
the two states of without stirrup and with stirrup. Inconsistency in the the design code provision, as it is, which accounts for SC limit. The
reliability of shear strength provisions is even more pronounced in ACI, latter is the same as the first case but without applying any SC limit. The
in which the reliability index for beams without stirrup is much higher professional factors for both cases are evaluated based on the same
than that for beams with stirrup (βT = 5 versus 3.1). number of specimens (i.e., 109 test data) and reported in Table 7. The
Similar to what has been done for ACI, the reliability of shear 109 data contain shear-compression specimens in addition to shear-
strength provisions of CSA for both ST and SC modes is studied as tension ones. The results are presented for both ACI and CSA codes. As
follows. Comparison of professional factors between states 7 and 5 in for ACI provision, it is seen from Table 7 that once SC limit is applied
Table 6 indicates that the shear strength due to FRP stirrups, Vf, as (denoted by state number 2), the reliability index for the ACI shear
provided by CSA is somewhat underestimated since by adding this term equation increases as compared to when SC limit is not considered
to Vc, the mean value of P becomes larger than unity and increases by (state number 9 in Table 7). This is because in the absence of a limit on
about 30%. The fact is that the tensile strain for FRP stirrup as con- the FRP stirrup capacity, the specimens which actually failed in SC are
sidered by CSA is 0.003 on average. It should be noticed that the tensile treated as ST ones, resulting in a predicted value for shear strength
strain used in the previous version of ACI 440 [79] was 0.002, which higher than it must be. On the contrary, when SC limit is incorporated
was justified to be very conservative and was increased to 0.004 in the into the shear strength provision, the model error in prediction of ex-
latest versions. As for SC strength of CSA, the reliability is assessed perimental values decreases totally. This means that the professional
using 19 test data failed in this mode and the index of reliability is factor after introducing SC limit is improved in terms of higher mean
derived to be equal to 1.39 (see state 8 in Table 6). Such a low relia- value with lower dispersion (see Table 7), and consequently, the re-
bility index is expected because the mean value of P is equal to 0.6. This liability index is increased. The above-mentioned conclusion highlights
shows that the SC limit of CSA is much higher than experimental ob- the key role of proposing a proper limit for SC strength in enhancing the
servations – a conclusion which was mentioned for ACI too. In fact, the reliability of shear strength provisions for FRP-RC beams with FRP
equation for SC currently employed in CSA S806 [21] is the same as stirrups.
that used in CSA A23.3-14 [80] for steel-RC beams with only one dif- A study similar to ACI is conducted on the shear regulations of CSA,
ference in which the coefficient of 0.22 is set instead of 0.25. Such a listed as states 6 and 14 in Table 7. Again, it is seen that the calculated
slight reduction of this coefficient is however insufficient. In brief, reliability index by FORM in the case that SC limit is applied becomes
comparison of the reliability indexes for states 4 and 8 in Table 6 in- higher than when this limit is not incorporated into the equation.
dicates that the SC limits suggested by CSA and ACI are not safe enough However, the amount of improvement in reliability index due to

Table 7
Reliability assessment of shear strength equations under different expressions for shear-compression limit for FRP-RC beams with stirrups.
State Number Vpre Description Number of specimens Vexp Pequ
P=
Vpre

FORM FOSML FOSMN

μ δ μ δ μ μ μ

9 Vc + Vf Vc and Vf are based on the current ACI code. 109 1.518 0.363 1.263 0.223 2.648 2.373 2.069
2 Vc + (Vf Vsc, ACI ) 109 1.564 0.311 1.359 0.203 3.106 3.087 2.667
10 Vc + (Vf Vsc1,ACI ) a 109 1.626 0.263 1.465 0.181 3.874 3.839 3.242
11 Vc + (Vf Vsc 2, ACI ) b 109 1.751 0.220 1.624 0.160 4.638 4.642 3.798
12 Vc + Vsc1, ACI a 19 1.243 0.188 1.175 0.143 3.910 3.938 3.512
13 Vc + Vsc 2, ACI b 19 1.502 0.250 1.366 0.175 3.810 3.781 3.341

14 Vc + Vf Vc and Vf are based on the current CSA code. 109 1.389 0.271 1.244 0.185 3.090 5.626 2.468
6 (Vc + Vf ) Vsc, CSA 109 1.390 0.270 1.246 0.185 3.373 4.818 2.663
15 Vc + (Vf Vsc1,CSA) a 109 1.403 0.256 1.270 0.178 3.793 4.501 3.033
16 (Vc + Vf ) Vsc2, CSA b 109 1.441 0.229 1.329 0.165 4.246 5.314 3.555
17 Vc + Vsc1, CSA a 19 0.980 0.212 0.913 0.156 3.204 3.401 2.815
18 Vsc 2, CSA b 19 1.097 0.248 0.999 0.174 3.143 3.856 2.896

Note: For ACI, the above results are presented under load combination of 1.2D + 1.6L and ϕ = 0.75; and for CSA, the calculations are for 1.25D + 1.5L and ϕc = 0.65
and ϕf = 0.75.
a Efv Efv '
f c' bw d and Vsc1, CSA = 0.17
2
Vsc1, ACI = f bw d
3 Es Es c
b
Vsc 2, ACI = 3Vc , in which Vc is based on the current provisions of ACI for concrete shear strength and Vsc 2, CSA = 4Vc , in which Vc is based on the current provisions of
CSA.

792
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

introduction of SC limit is less appreciated in CSA as compared to ACI. 0.97, and for beams with stirrup, ϕ = 0.66 and 0.56 are derived for
This is partly because the contribution of FRP stirrups to shear strength βT = 3.5 and 4.0, respectively.
as provided by CSA is less than that by ACI, as explained earlier. As for CSA, which is based on partial safety factors applied to the
In line with the above-mentioned discussion, Zadeh and Nanni [6] material strength, there are two coefficients as ϕc = 0.65 and ϕf = 0.75
have applied a limit of 3Vc to the calculated value of Vf, and determined for concrete and FRP materials, respectively. In Fig. 1, the variation of β
professional factor using the data in which the stirrup capacity was not for shear strength of beams without FRP stirrups in terms of different
greater than 3Vc, where Vc was based on the current provision of ACI values of ϕc is illustrated. It is noted that the horizontal axis of Fig. 1
440, and disregarded the data not satisfying this requirement. In other may refer to ϕ, ϕc or ϕf depending on the case under study. It can be
words, the specimens that were likely to fail in SC mode were not in- seen that for the CSA equation of beams without stirrup to reach
cluded in their reliability analysis. In here, we take 3Vc as a candidate βT = 3.5, 4.0, or 4.5, it is required to have ϕc = 0.70, 0.62, or 0.55,
for SC limit proposed by [6]. On the other hand, as explained earlier respectively. Thus, the current value of ϕc = 0.65 in CSA can barely
under state 4 in Table 6, the SC limit of ACI 440 overestimates the provide a target reliability index of about 4.0 for beams without stir-
experimental evidences. This issue has also been addressed by Shehata rups. Furthermore, the effects of varying values of ϕf, while the value of
[81], who suggested the SC limit of ACI for FRP-RC beams to be pro- ϕc is kept unchanged, on shear strength of beams with stirrups are
portional to the square root of Efv/Es, where Efv and Es are elastic demonstrated in Fig. 1 by three curves, each corresponds to a certain
modulus for FRP stirrup and steel, respectively. In Table 7, the effects of value of ϕc (i.e., 0.70, 0.62, and 0.55). As mentioned for state 6 in
SC limits proposed by Shehata [81] (denoted by Vsc1) as well as Zadeh Table 6, the current values of ϕc = 0.65 and ϕf = 0.75 in CSA yield a
and Nanni [6] (denoted by Vsc2) on the reliability of shear provisions of target reliability index of about 3.37 for the case of beams with stirrups.
both ACI and CSA codes are examined. As can be seen from states 10 In order to meet βT = 4.5 for FRP-RC beams in both cases of with and
and 11 in Table 7, the reliability index for Vsc2,ACI arrives at about 4.6 without stirrups, it is required to apply ϕc = 0.55 and ϕf = 0.46, si-
and its professional factor is better than that of Vsc1,ACI in terms of lower multaneously, as seen in Fig. 1. Also, for βT = 3.5, it is required to have
coefficient of variation (δ = 22% versus 26%). Meantime, δ for both of ϕc = 0.7 and ϕf = 0.66, whereas ϕc = 0.62 and ϕf = 0.55 are needed to
the newly proposed SC limits is less than that of the current SC limit of achieve βT = 4.0. The major point to note is that the current partial
ACI (see state 2 with δ = 31%). Comparing the mean values of P for safety factors (i.e., ϕc = 0.65 and ϕf = 0.75) are to be kept constant
Vsc,ACI , Vsc1,ACI , and Vsc2,ACI indicates that the earlier the SC limit throughout the CSA code, meaning that change of them affects the re-
mobilizes, the higher the reliability index becomes. States 15 and 16 in liability of not only shear provisions but also all other models in the
Table 7 apply SC limits of Vsc1 and Vsc2 on the shear provision of CSA code. In order to provide the desirable level of safety without altering
code. It is noted that in employing Shehata's approach for SC limit of the pre-defined values of ϕc and ϕf, the solution is to incorporate an-
CSA, denoted as Vsc1,CSA in Table 7, only the portion of FRP stirrup in other coefficient called as member factor or to modify the existing
the expression of SC limit, that is 0.17f'c bw d [8], is multiplied by the coefficients in the resistance equations. This approach has already been
square root of Efv/Es. Similarly, Vsc2,CSA is taken as 4Vc, of which 3Vc is employed in CSA A23.3-94 when modifying a coefficient existing in
the portion of FRP stirrup, and Vc is based on the current equation of shear strength equation for steel-RC beams without stirrup [82]. To
CSA. Similar to ACI, it is concluded that by introducing Vsc2,CSA as the summarize, to maintain a single value of ϕ for beams without and with
SC limit, the reliability index for the current provision of CSA can arrive stirrup in ACI, and also, to keep the present values of ϕc = 0.65 and
at about 4.3. It is emphasized that to achieve this improved reliability, ϕf = 0.75 in CSA; it is required to modify the coefficients existing in the
the only change made to the CSA equations is to apply a new SC limit current shear strength relations of ACI and CSA so as to meet a desirable
(i.e. Vsc2,CSA). reliability level. In this capacity, reliability-based modifications to the
For further assessment on accuracy of the different SC limits con- existing shear provisions of ACI and CSA will be conducted in the up-
sidered herein, the mean values for P are determined and listed as states coming section.
12–13 (for ACI) and 17–18 (for CSA) in Table 7. The professional factor
P is evaluated using 19 test data specifically failed in SC mode. The
mean values of P for all these states are greater or equal to unity as 5.7. Reliability-based modification of shear strength provisions for FRP-RC
opposed to the current SC limits in ACI and CSA codes, which over- beams
estimate the SC capacities (see states 4 and 8 in Table 6). Further ca-
libration of the considered SC limits is, however, needed to achieve the Central to this paper is to work under the framework of the current
desirable reliability index. codes which are internationally known for shear design of FRP-RC
beams. In doing so, three measures need to be taken as follows. First,
5.6. Calibration of strength reduction factors in the current provisions the format of equations including the involved variables and their

As explained earlier, the current shear provisions of ACI and CSA


yield the target reliability indexes in the range of 3.1–5. In order to
arrive at a desirable reliability index, a common solution is that the
strength reduction factors associated with the design equations are
tuned. To this aim, Fig. 1 depicts reliability indexes corresponding to
different values of resistance reduction factors ϕ. As for ACI, two curves
are presented in Fig. 1 for FRP-RC beams without and with stirrup. To
achieve, for instance, βT = 4.5 for beams without stirrup, Fig. 1 shows
that ϕ = 0.85 is required, whereas for beams with stirrup, ϕ = 0.47 is
needed. It goes without saying that the general trend of ACI is to apply a
unified value of ϕ for shear model of FRP-RC beams either without
stirrup or with stirrup. However, it is not possible in this case unless the
characteristic value of the shear strength equation itself is modified so
that just one ϕ can bring both equations (i.e., without and with stirrup)
to the same reliability level, as will be discussed later. For other values
of target reliability index, the corresponding ϕ can be determined using Fig. 1. Variation of β with respect to ϕ, ϕc or ϕf for the current provisions of ACI
Fig. 1. For example, for FRP-RC beams without stirrup, ϕ = 1.1 and 440.1R and CSA S806.

793
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

powers is kept unchanged while the reliability level is adjusted by words, the characteristic value of the shear strength is altered to
modifying the constant coefficient behind the equations and by cali- cancel the model bias associated with Vc. It is noted that the pro-
brating the accompanied resistance factors. It is herein emphasized not fessional factor is computed by use of 288 test data on shear-failed
to alter the general form of the equations so as not to undermine the FRP-RC beams without stirrup – the largest experimental database
mechanics underlying them. This paper takes such a view because even available in the open literature.
though the code equations are semi-empirical, they still follow some (2) In order to arrive at a desirable reliability index, the newly pro-
mechanical backgrounds [8,83]. Although the best way to improve the posed shear strength of concrete (i.e., μP × Vc) is further multiplied
professional factor P is to develop equations based upon sound me- by a coefficient to be calculated based on reliability analysis. This
chanical models with higher accuracy to predict experiments, devel- coefficient in ACI code is interpreted as the new resistance reduc-
opment of such models is however beyond the scope of this paper. On tion factor, which is replaced for the current value of ϕ = 0.75, and
the other hand, tuning the constant terms of the equations has been is kept constant throughout the reliability analysis of newly pro-
already adopted in both ACI and CSA [82,83]. Second, it was revealed posed equations of ACI. In CSA code, in which there is no global
in the previous sections that there is a significant inconsistency between safety factor, the coefficient determined above is considered as an
the reliability index offered by shear strength of concrete alone and that additional constant (say αc) used to tune the characteristic value.
of FRP stirrups in the current provisions of both ACI and CSA. In ad- Therefore, the characteristic value of shear strength of CSA is again
dition, the shear compression strength needs to be consistent with other altered to αc × μP × Vc. By using this approach, the mean value of
shear terms. To solve this, the constant terms in concrete and stirrup the professional factor remains equal to unity in ACI but changes in
shear strength expressions are tuned to arrive at almost the same mean, CSA, yet still close to unity. Needless to say that in reliability
and preferably as close as to unity, for the professional factors of beams analysis of all newly proposed equations of CSA, the partial safety
without and with stirrups. This itself reduces the dispersion of P for factors are kept the same as those existing in the current code (i.e.,
beams with stirrups. Third, as per ACI, there should be a single re- ϕc = 0.65, ϕf = 0.75).
duction factor for both equations of beams without and with stirrups. (3) Now, the shear strength equation for beams with FRP stirrups is
On the other hand, as per CSA which is based on predefined reduction studied. This equation consists of two parts: Vf plus Vc, where Vc is
factors for concrete and FRP material strengths, it is required to keep the newly proposed equation for the characteristic value of shear
ϕc = 0.65 and ϕf = 0.75 unchanged throughout the entire code in- strength contribution of concrete, as derived in above steps. In this
cluding shear strength provisions. case, the statistical characteristics of the professional factor are
Thereby, the proposed approach deals with the above mentioned examined using 32 test data failed in ST mode. The idea here is to
constraints simultaneously, and in this capacity, contributes to the modify the characteristic value of Vf in order that the mean of
novelty of the current work. If the last two constraints are relaxed de- professional factor for shear strength of beams with stirrup becomes
spite it is not an acceptable result, meaning that the constant terms in equal to that of beams without stirrup. This means that model error
each equation are tuned only on the basis of meeting the target relia- for shear strength of stirrups is essentially in the same order of that
bility index (i.e., equivalent to the change of ϕ factor), the results are as for shear strength of concrete.
indicated in Fig. 1, which yields different ϕ factors for beams with and (4) Finally, the SC limit is calibrated in order that the target reliability
without stirrups. It is noted that an alternative to improve P is to op- index can be achieved for the shear strength of FRP-RC beams with
timize the terms of equations as was done in [17]. This is, however, not FRP stirrups, which is now evaluated based on the newly proposed
within the scope of the current paper, as stated earlier. Even in such a characteristic values for both Vc and Vf, as derived in previous steps.
case, there is no guarantee to have the same ϕ factors for both beams In this case, the professional factor is evaluated by 109 specimens –
without and with stirrups. the largest experimental database available at the present time for
In order to fulfill the above-mentioned constraints and to arrive at a beams with FRP stirrups.
desirable reliability target, the following steps are taken:
The above steps are implemented for both ACI and CSA codes as
(1) The current equation for concrete contribution to the shear strength reported in Table 8. More details are explained as follows.
Vc is multiplied by the mean value (μP) of its professional factor. In As step 1 of the above procedure for ACI, the current equation of Vc,
this way, the newly proposed shear strength (i.e., μP × Vc) yields a which is very conservative with the mean value of P equal to 1.786, is
professional factor having the mean value of about unity. In other modified to 1.786Vc. In here, thanks to a test database of as large as 288

Table 8
Modified equations of shear strength provisions of ACI and CSA for FRP-RC beams (βT = 4.5).
State Number Vpre Description Number of Vexp Pequ
P=
specimens Vpre

FORM FOSML FOSMN

μ δ μ δ μ μ μ

19 1.786Vc a Vc and Vf are based on the current ACI 288 1.000 0.208 0.934 0.154 4.538 4.654 3.935
20 1.786Vc + (Vf 0.875Vsc, ACI ) a code. 109 1.178 0.243 1.076 0.172 4.541 4.552 3.496
21 1.786Vc + Vf 32 1.062 0.283 0.943 0.191 3.411 3.320 2.709
22 1.786Vc + 0.875Vsc, ACI 19 0.834 0.190 0.787 0.144 4.038 4.221 3.693
b
23 0.848Vc Vc and Vf are based on the current CSA 288 1.289 0.188 1.222 0.143 4.500 5.604 3.896
24 (0.848Vc + 1.346Vf ) 0.424Vsc, CSA b code. 109 1.487 0.223 1.361 0.162 4.499 4.844 3.447
25 0.848Vc + 1.346Vf 32 1.289 0.305 1.125 0.200 2.530 5.033 2.094
26 0.424Vsc, CSA 19 1.414 0.215 1.315 0.158 4.499 4.648 3.436

Note: For ACI, the above results are presented under load combination of 1.2D + 1.6L and ϕ = 0.47; and for CSA, the calculations are for 1.25D + 1.5L and ϕc = 0.65
and ϕf = 0.75.
a
For βT = 3.5 and 4.0 in ACI, ϕ should be taken as 0.62 and 0.55, respectively.
b
For βT = 3.5 and 4.0 in CSA, the equation is multiplied by kc = 1.278 and 1.131, respectively.

794
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

Modified ACI without stirrup Modified CSA without stirrup 6.0


Modified ACI with stirrup Modified CSA with stirrup
4.5
5.0

Reliability index,
4.0
Reliability index,

3.0
4.0
Current ACI without stirrup
(k c ) (k c ) Current ACI with stirrup
2.0
Current CSA without stirrup
Current CSA with stirrup
1.0
0.0 0.2 0.4 0.6 0.8 1.0
3.5 (a) Live load ratio, L / (L+D)
0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Strength reduction factor\Calibration factor
6.0
Fig. 2. Variation of β in terms of ϕ or kc for the modified versions of ACI and
CSA provisions. 5.0

βT=4.5

Reliability index,
specimens exclusively for FRP-RC beams, Vc could be refined with more 4.0
β =4.0
T
accuracy. It should be noted that the concrete shear strength currently
employed in ACI 440 (2015) is based on the work by Tureyen and 3.0
βT=3.5
Frosch [83], who proposed shear strength equations to be applicable for Modified ACI without stirrup
2.0
both steel and FRP-reinforced beams by calibrating against 370 test Modified ACI with stirrup
specimens of which only 44 data were FRP-RC beams and the re-
1.0
mainder were steel-reinforced concrete beams. Also, as stated by [83], 0.0 0.2 0.4 0.6 0.8 1.0
the proposed equation contained more conservatism for FRP-RC beams (b) Live load ratio, L / (L+D)
compared with steel-RC beams due to simplification made in equation
as well as the limited data for FRP that was available at that time. In 6.0
this regard, the coefficient of 0.54 in their initially proposed equation
was then altered to 0.4 to account for conservatism. However, in the 5.0
context of reliability analysis, the required conservatism should not be βT=4.5
Reliability index,

engaged at the stage of model development but after an accurate model 4.0 β =4.0
T
is achieved, the resistance reduction factors are calibrated to meet a β =3.5
target reliability. In this capacity, the present paper firstly develops a 3.0 T
predictive equation such that its professional factor is as close as to
2.0 Modified CSA without stirrup
unity while noticing that the original equation by Tureyen and Frosch Modified CSA with stirrup
[83] had been overly conservative for FRP-RC beams. Then, the re-
1.0
duction factors are determined by reliability analysis to bring the pro- 0.0 0.2 0.4 0.6 0.8 1.0
posed equation to the level of the desirable safety. Thereby, the con- (c) Live load ratio, L / (L+D)
servatism related to different values of target reliability is achieved by
Fig. 3. Variation of β against live load ratios (L/L + D) for: (a) the current
the present paper through the recommended reduction factors applied provisions of ACI and CSA; (b) the modified equations (see Table 9) of ACI for
to the proposed equations. βT = 3.5, 4.0 and 4.5; (c) the modified equations (see Table 9) of CSA for
To facilitate the following explanations on the proposed procedure, βT = 3.5, 4.0 and 4.5.
there is a need to assume a certain target reliability index. However, the
final equations can be readily adjusted for other target values of the
shear strength reduction factor of ϕ = 0.47 proposed herein is sub-
reliability index. As for step 2 and with the aim of reaching a target
stantially lower than the current value of 0.75 in ACI 440. The reduc-
reliability index of 4.5 (as an example of βT), a resistance reduction
tion factor of 0.75 was suggested basically for shear strength of steel-RC
factor of ϕ = 0.47 is required. This is readily computed as 0.85/
beams in ACI 318 but was adopted by ACI 440 for FRP reinforcement.
1.786 = 0.47, in which the coefficient of 0.85 was derived in the past
However, considering that the current reduction factor for flexure in
section (see Fig. 1) to calibrate the current shear provision of ACI for
ACI 440 is either 0.55 for FRP rupture or 0.65 for concrete crushing,
beams without stirrup so as to meet βT = 4.5. It should be noticed that
and that the common approach in design guidelines is to apply a more

Table 9
Modified equations of shear strength provisions of ACI and CSA for FRP-RC beams at different target reliability indexes.
Modified Equations Description βT = 3.5 βT = 4.0 βT = 4.5

Afv ACI (modified) = 0.62 = 0.55 = 0.47


Vc = 0.714 f c' b w (kd), Vf = f d
s fv

Vsc, ACI = 0.583 f c' b w d

1 CSA (modified) A1 = 0.119 A1 = 0.106 A1 = 0.093


Vc = A1 f c' b w dv A2 km kr (fc' ) 3 b w dv A3 fc' bw dv
Afv A2 = 0.054 A2 = 0.048 A2 = 0.042
Vf = B f d cot( )
s fv v A3 = 0.238 A3 = 0.211 A3 = 0.187
Vsc, CSA = Cf c' b w dv B = 1.720 B = 1.522 B = 1.346
( = 0.65, = 0.75) C = 0.119 C = 0.105 C = 0.093
c f

Note: Various notations are the same as those appeared in Table 1.

795
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

reduced factor to shear strength compared with flexural strength, the reliability level for the proposed shear strength provisions of FRP-RC
existing value of 0.75 for shear reduction factor in the case of FRP bars beams is important as opposed to the existing shear strength equations
needs to be lowered, as stated by others as well [6]. To this end, the in the current codes of ACI and CSA, which have reliability indexes
value of 0.47 suggested in this paper appears to be a rational value. As varying in the range of 3.1–5, as presented in Table 6. Meantime, it is
for step 3, Table 8 (state 21) gives the statistical descriptors for the noticed that the target reliability for shear failure is commonly con-
professional factor of shear strength for beams with stirrups, in which sidered to be different from that of flexural failure due to brittleness of
concrete shear strength is 1.786Vc as proposed above and stirrup shear shear modes [24]. This is because the target reliability index is de-
strength is the current equation of ACI for Vf. As can be seen, the mean termined based on various concerns including the failure consequences
value of P for this case is about unity, meaning that the ACI provision which are obviously affected by brittle or ductile behavior. Yet, as
for Vf is accurate enough. It is recalled that the equation of Vf in the mentioned earlier, decision on the target reliability index is not within
present version of ACI 440 is based on the tensile strain of 0.004 for FRP intention of this work.
shear reinforcement, whereas the previous versions of ACI 440-01 [79]
were based on the FRP strain of 0.002 that was justified to be very
conservative. In agreement with the current approach of ACI 440, the 5.8. Reliability analysis of shear strength provisions under different load
results herein endorse the suitability of tensile strain of 0.004 for FRP combinations
stirrups. Finally, at step 4, the current SC limit of ACI is calibrated to
bring the shear strength equation of beams with stirrups to the target In order to assess the effect of varying ratios of loads in different
reliability index of 4.5, and the results are reported as state 20 in load combinations on the safety of shear strength equations, the relia-
Table 8. It is again emphasized that by incorporating a properly cali- bility index is determined across a wide range of ratio of principal load
brated SC limit to the shear strength equation, the desirable reliability to the total load, for instance L/L + D. The reliability curves, that de-
index can be achieved. monstrate variation of reliability index with respect to load ratios, are
Table 8 also summarizes the modified equations for shear capacity presented in Figs. 3–6 for live, snow, wind and earthquake loads, re-
based on CSA code. Steps 1 and 2 of the above-mentioned procedure spectively. The statistical descriptors of each load type are listed in
propose the concrete shear strength of 0.848Vc, where Vc is the current
6.0
provision of CSA. According to step 3, Vf of CSA should be multiplied by
a factor of 1.346 in order that the mean values of P for shear strengths
5.0
of beams with and without stirrups become equal (i.e., 1.289), as seen
in states 23 and 25 in Table 8. The coefficient of 1.346, which is herein
Reliability index,

4.0
proposed to increase the characteristic value of Vf in CSA, is justified
because the current provision of CSA for Vf is based on an FRP strain of
3.0
about 0.003 as compared to the strain of 0.004 employed in ACI. At Current ACI without stirrup
last, the SC limit is calibrated to capture the target reliability index. Current ACI with stirrup
2.0
The above analysis for ACI and CSA has focused on target reliability Current CSA without stirrup
Current CSA with stirrup
of 4.5. In order to meet other target reliability indexes of 3.5 and 4 for 1.0
ACI, it would suffice to replace ϕ by 0.62 and 0.55, respectively. The 0.0 0.2 0.4 0.6 0.8 1.0
(a) Snow load ratio, S / (S+D)
variation of target reliability index with respect to resistance reduction
factor for modified equations of ACI is illustrated in Fig. 2. As for CSA,
the equations in Table 8 need to be multiplied by 1.278 and 1.131 to 6.0
achieve βT = 3.5 and 4, respectively (see Fig. 2). It is noted that the CSA
partial safety factors of ϕc = 0.65 and ϕf = 0.75 were kept constant for 5.0
modified versions of CSA code. To summarize, Table 9 presents the βT=4.5
Reliability index,

finally modified equations of shear strength for both ACI and CSA under 4.0
β =4.0
T
different target reliability indexes of βT = 3.5, 4.0 and 4.5. As listed in
3.0
Table 9, the modified SC limit for ACI is proposed to be 0.583 f c' bw d βT=3.5
with ϕ = 0.62, 0.55 and 0.47 for βT = 3.5, 4.0 and 4.5, respectively, as Modified ACI without stirrup
2.0
compared to the current SC limit of ACI, that is 2/3 f c' bw d with Modified ACI with stirrup
ϕ = 0.75. For CSA, the SC limit suggested herein is in the range of 1.0
0.12f'c bw d to 0.09f'c bw d (for βT = 3.5–4.5) in contrast to 0.22f'c bw d 0.0 0.2 0.4 0.6 0.8 1.0
(b) Snow load ratio, S / (S+D)
currently used in CSA.
With reference to Table 6, which presents the dispersion δ of the
calculated β for the current equations, it is seen that δ is small and less 6.0
than 3% in all states by FORM method. When the modifications are
made and by comparing Tables 6 and 8, it is seen that the mean of 5.0
βT=4.5
professional factor is reduced to unity. Besides, δ of P for the case of
Reliability index,

beams with stirrup is also reduced compared to the original equations, 4.0 β =4.0
T
as a result of the proposed procedure. Further decrease in δ for P re-
3.0
quires substantial change of the equations, as discussed earlier. None- βT=3.5
theless, all values of δ for calculated β are small enough in this study on Modified CSA without stirrup
2.0
the design space. Modified CSA with stirrup
As stated in above, the professional factors for Vc, Vsc, and Vf are
1.0
adjusted in the modified equations (Table 8) to develop predictive 0.0 0.2 0.4 0.6 0.8 1.0
models which are as accurate as possible. Then, the resistance reduction (c) Snow load ratio, S / (S+D)
factors are calibrated on the basis of one common value of target re-
Fig. 4. Variation of β against snow load ratios (S/S + D) for: (a) the current
liability index (see Table 9) regardless of the mode of shear failure (i.e.,
provisions of ACI and CSA; (b) the modified equations of ACI for βT = 3.5, 4.0
SC or ST). This aspect of the present work in providing the same
and 4.5; (c) the modified equations of CSA for βT = 3.5, 4.0 and 4.5.

796
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

6.0 6.0
Current ACI without stirrup
5.0 5.0 Current ACI with stirrup
Current CSA without stirrup
Current CSA with stirrup

Reliability index,
Reliability index,

4.0 4.0

3.0 3.0
Current ACI without stirrup
2.0 Current ACI with stirrup 2.0
Current CSA without stirrup
Current CSA with stirrup
1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(a) Wind load ratio, W / (W+D) (a) Earthquake load ratio, E / (E+D)

6.0 6.0

5.0 5.0 Modified ACI without stirrup


Modified ACI with stirrup
βT=4.5
Reliability index,

Reliability index,
4.0 βT=4.5
β =4.0 4.0
T βT=4.0
3.0 3.0
βT=3.5
βT=3.5
2.0 Modified ACI without stirrup
2.0
Modified ACI with stirrup
1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(b) Wind load ratio, W / (W+D) Earthquake load ratio, E / (E+D)
(b)
6.0 6.0

5.0 5.0 Modified CSA without stirrup


βT=4.5 Modified CSA with stirrup
Reliability index,

Reliability index,

4.0 β =4.0 4.0


T
βT=4.5
3.0 3.0 βT=4.0
βT=3.5

2.0 Modified CSA without stirrup 2.0


Modified CSA with stirrup βT=3.5
1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(c) Wind load ratio, W / (W+D) Earthquake load ratio, E / (E+D)
(c)
Fig. 5. Variation of β against wind load ratios (W/W + D) for: (a) the current
Fig. 6. Variation of β against earthquake load ratios (E/E + D) for: (a) the
provisions of ACI and CSA; (b) the modified equations of ACI for βT = 3.5, 4.0
current provisions of ACI and CSA; (b) the modified equations of ACI for
and 4.5; (c) the modified equations of CSA for βT = 3.5, 4.0 and 4.5.
βT = 3.5, 4.0 and 4.5; (c) the modified equations of CSA for βT = 3.5, 4.0 and
4.5.
Table 2. For each load type, the reliability curves are derived for both
current provisions as well as modified equations of ACI and CSA. For
specific shear provision in a design code is not affected by different
each design code, the related load combinations, as listed in Table 5,
ratios of dead and live loads. The above-mentioned conclusions are also
are employed.
valid for the load combinations containing snow or wind, as illustrated
Fig. 3(a) compares the current shear provisions of ACI and CSA
in Figs. 4 and 5.
under basic load combination of dead and live loads. As can be seen
As for earthquake load combination, Fig. 6 illustrates the variation
from Fig. 3(a), the levels of reliability are different among different
of β in terms of the ratio of earthquake load to the total load, E/E + D. It
curves, meaning that from one hand, the safety levels between the two
is observed that as this ratio increases, the reliability index for each of
codes of ACI and CSA for beams without stirrup are different from each
the shear provisions, either original or modified version, drops sud-
other, and from other hand, in each code the average reliability of
denly. This is because the coefficient of variation for earthquake load is
beams without stirrup is not identical to that with stirrup – a conclusion
a high value as seen in Table 2. Besides, the probabilistic distribution
which was derived earlier too. Fig. 3(b) and (c) show the reliability
function for earthquake load is Extreme Value Type 2 with a heavier
curves for the modified versions of ACI and CSA, respectively, at dif-
upper tail in which there is a high probability to exceed the design
ferent target reliability indexes of 3.5, 4.0 and 4.5. It can be seen that at
point, leading to lower the reliability index. Nonetheless, it is assumed
each level of target reliability index, the reliability offered by modified
that earthquake is resisted by a parallel system consisting of several
shear strength for beams without stirrup coincides with that for beams
members rather than just one member. Assessment of reliability for a
with stirrups over a wide range of load ratios. This result shows an
system containing more than a member is however beyond the scope of
improvement as compared to the current provisions of the considered
this paper. On the other hand, the target reliability index for earthquake
design codes, where the reliability curves of beams with stirrup differ
load may be relaxed as compared to that for other load combinations
from those of beams without stirrups. It is also seen that across a
because some amount of damage may be allowed under earthquake
practical range of load ratios (say 0.3–0.7), the level of reliability for a
event. To this end, the current paper is not intended to propose the
particular equation of shear strength corresponding to a target relia-
reliability index for earthquake load combinations.
bility index is not varied that much. This indicates that the safety of a

797
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

6.0 6.0
W L
5.0 5.0 (W+D) (L+D)
E S
Reliability index,

Reliability index,
4.0 4.0 (E+D) (S+D)

3.0 W L 3.0
(W+D) (L+D)
E S
2.0 2.0
(E+D) (S+D)

1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(a) Load ratio (b) Load ratio

6.0 6.0

5.0 5.0
Reliability index,

Reliability index,
4.0 4.0

3.0 W L 3.0 W L
(W+D) (L+D) (W+D) (L+D)
2.0 E S 2.0 E S
(E+D) (S+D) (E+D) (S+D)
1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
(c) Load ratio (d) Load ratio

Fig. 7. Reliability index β versus varied load ratios for live, snow, wind and earthquake load combinations in ACI: (a) the current provisions of ACI for beams without
stirrup; (b) the current provisions of ACI for beams with stirrup; (c) the modified equations of ACI for beams without stirrup; (d) the modified equations of ACI for
beams with stirrup.

6.0 6.0
W L W L
(W+D) (L+D) (W+D) (L+D)
5.0 E S 5.0
E S
(E+D) (S+D)
Reliability index,

(S+D)
Reliability index,

(E+D)
4.0 4.0

3.0 3.0

2.0 2.0

1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Load ratio Load ratio
(a) (b)
6.0 6.0

5.0 5.0
Reliability index,

Reliability index,

4.0 4.0

3.0 3.0
W L W L
(W+D) (L+D) (W+D) (L+D)
2.0 2.0 E S
E S
(E+D) (S+D) (E+D) (S+D)
1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Load ratio Load ratio
(c) (d)
Fig. 8. Reliability index β versus varied load ratios for live, snow, wind and earthquake load combinations in CSA: (a) the current provisions of CSA for beams
without stirrup; (b) the current provisions of CSA for beams with stirrup; (c) the modified equations of ACI for beams without stirrup; (d) the modified equations of
CSA for beams with stirrup.

Fig. 7(a–d) depict the effect of varied ratios of different load types also achieved over a practical range of load ratios.
(i.e., live, snow, wind and earthquake) altogether on reliability index of
the current provision along with the modified version (βT = 4.5) for 6. Conclusions
ACI. Fig. 8 shows similar information for CSA. In addition to confirming
the previously mentioned results, Figs. 7 and 8 indicate that the average In the present paper, the shear strength provisions of ACI 440 and
reliability index of a shear provision under live load in a practical range CSA S806-2 for FRP-RC beams have been evaluated by reliability ana-
of load ratios is almost the same as that under the other load combi- lysis under variety of load combinations including dead, live, snow,
nations of snow and, to a lesser extent, wind. This implies that by sa- wind and earthquake loads. The probabilistic characteristics of all
tisfying the safety of shear strength equations under the basic load random variables involved in resistance models and load combinations
combination of dead and live loads, the safety under snow or wind is have been considered. The professional factors (i.e., model uncertainty)

798
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

for shear strength equations have been assessed using a large database by satisfying the safety for basic load combination of dead and live
of shear-failed test results, which was assembled from literature and loads, the safety for snow or wind load combinations is also
contained 288 FRP-RC beams without stirrups and 109 beams with FRP achieved over a practical range of load ratios.
stirrups. The conclusions are as follows:
Future research may focus on reliability analysis of shear resistance
(1) The reliability indexes for shear strength provisions of ACI for FRP- of FRP-RC beams under earthquake load combination taking into ac-
RC beams without stirrup and with stirrup are equal to 5.0 and 3.1, count reliability assessment of a system consisting of more than a single
respectively, under the basic load combination of dead and live member. Besides, the reliability analysis on shear design formulas of
loads. As for CSA, the reliability indexes of 3.8 and 3.4 are de- other codes such as Eurocode and fib is suggested as future research
termined for beams without and with stirrup, respectively. Thus, in needs, which can provide a comparison with the outcomes of the cur-
ACI, the safety level offered for beams without stirrup is sub- rent paper.
stantially higher than that for beams with stirrup. Likewise, in CSA,
the difference exists in reliability for beams without versus with 7. Funding
stirrup but to a lesser extent as compared to ACI.
(2) The professional factors evaluated using available experimental This research received no specific grant from any funding agency in
data show that the concrete shear contribution in FRP-RC beams as the public, commercial, or not-for-profit sectors.
provided by CSA is accurate enough (with the mean of professional
factor, μP = 1.09), whereas that by ACI is overly conservative 8. Declaration of conflicting interests
(μP = 1.79). On the other hand, the FRP stirrup shear contribution
suggested by ACI appears to be proper while that by CSA is some- The authors hereby declare that there is no conflict of interest.
what conservative.
(3) Through reliability assessment of different SC limits, it is revealed References
that incorporation of an accurate SC limit into the shear capacity
equation of beams with stirrup is of central importance to enhance [1] Fico R, Prota A, Manfredi G. Assessment of Eurocode-like design equations for the
the reliability of the existing shear provisions. In the absence of any shear capacity of FRP RC members. J Compos Part B Eng 2008;39(5):792–806.
[2] El-Sayed A, Soudki K. Evaluation of shear design equations of concrete beamswith
SC limit or by employing an unrealistically high amount for the SC FRP reinforcement. J Compos Constr ASCE 2011;15(1):9–20.
limit, the reliability index of the shear resistance of beams with [3] Nehdi M, El Chabib H, Aly Said A. Proposed shear design equations for FRP-re-
stirrup drops since shear-compression failed mode is not dis- inforced concrete beams based on genetic algorithms approach. J Mater Civil Eng
ASCE 2007;19(12):1033–42.
tinguished from shear-tension failed one. The current SC limits [4] Ashour AF, Kara IF. Size effect on shear strength of FRP reinforced concrete beams.
provided by ACI and CSA for FRP-RC beams, which are essentially J Compos Part B Eng 2014;60:612–20.
identical to those for steel-RC beams, prove to be high values, [5] Nasrollahzadeh K, Basiri MM. Prediction of shear strength of FRP reinforced con-
crete beams using fuzzy inference system. J Expert Syst Appl 2014;41(4):1006–20.
leading to lower the reliability of the existing relations for shear
[6] Zadeh HJ, Nanni A. Reliability analysis of concrete beams internally reinforced with
capacity design. fiber-reinforced polymer bars. ACI Struct J 2013;110(6):1023–32.
(4) In order that the existing shear provisions of ACI can meet βT = 3.5, [7] Machial R, Alam MS, Rteil A. Revisiting the shear design equations for concrete
beams reinforced with FRP rebar and stirrup. Mater Struct 2012;45(11):1593–612.
4 and 4.5, the value of ϕ for beams without stirrup shall be 1.1,
[8] Razaqpur AG, Spadea S. Shear strength of FRP reinforced concrete members with
0.97, and 0.85 whereas for beams with stirrup, ϕ shall become 0.66, stirrups. J Compos Constr 2014;19(1):04014025.
0.56, and 0.47, respectively. To follow the ACI approach in em- [9] Bashir R, Ashour A. Neural network modelling for shear strength of concrete
ploying a single value of ϕ for both beams with and without stirrup members reinforced with FRP bars. J Compos Part B Eng 2012;43(8):3198–207.
[10] Kara IF. Prediction of shear strength of FRP-reinforced concrete beams without
to arrive at the same reliability index, the shear strength relations stirrups based on genetic programming. J Adv Eng Softw 2011;42(6):295–304.
need to be altered. As for the current shear equations of CSA, [11] Nehdi M, El Chabib H, Said A. Evaluation of shear capacity of FRP reinforced
ϕc = 0.7 and ϕf = 0.66 are required for βT = 3.5; ϕc = 0.62 and concrete beams using artificial neural networks. Smart Struct Syst
2006;2(1):81–100.
ϕf = 0.55 for βT = 4.0; and ϕc = 0.55 and ϕf = 0.46 are needed for [12] Pham HB, Al-Mahaidi R. Reliability analysis of bridge beams retrofitted with fibre
βT = 4.5. Alternatively, ϕc = 0.65 and ϕf = 0.75, which are cur- reinforced polymers. Compos Struct 2008;82(2):177–84.
rently used in the entire CSA code, can be kept unchanged provided [13] Zeng Y, Botte W. Caspeele R. Reliability analysis of FRP strengthened RC beams
considering compressive membrane action. Constr Build Mater
that the shear equations themselves are to be modified to satisfy a 2018;30(169):473–88.
specific value of βT. [14] He Z, Qiu F. Probabilistic assessment on flexural capacity of GFRP-reinforced
(5) In this paper, modifications are proposed to the coefficients existing concrete beams designed by guideline ACI 440.1 R-06. Constr Build Mater
2011;25(4):1663–70.
in shear design relations while maintaining the format of the cur- [15] Ribeiro SEC, Diniz SMC. Reliability-based design recommendations for FRP-re-
rent equations in ACI and CSA for FRP-RC beams. The modified inforced concrete beams. Eng Struct 2013;52:273–83.
shear strength expressions (see Table 9) have the following features: [16] Shield CK, Galambos TV, Gulbrandsen P. On the history and reliability of the
flexural strength of FRP reinforced concrete members in ACI 440.1 R. In: Proc, 10th
First, Vc and Vf in each code are refined to yield a professional factor
int symp on fiber-reinforced polymer reinforcement for concrete structures, vol.
with the mean value of about unity; Second, in line with the ACI 275. Farmington Hills (MI): ACI; 2011. p. 1–18.
approach, a single value of ϕ will suffice to offer the same reliability [17] Shahnewaz M, Machial R, Alam MS, Rteil A. Optimized shear design equation for
for beams with and without stirrup. As for modified shear strength slender concrete beams reinforced with FRP bars and stirrups using Genetic
Algorithm and reliability analysis. Eng Struct 2016;107:151–65.
equations of CSA, ϕc = 0.65 and ϕf = 0.75, which are currently [18] ACI Committee 440. Guide for the design and construction of structural concrete
adopted throughout the CSA code, are applied for all target relia- reinforced with FRP bars. Farmington Hills (Michigan): American Concrete
bility indexes. Institute; 2006. ACI 440.1R-06.
[19] CSA. Design and construction of building components with fibre-reinforced poly-
(6) The shear compression limit for shear provisions of ACI is proposed mers. Rexdale (Ont): Canadian Standards Association; 2002. CSA standard CAN/
to be 0.583 f c' bw d with ϕ = 0.62, 0.55 and 0.47 for βT = 3.5, 4.0 CSA S806-02.
[20] ACI Committee 440. Guide for the design and construction of structural concrete
and 4.5, respectively as compared to the current SC limit of ACI,
reinforced with FRP bars. Farmington Hills (Michigan): American Concrete
that is 2/3 f c' bw d with ϕ = 0.75. For CSA, the SC limit suggested Institute; 2015. ACI 440.1R-15.
herein is in the range of 0.12f'c bw d to 0.09f'c bw d (for βT = 3.5–4.5) [21] CSA. Design and construction of building components with fibre-reinforced poly-
mers. Rexdale (Ont): Canadian Standards Association; 2012. CSA standard CAN/
as opposed to 0.22f'c bw d currently used in CSA. CSA S806-12.
(7) Reliability analysis of the proposed shear design equations under [22] Mirza SA, MacGregor JG. Probabilistic study of strength of reinforced concrete
different load combinations containing snow or wind indicates that members. Can J Civ Eng 1982;9(3):431–48.
[23] Zadeh HJ, Mejia F, Nanni A. Strength reduction factor for flexural RC members

799
K. Nasrollahzadeh, R. Aghamohammadi Engineering Structures 176 (2018) 785–800

strengthened with near-surface-mounted bars. J Compos Constr [55] Gross SP, Dinehart DW, Yost Jr, Theisz PM. Experimental tests of high-strength
2013;17(5):614–25. concrete beams reinforced with CFRP bars. In: 4th International conference on
[24] Nowak AS, Szerszen MM. Calibration of design code for buildings (ACI 318): Part advanced composite materials in bridges and structures; 2004. p. 20–3.
1—Statistical models for resistance. ACI Struct J 2003;100(3):377–82. [56] Tariq M, Newhook JP. Shear testing of FRP reinforced concrete without transverse
[25] Kocaoz S, Samaranayake VA, Nanni A. Tensile characterization of glass FRP bars. reinforcement. In: Proceedings Annual conference – Canadian society for civil en-
Compos Part B Eng 2005;36(2):127–34. gineering; 2003. p. 1330–9.
[26] Michaluk R, Rizkalla S, Tadros G, Benmokrane B. Flexural behavior of one-way [57] Deitz DH, Harik IE, Gesund H. One-way slabs reinforced with glass fiber reinforced
concrete slabs reinforced by fiber reinforced plastic reinforcement. ACI Struct J polymer reinforcing bars, fiber reinforced polymer reinforcement for reinforced
1998;95(3):353–65. concrete structures. Dolan CW, editor. Proceedings of the 4-th international con-
[27] El-Sayed AK, El-Salakawy EF, Benmokrane B. Shear strength of FRP-reinforced ference, SP-188. Farmington Hills (Mich): American Concrete Institute; 1999. p.
concrete beams without transverse reinforcement. ACI Struct J 279–86.
2006;103(2):235–43. [58] Mizukawa Y, Sato Y, Ueda T, Kakuta Y. A study on shear fatigue behavior of con-
[28] El-Sayed A, El-Salakawy E, Benmokrane B. Shear strength of one-way concrete slabs crete beams with FRP beams with FRP rods. Non-metallic (FRP) reinforcement for
reinforced with fiber-reinforced polymer composite bars. J Compos Constr concrete structure, V. 2, proceedings of the third international symposium on non-
2005;9(2):147–57. metallic (FRP) reinforcement for concrete structure Japan Concrete Institute; 1997.
[29] Nowak AS, Collins KR. Reliability of structures. Boston: McGraw-Hill; 2000. p. 309–16.
[30] Forouzannia F, Gencturk B, Dawood M, Belarbi A. Calibration of flexural resistance [59] Jang H, Kim M, Cho J, Kim C. Concrete shear strength of beams reinforced with FRP
factors for load and resistance factor design of concrete bridge girders prestressed bars according to flexural reinforcement ratio and shear span to depth ratio. In:
with carbon fiber–reinforced polymers. J Compos Constr 2015. 04015050. FRPRCS-9, Sidney, Australia; 2009.
[31] El-Sayed AK, El-Salakawy EF, Benmokrane B. Shear capacity of high-strength [60] Yost J R, Gross S, Masone G C. Sustained load effects on deflection and ultimate
concrete beams reinforced with FRP bars. ACI Struct J 2006;103(3):383–9. strength of concrete beams reinforced with GFRP. In: CCC2005, Hamelin P, editor.
[32] Ashour AF. Flexural and shear capacities of concrete beams reinforced with GFRP Lyon (France): ACMBS; 2005.
bars. Constr Build Mater 2006;20(10):1005–15. [61] Kilpatrick A E, Dawborn R. Flexural shear capacity of high strength concrete slabs
[33] Razaqpur A, Isgor B, Greenaway S, Selley A. Concrete contribution to the shear reinforced with longitudinal GFRP bars. In: FIB, Naples, Italy; 2006; 1–10.
resistance of fiber reinforced polymer reinforced concrete members. J Compos [62] Steiner S, El-Sayed, AK, Benmokrane B, Matta F, Nanni A. Shear strength of large-
Constr 2004;8(5):452–60. size concrete beams reinforced with glass FRP bars. In: 5th International conference
[34] Tureyen AK, Frosch RJ. Shear tests of FRP-reinforced concrete beams without on advanced composite materials in bridges and structures, Winnipeg, Manitoba,
stirrups. ACI Struct J 2002;99(4):427–34. Canada; 2008.
[35] Yost JR, Goodspeed CH, Schmeckpeper ER. Flexural performance of concrete beams [63] Nagasaka T, Fukuyama H, Tanigaki M. Shear performance of concrete beams re-
reinforced with FRP grids. J Compos Constr 2001;5(1):18–25. inforced with FRP stirrups. International symposium on fiber-reinforced-plastic
[36] Alkhrdaji T, Wideman M, Belarbi A, Nanni A. Shear strength of RC beams and slabs. reinforcement for concrete structures, SP-138. 1993. p. 789–811.
Composites in construction 2001:409–14. [64] Okamoto T, Nagasaka T, Tanigaki M. Shear capacity of concrete beams using FRP
[37] Duranovic N, Pilakoutas K, Estes Waldron P. on concrete beams reinforced with reinforcement. J Struct Constr Eng 1994;455:27–136.
glass fiber reinforced plastic bars. Non-metallic (FRP) reinforcement for concrete [65] Nakamura H, Higai T. Evaluation of shear strength of concrete beams reinforced
structure, V. 2, proceedings of the third international symposium on non-metallic with FRP. Concr Libr Int 1995;26:111–23.
(FRP) reinforcement for concrete structure Japan Concrete Institute; 1997. p. [66] Tottori S, Wakui H. Shear capacity of RC and PC beams using FRP reinforcement. In:
479–86. Proceedings of the international symposium on fiber-reinforced-plastic reinforce-
[38] Alam MS, Hussein A. Effect of member depth on shear strength of high-strength ment for concrete structures, SP-138; 1993. p. 615–31.
fiber-reinforced polymer–reinforced concrete beams. J Compos Constr [67] Maruyama K, Zhao WJ. Flexural and shear behaviour of concrete beams reinforced
2012;16(2):119–26. with FRP rods. In: Proceedings in the international conference on corrosion and
[39] Bentz EC, Massam L, Collins MP. Shear strength of large concrete members with corrosion protection of steel in concrete; 1994. p. 24–8.
FRP reinforcement. J Compos Constr 2010;14(6):637–46. [68] Maruyama K, Zhao W J. Size effect in shear behavior of frp reinforced concrete
[40] Guadagnini M, Pilakoutas K, Waldron P. Shear resistance of FRP RC beams: ex- beam. In: El-Badry M, editor. 2nd int. conf. on advanced composite materials in
perimental study. J Compos Constr 2006;10(6):464–73. bridges and structures, ACMBS; 1996; 227–34.
[41] Kim CH, Jang HS. Concrete shear strength of normal and lightweight concrete [69] Alsayed S, Al-Salloum Y, Almusallam T, Amjad M. Evaluation of shear stresses in
beams reinforced with FRP bars. J Compos Constr 2013;18(2):04013038. concrete bearns reinforced by FRP bars. In: ACMBS-II, Montreal, Québec; 1996;
[42] Zeidan M, Barakat MA, Mahmoud Z, Khalifa A. Evaluation of concrete shear 173–79.
strength for FRP reinforced beams. In: Proc, structures congress. Las Vegas (NV): [70] Ascione L, Mancusi G, Spadea S. Flexural behavior of concrete beams reinforced
ASCE; 2011. p. 1816–26. with GFRP bars. Strain 2010;46(5):460–9.
[43] Abdul-Salam B, Farghaly AS, Benmokrane B. Evaluation of shear behavior for one- [71] ACI Committee 318. Building code requirements for structural concrete (ACI
way concrete slabs reinforced with carbon-FRP bars. G E N 2013;243:1–11. 318–14) and commentary (318R–14). Farmington Hills (MI): American Concrete
[44] Yost JR, Gross SP, Dinehart DW. Associate members, ASCE. Shear strength of Institute; 2014.
normal strength concrete beams reinforced with deformed GFRP bars. J Compos [72] Canadian Standards Association. CAN/CSA A23.3-04 design of concrete structures.
Constr 2001;5(4):268–75. Rexdale (Ontario, Canada): CSA; 2004.
[45] Pantelides CP, Besser BT, Liu R. One-way shear behavior of lightweight concrete [73] Okeil A, Belarbi A, Kuchma D. Reliability assessment of FRP-strengthened concrete
panels reinforced with GFRP bars. J Compos Constr 2012;16(1):2–9. bridge girders in shear. J Compos Constr ASCE 2013:91–100.
[46] Razaqpur AG, Shedid M, Isgor B. Shear strength of fiber-reinforced polymer re- [74] Szerszen MM, Nowak AS. Calibration of design code for buildings (ACI 318): Part
inforced concrete beams subject to unsymmetric loading. J Compos Constr 2—Reliability analysis and resistance factors. ACI Struct J 2003;100(3):383–91.
2010;15(4):500–12. [75] Der Kiureghian A. First- and second-order reliability methods. In: Nikolaidis E,
[47] Gross SP, Yost JR, Dinehart DW, Svensen E, Liu N. Shear strength of normal and Ghiocel DM, Singhal S, editors. Engineering design reliability handbook. Raton
high strength concrete beams reinforced with GFRP bars. In: Proceeding on high (FL): CRC Press, Boca; 2005.
performance of materials in bridges. ASCE; 2003. p. 426–37. [76] Rackwitz R, Fiessler B. Structural reliability under combined random load se-
[48] Matta F, El-Sayed AK, Nanni A, Benmokrane B. Size effect on concrete shear quences. Comput Struct 1978;9(5):489–94.
strength in beams reinforced with fiber-reinforced polymer bars. ACI Struct J [77] ACI Committee 318. Building code requirements for structural concrete (ACI
2013;110(4):617–28. 318–11) and commentary (318R–11). Farmington Hills (MI): American Concrete
[49] Alam MS, Hussein A. Size effect on shear strength of FRP reinforced concrete beams Institute; 2011.
without stirrups. J Compos Constr 2012;17(4):507–16. [78] The Joint Committee on Structural Safety (JCSS).. Probabilistic model code.
[50] Zhao W, Maruyama K, Suzuki H. Shear behaviour of concrete beams reinforced by Lausanne (Switzerland): Joint Committee on Structural Safety; 2001.
FRP rods as longitudinal and shear reinforcement. In: Proceedings of the second [79] ACI Committee 440. Guide for the design and construction of structural concrete
international RILEM symposium on non-metallic (FRP) reinforcement for concrete reinforced with FRP bars. Farmington Hills (MI): American Concrete Institute;
structures, FRPRCS-2, Ghent; 1995. p. 352–9. 2001. ACI 440.1R-01.
[51] Olivito RS, Zuccarello FA. On the shear behavior of concrete beams reinforced by [80] Canadian Standards Association.. CAN/CSA A23.3-14 design of concrete structures.
carbon fibre-reinforced polymer bars: an experimental investigation by means of Rexdale (Ontario, Canada): CSA; 2014.
acoustic emission technique. Strain 2010;46(5):470–81. [81] Shehata EFG. Fiber reinforced polymer (FRP) for shear reinforcement in concrete
[52] Zhang B, Masmoudi R, Benmokrane B. Behaviour of one-way concrete slabs re- structures PhD. thesis Manitoba (Canada): Department of Civil and Geological
inforced with CFRP grid reinforcements. Constr Build Mater 2004;18(8):625–35. Engineering, Univ. of Manitoba; 1999.
[53] Massam L. The behavior of GFRP reinforced concrete beams in shear Master thesis [82] MacGregor JG, Bartlett FM. Reinforced concrete mechanics and design. 1st
Ont (Canada): Dept of Civil Engineering, Univ of Toronto; 2001 Canadian ed. Ontario: Prentice-Hall; 2000.
[54] Kilpatrick AE, Easden L. Shear capacity of GFRP reinforced high strength concrete [83] Tureyen A, Frosch RJ. Concrete shear strength: another perspective. ACI Struct J
slabs. Deeks AJ, Hao H, editors. Developments in mechanics of structures and 2003;100(5):609–15.
materials, vol. 1. London: Taylor & Francis; 2005. p. 119–24.

800

You might also like