Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Structures 28 (2020) 1580–1588

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Design and reliability analysis of FRP-reinforced concrete columns


Ahmad Tarawneh a, *, Sereen Majdalaweyh b
a
Civil Engineering Department, Hashemite University, Zarqa, Jordan
b
Glenn Department of Civil Engineering, Clemson University, Lowry Hall, SC 2963, United States

A R T I C L E I N F O A B S T R A C T

Keywords: ACI 440-15 does not provide design provisions for columns and requires ignoring the compression contribution
FRP-Reinforced columns of FRP bars. However, this paper provides a design recommendation to include FRP bars’ compression for more
Compressive contribution accurate predictions. The compressive contribution is included by limiting the strain to 0.002 for compression
Reliability analysis
bars. This recommendation is valid for circular and rectangle columns subjected to concentric and eccentric
Monte Carlo Simulation
GFRP
loading. This paper supports the design recommendation by reliability analyses using Monte Carlo simulations.
CFRP The analyses are based on the results of 194 column tests surveyed from literature. The analysis has been
conducted for a wide range of live to dead load ratios. Generally, the results show that FRP-reinforced columns
exhibit an acceptable levels of reliability index. Additionally, the reliability index was not sensitive to rein­
forcement ratio, eccentricity level, or concrete strength. However, the results show that CFRP columns have a
higher reliability index than GFRP columns. Consequently, a lower strength reduction factor for GFRP bars can
be utilized to achieve a consistent safety level.

1. Introduction in bridges’ decks and girders [4]. However, these design guidelines do
not provide provisions of using FRP for compression resistance and
Reinforced concrete structural elements in harsh corrosive environ­ require ignoring the contribution of FRP compression reinforcement in
ments, especially bridges, suffer from deterioration because of rein­ beams, and columns. The reason behind such requirements is the com­
forcing steel corrosion [1]. The cost of replacing or repairing bined effect of the lower FRP elastic modulus and strength in
deteriorated reinforced concrete elements can be considerably high. compression compared to the elastic modulus and strength in tension
Consequently, Fiber reinforced polymer (FRP) bars have emerged as a [7]. This results in a small contribution of FRP bars in compression at
promising alternative of conventional steel reinforcing bars for their ultimate concrete strain (εcu = 0.003) [5].
advantageous properties; primarily corrosion resistance. In addition, the Over the years, several research works attempted to improve the
use of FRP reinforcing bars possess electromagnetic transparency that understanding of FRP bars behavior in compression zones. Studies
makes them suitable for structures with magnetic resonance imaging showed that including the FRP compression contribution in the analysis
units [2]. While, FRP bars comprise of aligned fibers made of glass fibers can result in a better prediction for the capacity, while neglecting the
(GFRP), carbon fibers (CFRP), and Aramid fibers (AFRP). FRP bars have contribution of FRP bars in compression leads to additional conserva­
an elastic brittle stress–strain curve [3] with an elastic modulus of 20 to tism [8–11]. Different design recommendations were suggested to
25 percent of steel elastic modulus for GFRP, 60 to 80 percent for CFRP, include the FRP bars compression contribution in concentrically loaded
and 20 to 60 percent for AFRP. The ultimate tensile strength range be­ columns. Afifi et al., [12] and Tobbi et al., [13] recommended including
tween 413 and 2070 MPa (60 to 300 ksi) [3]. the compression strength of GFRP bars equals to 0.35 of the tensile
Tracing the roots of the research conductions in the 1990 s, an strength, which would improve the accuracy of predicted nominal
extensive experimental and analytical research works have been con­ compressive strength of GFRP-reinforced circular and rectangular col­
ducted to investigate the flexural and shear behavior of FRP-reinforced umns. Tobbi et al. reported that FRP bars contributed 10% of the column
flexural concrete elements [3–4]. The research studies formed the FRP- capacity which is close to steel bars contribution of 12%. Mohamed
reinforced flexural design guidelines (ACI 440.1R-15 [5], CAN/CSA et al., [14] tested 14 full scale FRP-reinforced circular columns and
S806-12 [6]for engineers. Consequently, FRP bars have been wildly used proposed to limit the FRP bars strain in compression to the initiation of

* Corresponding author.
E-mail addresses: ahmadn@hu.edu.jo (A. Tarawneh), smajdal@g.clemson.edu (S. Majdalaweyh).

https://doi.org/10.1016/j.istruc.2020.10.009
Received 2 August 2020; Received in revised form 28 September 2020; Accepted 6 October 2020
Available online 13 October 2020
2352-0124/© 2020 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

Table 1
Concentrically loaded columns data.
Reference No. of specimens b × h or Dia., mm f’c, MPa FRP type Ef, GPa ρf %
Afifi et al. [12] 9 300 42.9 GFRP 55.4 1.1–3.2
De Luca et al. [16] 4 610 × 610 33–43.7 GFRP 44.2–44.4 1
Elchalakani and Ma [17] 3 260 × 160 32.75 GFRP 46.3 1.83
Elchalakani et al. [18] 3 260 × 160 26 GFRP 46.3 2.37
Hadhood et al. [19] 1 305 35 CFRP 141 2.17
Hadhood et al. [10] 2 305 70.2 GFRP 54.9 2.18–3.27
Hadi and Youssef [20] 1 210 × 210 33 GFRP 67.9 1.15
Khorramian and Sadeghian [8] 2 150 × 150 37 GFRP 41.2 5.36
Mohammad et al. [14] 9 300 42.9 CFRP 55.4–140 1.79,2.24
Othman et al. [21] 5 150 × 150 44.7 CFRP 145–151 1.4–3.6
Tobbi et al. [13] 5 350 × 350 32.6 GFRP 47.6–48.2 1.87, 1.95
Xue et al. [11] 3 300 × 300 39 GFRP 39 1.34

plastic deformation in the concrete of 0.002 to provide more accurate Design (LRFD) methodology, which targets a certain reliability level by
results. Despite the fact that the ultimate limit state of the strain of the calibrating load and resistance factors. Limited research has been done
concrete is equal to 0.003 according to ACI code and 0.0035 according to find the reliability indices for column design: Peng et al., [4] con­
to the Canadian code. Mohamed et al. concluded that the use of FRP bars ducted a detailed reliability analysis for rectangular short eccentric
as lateral confinement (spiral and hoops) was effective in confining columns in order to calibrate the strength reduction factor for short
concrete. Guérin et al. [9] reported that the FRP-reinforced column’s columns. Two methodologies for reliability analysis were used in this
strength and contribution of FRP bars were significantly affected by the study; first-order second moment and Monte Carlo simulation, followed
eccentricity level. The study reported FRP bars contribution of 3%, 5%, by sensitivity analysis. They concluded that including FRP compression
and 13% of the peak load for specimens with low eccentricity (0.1–0.2 contribution results in accurate prediction and avoid high conservatism.
h), medium eccentricity (0.4 h), and high eccentricity (0.8 h) respec­ Additionally, the study showed that the reliability index was not sensi­
tively, where h is the column dimension perpendicular to the axis of tive to the eccentricity ratio or concrete strength but was sensitive to the
bending. The study also indicates that the behavior of steel and FRP- live to dead load ratio expressed as D/(D + L). This study aimed to
reinforced columns were similar under same eccentricity level. The achieve a uniform reliability index of 4.0 by recommending a strength
conclusions were based on testing 12 full-scale GFRP-reinforced reduction factor of 0.6 for columns with axial stiffness less than or equal
columns. 2 GPa (290 ksi), and 0.65 for those with axial stiffness more than 4 GPa
Several studies investigated the behavior and design of rectangle and (580 ksi), with linear interpolation in the transition region between the
circular columns under eccentric loading. Xue et al. [11] investigated two.
the behavior of slender FRP-reinforced columns using experimental and As mentioned earlier, design codes do not provide provisions
finite element analysis. They found that the behavior of FRP reinforced regarding FRP reinforced columns neither reliability-based calibration.
columns are similar to steel reinforced columns but with less contribu­ Accordingly, this study presents a general design procedure and LRFD
tion for FRP bars. Additionally, it was concluded that the failure was not based reliability analysis for circular and rectangular FRP-reinforced
triggered by the rupture of tensile FRP bars, rather by concrete crushing columns under concentrically and eccentrically loading. The design
even for columns with low reinforcement ratio and high eccentricity procedure and reliability analysis are based on large experimental data
level. Choo et al., [15] developed an analytical approach to investigate of circular and rectangle columns under eccentric and concentric
the axial-bending interaction diagrams for FRP-reinforced concrete loading. The reliability analysis was conducted using LRFD based Monte
columns. Choo et al. concluded that rectangle FRP reinforced columns Carlo simulation.
with reinforcement ratio between 1% and 8% do not exhibit balance
points unlike steel reinforced columns. This is due to the well-defined 2. Research significance
yielding point for steel bars, unlike FRP bars where the behavior re­
mains linear elastic until failure. Their study also indicated that ignoring Currently, design codes do not provide design provisions for FRP-
FRP bars contribution leads to conservatism. reinforced concrete columns and require ignoring the compressive
Including FRP bar contribution in compression zones has to be contribution of FRP bars. This study will assess researchers in updating
investigated through the state-of-the-art Load and Resistance Factor ACI-440 design provisions by presenting design guidelines supported by

Table 2
Eccentrically loaded rectangle columns data.
Reference No. of specimens b × h mm f’c, MPa FRP type Ef, GPa ρf % e/h

Elchalakani and Ma [17] 4 260 × 160 32.8 GFRP 46.3 1.83 0.16–0.22
Elchalakani et al. [18] 6 260 × 160 26 GFRP 46.3 2.37 0.16–0.5
Gong and Zhang [22] 10 180 × 250 24,29 CFRP 139 0.46, 0.9 0.1–1.44
Guérin [22] 8 405 × 405 42.3 GFRP 48.2–53.3 1.1 0.1–0.8
Guérin [23] 8 405 × 405 42.3 GFRP 54.4 1.39–2.47 0.1–0.8
Hadi and Youssef [20] 2 210 × 210 33 GFRP 67.9 1.15 0.12–0.24
Issa et al. [24] 2 150 × 150 24.7 GFRP 32.7 2.01 0.16,0.33
Khorramian and Sadeghian [8] 6 150 × 150 37 GFRP 39 5.33 0.1–0.3
Khorramian and Sadeghian [25] 1 305 × 203 48 GFRP 46 4.7 0.25
Othman et al. [21] 10 150 × 150 44.7 CFRP 145–151 1.4–3.6 0.5,1
salah Eldin et al. [26] 4 400 × 400 71.2 GFRP 62.7 1 0.2–0.6
Sharbatdar [27] 5 230 × 230 54.1 CFRP 147 0.38 0.26, 0.33
Sun et al. [28] 9 180 × 250 26.8 GFRP 92.4 1.05 0.3–0.7
Tikka et al. [29] 8 150 × 150 35.7 GFRP 53 2.3, 3.4 0.2–0.8
Xue et al. [11] 12 300 × 300 29–55.2 GFRP 39–44 1.3–2.6 0.2–1

1581
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

LRFD based reliability analysis to evaluate the safety level. The result (Table 3). The data comprises FRP-RC column subjected to both
shows that including the compressive contribution of FRP bars will concentric and eccentric loading with respect to most influential pa­
result in more accurate results with an acceptable safety level. rameters including: concrete strength (f’c, MPa), reinforcement ratio (ρf
%), FRP elastic modulus (Ef), eccentricity level (e/h), and column size
3. Experimental data defined by the dimensions of the cross sectional area: width (b) and
height (h) of rectangular columns, and the diameter (D) for circular
Proposing a dependable design recommendation for the use of FRP columns. Concentrically loaded columns data comprises 47 tests of both
bars in columns requires conducting analytical and reliability assess­ rectangle and circular columns, while eccentrically loaded rectangle
ment using a wide range of experimental data. Accordingly, experi­ columns and circular columns data comprises 95 tests and 52 tests,
mental data of FRP-RC columns under concentric and eccentric loading respectively. In addition, the collected data included both GFRP and
has been collected. The experimental data is divided into three cate­ CFRP reinforced columns.
gories: concentrically loaded columns (Table 1), eccentrically loaded
rectangle column (Table 2), and eccentrically loaded circular columns

Table 3
Eccentrically loaded circular columns data.
Reference No. of specimens Dia, mm f’c, MPa FRP type Ef, GPa ρf % e/h

Hadhood [30] 36 305 35, 70.2 GFRP 54.9 2.18, 3.27 0.08–0.66
Hadhood [31] 8 305 35, 70.3 CFRP 141 2.18 0.08–0.66
Hadi et al. [32] 4 205 37 GFRP 50 2.3 0.08–0.163
Hadi et al. [33] 4 210 85 GFRP 52 2.19 0.12–0.24

Fig. 1. Effect of FRP compression bars contribution in predicting axial capacity of concentrically loaded columns: (a) neglecting FRP bars contribution; (b) including
FRP bars contribution (Note: 1 ksi = 6.895 MPa).
1582
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

Table 4 predicting the capacity of concentrically and eccentrically loaded


Statistics for investigated cases in predicting axial capacity of FRP-reinforced columns.
columns. ( )
Po = 0.85f ’c Ag − Af + 0.002Ef Af (3)
Concentrically loaded columns

Case Axial Capacity Formula Average Pexp/ COV RMSE The effectiveness of including FRP bars compressive contribution to
Pth (%) the nominal axial capacity is studied using the ratio of experimental
1
(
Po = 0.85f’c Ag − Af
)
1.15 9.55 476 loading capacity to the analytical axial loading capacity calculated from
2
( )
1.06 9.06 306
the aforementioned equations (Pexp/Pth). When the ratio is closer to
Po = 0.85f’c Ag − Af + 0.002Ef Af
unity means more accurate prediction of nominal capacity. Fig. 1 pre­
Eccentrically loaded rectangular columns
Case Case Description Average Pexp/ COV RMSE sents the relationship of this ratio with respect to concrete strength, FRP
Pth (%) elastic modulus, and reinforcement ratio if the FRP contribution is
1 Neglect FRP in compression, Subtract 1.08 14.7 165 neglected (Fig. 1 (a)), and if the FRP contribution is included (Fig. 1 (b))
FRP area
in concentrically loaded columns. The ratio (Pexp/Pth) values are more
2 Include FRP, fFRP = εEf ; ε < 0.002 1.01 13.7 154
scattered and higher in Fig. 1 (a) compared with ratio values Fig. 1 (b),
Eccentrically loaded circular columns
Case Case Description Average Pexp/ COV RMSE
which indicates that including FRP contribution in axially loaded col­
Pth (%) umns results in a more accurate prediction. Also, Pexp/Pth values do not
1 Neglect FRP in compression, subtract 1.05 12.8 224 follow a certain trend with respect to different parameters.
FRP area In order to further understand the effect of FRP contribution in
2 Include FRP, fFRP = εEf < 0.002 0.99 10.0 157
nominal axial capacity, different statistics figures are calculated (mean,
Coefficient of Variance (COV), and Root Mean Square Error (RMSE)) for
the three categories: concentrically loaded columns; eccentrically
4. Nominal resistance
loaded rectangular columns; and finally, eccentrically loaded circular
columns. Neglecting FRP compression contribution to the nominal axial
The nominal axial capacity of concentrically loaded short columns
capacity results in higher COV and higher (Pexp/Pth) values. A similar
reinforced with mild steel longitudinal bars can be computed by sum­
conclusion is drawn for eccentrically rectangle and circular columns as
ming the contribution of the concrete and steel in compression. At the
shown in Table 4.
ultimate limit state, the steel bars yield (i.e. εsteel ≥ 0.002) and the col­
The axial capacity of eccentrically loaded rectangular columns can
umn capacity can be computed as shown in Eq. (1). The ACI-440 and
be computed by establishing two basic equilibrium equations: the sum of
CAN/CSA S806 state that the contribution of FRP reinforcement should
vertical forces equals to zero (Eq. (4)), and the sum of moments about
be neglected in compression. Based on this recommendation, the nom­
tension reinforcement equals to zero (Eq. (5)) as illustrated in Fig. 2. The
inal axial capacity of FRP-RC columns can be computed as given in Eq.
solution of these equilibrium equations can be iterative by assuming a
(2) which ignores the FRP contribution completely.
( ) value for the depth of neutral axis ‘c’ that satisfy Eq. (4) and (5),
Po = 0.85f ’c Ag − As + As fy (1)
Po − C c − C s + T = 0 (4)
( )
Po = 0.85f ’c Ag − Af (2) 1[ ( a) ]
Po = ’
Cc d − + Cs (d − d’ ) = 0 (5)
e 2
where, Po is the nominal axial capacity, Ag is the columns’ gross area, Af
is the area of FRP bars, Ef is the elastic modulus of FRP bars. where: Cc is the compression force in concrete (Cc = 0.85f’c ab); Cs is the
( ( ))
In the surveyed literature, several researchers suggested including compression force in compression bars Cs = A’ s f ’ s − 0.85f ’ c ; T is the
the FRP effect in nominal axial capacity through different approaches. tension force in tension bars (T = As fs ); fs is compressive stress in
One approach is to obtain the stress in the FRP bars by assuming a linear ( )
compression bars (f ’ s = ε’ t E); e’ is the applied load moment arm; d − 2a
stress–strain relationship (i.e. σ FRP = Ef × ε) and limiting the strain to
0.002 (Eq. (3)). is the concrete compression force moment arm; (d − d’ ) = bars
In this study, limiting the strain to 0.002 criteria will be utilized in compressive force moment arm (refer to Fig. 2 for variables definitions).

Fig.2. Equilibrium of forces at ultimate limit states in eccentrically loaded columns.

1583
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

Fig. 3. Normalized Pexp and Pth for all surveyed data based on limiting FRP strain to 0.002.

Alternatively, the analysis can be reduced into one cubic equation to ( )


c − d’
be solved directly in the form of Eq. (6), assuming compression- ε’t = 0.003 (7)
controlled failure. The values A, B, C, D are calculated then ‘a’ value c
can be determined and substituted in Eq. (5) to estimate the nominal The analysis of circular columns has been conducted utilizing
axial capacity (refer to Fig. 2 for variables definitions). sectional analysis, where the cross section is divided into layers and
analyzed using strain compatibility. The analysis was conducted using
Aa3 + Ba2 + Ca + D = 0 (6)
Response-2000 software [36].
Where Fig. 3 shows the normalized experimental versus the analytical ca­
pacity for FRP columns under concentric, eccentric loading, and com­
0.85f ’ c b
A= bined data. FRP bars’ contribution has been added with a 0.002 strain
2
limit. There is a good agreement between experimental and analytical
B = 0.85f ’ c b(e’ − d) values with a low COV for all loading cases and a maximum error margin
of 30%.
C = A’ s (f ’ s − 0.85f ’ c )(e’ − d + d’ ) + 0.003EA’ s e’
5. Proposed reliability analysis
D = 0.003EAs e’ β1 d
The level of safety in FRP-reinforced concrete columns is assessed by
The stress in the compression FRP bars can be computed using strain calculating the probability of failure, and the reliability index. The
compatibility as in Eq. (7). In addition, the contribution of the FRP in reliability index represents the number of standard deviations between
compression can be either neglected as recommended by design codes or zero and the mean of the (Resistance – Demand) distribution. The
limited by specific stain value (e.g., 0.002) similar to concentrically probability of failure and the safety index is calculated according to Eq.
loaded columns. In the analysis, the compression elastic modulus of FRP (8) and Eq. (9) for normal distributions. Where Pf is the probability of
is assumed to equal the tensile modulus of elasticity, [34–35] unless the failure, Φ is cumulative distribution function, β is the reliability index,
compressive elastic modulus is reported in the study. However, the re­ μR and μL are the mean of the resistance and load, respectively, σR and σ L
ported compressive elastic modulus in the literature is within a 6% are the standard deviation of the resistance and load, respectively.
difference in the tensile elastic modulus.

1584
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

1) Start with assuming a dead load distribution with a mean of one.


Szerszen et al., [38] reported that dead load exhibit normal distri­
bution with bias and COV of 1.05 and 0.1, respectively. The nominal
live load mean is determined based on the selected live to dead load
(L/D) ratio. Szerszen et al., [38] reported that live load exhibit
extreme type I distribution with bias and COV of 1.00 and 0.18,
respectively. The actual mean equal nominal mean multiplied by the
bias.

Extreme type I distribution is also known as Gumbel distribution.


Extreme type I distribution is used to model the maximum (or the
minimum) of a number of samples of various distributions.

2) The nominal resistance distribution is computed using the LRFD


equation (Eq. (10)). The nominal resistance mean is equal to factored
load (using the controlling load combination) divided by strength
reduction factor similar to steel reinforced columns of (0.65). The
bias and COV of the resistance distribution are calculated from the
surveyed experimental data.

Fig.4. Cumulative distribution function for the resistance model fitted to a


normal distribution.

Pf = Φ(β) (8)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
β = μR − μL / σ 2R + σ2L (9)

The current design practices for concrete accept a reliability index


between 3.0 and 3.5 for ductile failures and between 3.5 and 4 for brittle
and sudden failures [3]. Several reliability assessment methodologies
can be used to calculate β such as First-order Second Moment (FOSM)
method, or Monte Carlo Simulation (MCS).
In this paper a modified methodology of Farrow et al., [39] is
adopted to overcome different shortcomings in Farrow et al. reliability
assessment; which are: 1) assuming normal distribution for the demand
and capacity parameters; 2) considering live load only in the demand
model; 3) small number of simulations (10,000) were performed in the
MCS which might not guarantee convergence, thus it will give inaccu­
rate β values. Accordingly, to overcome these deficiencies this study
utilizes a more realistic distributions load and resistance, considering
live and dead load with a range of live to dead load ratios, consider two
load combinations, and a higher number of simulations (up to 109)
which ensures convergence is performed. The simulations have been
conducted using script loaded supercomputer. Fig. 6. Number of simulations versus the reliability index and probability of
The procedure of the reliability assessment used in this study is failure, the red line refers to the mean value. (For interpretation of the refer­
illustrated as follows: ences to colour in this figure legend, the reader is referred to the web version of
this article.)

Fig. 5. Illustrative diagram for determining loads and resistance distributions (for L/D ratio of 3).

1585
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

Fig.7. Reliability index of FRP reinforced concrete columns versus load ratio with respect to dataset, reinforcement ratio, concrete strength, eccentricity level, and
FRP type.

1586
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

{
1.4PD eccentricity level as the failure mode change from compression-
max ≤ ∅Rn (10)
1.2PD + 1.6PL controlled to tension-controlled [4] while in FRP-reinforced columns
compression-controlled failure only was considered. Similar findings
can be found in Fei Peng et al. [4] reporting no effect of eccentricity level
3) Kolmorgorov-Smirnov (KS) test [37] is performed for the resistance
on the reliability index.
model to examine if the resistance data are normally distributed, by
In addition, the analysis results show some variation in reliability
comparing the maximum absolute error between resistance Cumu­
index with respect to concrete strength. However, the reliability index
lative Distribution Function (CDF) and the CDF values of fitted
exceeds 3.5 and does not show significant change or trend with respect
normal distribution which is referred to as KS with the area of
to concrete strength. On the other hand. Fig. 7e shows the reliability
rejection which is called as KScritical. Fig. 4 shows the cumulative
index for CFRP and GFRP reinforced columns. GFRP-reinforced col­
distribution function for the resistance model fitted to a normal
umns; with FRP elastic modulus generally ranges from 30 to 70 GPa,
distribution for a significance level of 0.05, where KS (0.049) is less
exhibit lower reliability index compared to CFRP reinforced columns;
than KS critical of 0.097, which proves the normality of the resis­
with FRP elastic modulus ranges from 139 to 151 GPa.
tance model.
In summary, the results show an acceptable level of safety for FRP-
reinforced columns under concentric and eccentric loading. In addi­
Measurements of live loads in building codes are generally higher
tion, a consistent level of reliability is maintained for investigated pa­
than what the building actually experiences. For example, building
rameters except for FRP type. To account for the change in reliability
codes specify a live load at the 95 percentile for offices [39]. This means
based on FRP type; different strength reduction factor for GFRP rein­
the design load is higher than or equal to 95 percent of the observed live
forced columns is proposed. Fig. 7f shows reliability index based on FRP
loads. This value corresponds to 1.325 for live load extreme type I dis­
type with strength reduction factor of 0.6 for GFRP reinforced columns
tribution. Fig. 5 presents a simple illustrative diagram of the adopted
to achieve more consistent safety level.
procedure assuming (L/D) ratio of three.
6. Conclusions
4) With three distributions determined, MCS is performed for a wide
range of simulations starting with 100 to 2 billion, to reach the
The study provides design recommendation and reliability analysis
optimal number of simulations. Then the limit state function in Eq.
for FRP-reinforced columns under concentric and eccentric loading, the
(11) is used to find the probability of failure at each simulation.
following conclusions have pertained:
Where R is the ultimate capacity for the column, PD and PL are dead
and live loads, respectively.
1- Including the compressive contribution of FRP bars in columns
G = R − PD − P L (11) provides more accurate capacity predictions compared to ignoring
their contribution. The contribution can be included by limiting the
A negative value for G function represents a failure state while a safe
strain of compression bars to 0.002. This recommendation is valid for
state otherwise. The probability of failure is calculated by dividing the
circular and rectangular columns under eccentric and concentric
frequency for which the G function present failure state by the total
loading. Including the FRP bars contribution resulted in an average
number of simulations.
increase in capacity of 8% for both concentrically and eccentrically
The reliability analysis has been conducted for a range of L/D from
loaded columns and up to 17%.
0 to 4. The live to dead ratio is usually expressed by the term D/(D + L).
2- The reliability analysis for concentrically loaded columns, rectan­
The adopted range of L/D ratio corresponds to D/(D + L) ratio from 0.2
gular columns under eccentric loading, and circular columns under
to 1. The most probable load ratios D/(D + L) for columns are with the
eccentric loading provides reliability index β higher than 3.5 for a
range from 0.4 to 0.9 [40]. Reliability index β is evaluated using Eq. (8).
probable range of load.
3- Reliability index was not sensitive to changes in reinforcement ratio
5) A convergence study has been conducted for each simulation. Fig. 6
or eccentricity level. However, the reliability index exhibited some
presents a sample of the convergence study that shows the rela­
fluctuations with concrete strength but with no apparent trend.
tionship between the number of simulations versus Pf and β, it can be
4- CFRP-reinforced columns (E = 139 to 151 GPa) exhibited higher
noticed that the optimal number of simulations for this sample is 10
reliability index compared to GFRP-reinforced column (E = 30 to 70
million simulations.
GPa). To achieve consistent reliability level, a 0.6 strength reduction
factor can be used for columns with GFRP compared to 0.65 for
Fig. 7a presents the reliability index for each data set as a function of
columns with CFRP bars.
D/(D + L). For the probable range of load, the columns exhibit reliability
index higher than 3.5. It also can be noted how reliability index is
affected by the COV where the data with the lowest COV has a higher Declaration of Competing Interest
reliability index and vice versa. In addition, the figure shows a trend of
decreasing in reliability index with D/(D + L) ratio of 0.88. This change The authors declare that they have no known competing financial
is due to switching in the controlling load combination (Eq. (10)). A interests or personal relationships that could have appeared to influence
similar finding was also observed in steel-RC members [38,40] and FRP- the work reported in this paper.
reinforced members [4].
To investigate the column properties’ effect on the reliability index, References
the reliability analysis discussed above has been conducted after sorting
data based on reinforcement ratio, concrete strength, eccentricity level, [1] Soltani M, Safiey A, Brennan A. A State-of-the-Art Review of Bending and Shear
Behaviors of Corrosion-Damaged Reinforced Concrete Beams. Struct J 2019;116
and FRP type. Fig. 7b, 7c, 7d show the reliability index for different (3).
reinforcement ratios, concrete strengths, and eccentricity levels with [2] De Luca, A.; Matta, F.; and Nanni, A., “Behavior of Full-Scale Glass Fiber-
respect to load ratio. It can be noted from Fig. 7b and Fig. 7d that the Reinforced Polymer Reinforced Concrete Columns under Axial Load,” ACI Struct J,
107, (5), Sept.-Oct. 2010, pp. 589-596.
reliability index was not affected significantly by the reinforcement ratio
[3] Wight JK. Reinforced concrete mechanics and design. 7th edition. Hoboken, New
nor eccentricity level. The results indicate a reliability index exceeds 3.5 Jersey: Pearson Education Inc; 2016. p. 07030.
in all figures for the probable load range for columns. It should be noted [4] Peng F, Xue W. Reliability Analysis of Eccentrically Loaded Concrete Rectangular
Columns Reinforced with Fiber-Reinforced Polymer Bars. ACI Struct J 2019;116
that steel reinforced columns’ reliability is highly affected by the
(4):275–84.

1587
A. Tarawneh and S. Majdalaweyh Structures 28 (2020) 1580–1588

[5] ACI Committee 440, 2015, “Guide for the Design and Construction of Structural [21] Othman, Z.S. Mohammad, A.H., 2019. Behaviour of Eccentric Concrete Columns
Concrete Reinforced with FRP Bars (ACI 440.1R-15),” American Concrete Institute, Reinforced with Carbon Fibre-Reinforced Polymer Bars. Adv Civil Eng; 2019.
Farmington Hills, MI, pp. 83. [22] Gong Y, Zhang J. “Experimental Study of Reinforced Concrete Eccentric
[6] Canadian Standards Association (CSA), 2012, “Design and Construction of Building Compression Columns with CFRP Tendons. China Civ Eng J 2009;42(10):46–52.
Components with Fiber Reinforced Polymers (CAN/CSAS806-12),” CSA Group, https://doi.org/10.15951/j.tmgcxb.2009.10.012. in Chinese.
Rexdale, ON, Canada, pp. 208. [23] Guérin M, Mohamed HM, Benmokrane B, Shield CK, Nanni A. Effect of Glass Fiber-
[7] Kobayashi, K., Fijisaki, T., Compressive behavior of FRP reinforcement in non- Reinforced Polymer Reinforcement Ratio on Axial-Flexural Strength of Reinforced
prestresses concrete members. Proceedings of the Second International RILEM Concrete Columns. ACI Struct J 2018;115(R4):1049–61. https://doi.org/
Symposium on Non-metallic (FRP) Reinforcement for Concrete Structures 10.14359/51701279.
(FRPRCS-2). Ghent, Belgium; 1995: pp. 267-274. [24] Issa MS, Metwally IM, Elzeiny SM. Structural performance of eccentrically loaded
[8] Khorramian K, Sadeghian P. Experimental and Analytical Behavior of Short GFRP reinforced concrete columns. Int J Civil Struct Eng 2011;2(1):395–406.
Concrete Columns Reinforced with GFRP Bars under Eccentric Loading. Eng Struct [25] Khorramian K, Sadeghian P. April. Behavior of Slender GFRP Reinforced Concrete
2017;151. Columns. ASCE-SEI Structures Congress 2019. American Society of Civil Engineers;
[9] Guérin M, Mohamed HM, Benmokrane B, Nanni A, Shield CK. Eccentric Behavior 2019.
of Full-Scale Reinforced Concrete Columns with Glass Fiber-Reinforced Polymer [26] Salah-Eldin A, Mohamed HM, Benmokrane B. Structural performance of high-
Bars and Ties. ACI Struct J 2018;115(2):489–99. https://doi.org/10.14359/ strength-concrete columns reinforced with GFRP bars and ties subjected to
51701107. eccentric loads. Eng Struct 2019;185:286–300.
[10] Hadhood A, Mohamed HM, Benmokrane B. Axial Load-Moment Interaction [27] Sharbatdar MK. Concrete Columns and Beams Reinforced with FRP Bars and Grids
Diagram of Circular Concrete Columns Reinforced with CFRP Bars and Spirals: under Monotonic and Reversed Cyclic Loading. PhD dissertation. Ottawa, ON,
Experimental and Theoretical Investigations. J Compos Construct, ASCE 2017;21 Canada: University of Ottawa; 2003. p. 371.
(2):04016092. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000748. [28] Sun L, Wei M, Zhang N. Experimental Study on the Behavior of GFRP Reinforced
[11] Xue W, Peng F, Fang Z. Behavior and Design of Slender Rectangular Concrete Concrete Columns under Eccentric Axial Load. Constr Build Mater 2017;152:
Columns Longitudinally Reinforced with Fiber-Reinforced Polymer Bars. ACI Struct 214–25. https://doi.org/10.1016/j.conbuildmat.2017.06.159.
J 2018;115(2):311–22. https://doi.org/10.14359/51701131. [29] Tikka, T. K., Francis, M., Teng, B., “Strength of Concrete Beam Columns Reinforced
[12] Afifi MZ, Mohamed HM, Benmokrane B. Axial Capacity of Circular Concrete with GFRP Bars,” 2nd International Structures Specialty Conference, Winnipeg,
Columns Reinforced with GFRP Bars and Spirals. J Compos Constr 2014;18(1): MB, Canada; 2010: pp. 1194-1203.
04013017. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000438. [30] Hadhood, A.A., Behavior, strength and flexural stiffness of circular concrete
[13] Tobbi, Hany, Farghaly, Ahmed Sabry, & Benmokrane, Brahim. Concrete Columns columns reinforced with FRP bars and spirals/hoops under eccentric loading; 2017.
Reinforced Longitudinally and Transversally with Glass Fiber-Reinforced Polymer [31] Hadhood A, Mohamed HM, Benmokrane B, Nanni A, Shield CK. Assessment of
Bars. ACI Struct J ; 2012: 109(4). Doi: 10.14359/51683874. design guidelines of concrete columns reinforced with glass fiber-reinforced
[14] Mohamed HM, Afifi MZ, Benmokrane B. Performance Evaluation of Concrete polymer bars. ACI Struct J 2019;116(4):193–207.
Columns Reinforced Longitudinally with FRP Bars and Confined with FRP Hoops [32] Hadi MN, Karim H, Sheikh MN. Experimental investigations on circular concrete
and Spirals under Axial Load. J Bridge Eng 2014;19(7):04014020. https://doi.org/ columns reinforced with GFRP bars and helices under different loading conditions.
10.1061/(ASCE)BE.1943-5592.0000590. J Compos Constr 2016;20(4).
[15] Choo, C.C., Harik, I. E., Gesund, H., Strength of Rectangular Concrete Columns [33] Hadi MN, Hasan HA, Sheikh MN. Experimental investigation of circular high-
Reinforced with Fiber-Reinforced Polymer Bars. ACI Struct J, 103 (3), May-June; strength concrete columns reinforced with glass fiber-reinforced polymer bars and
2006: p. 452-459. helices under different loading conditions. J Compos Constr 2017;21(4):04017005.
[16] De Luca, A., Matta, F., Nanni, A., Behavior of Full-Scale Glass Fiber-Reinforced [34] Deitz D, Harik I, Gesund H. Physical Properties of Glass Fiber Reinforced Polymer
Polymer Reinforced Concrete Columns under Axial Load. ACI Struct J, 107, (5), Rebars in Compression. J Compos Construct ASCE 2003;7(4):363–6. https://doi.
Sept.-Oct; 2010, pp. 589-596. org/10.1061/(ASCE)1090-0268(2003)7:4(363).
[17] Elchalakani M, Ma G. Tests of Glass Fibre Reinforced Polymer Rectangular [35] Tavassoli A, Liu J, Sheikh S. Glass Fiber-Reinforced Polymer-Reinforced Circular
Concrete Columns Subjected to Concentric and Eccentric Axial Loading. Eng Struct Columns under Simulated Seismic Loads. ACI Struct J 2015;112(1):103–14.
2017;151:93–104. https://doi.org/10.1016/j.engstruct.2017.08.023. https://doi.org/10.14359/51687227.
[18] Elchalakani, M., Karrech, A., Dong, M., Ali, M., Yang, B., Experiments and Finite [36] Bentz EC. Sectional Analysis of Reinforced Concrete Members. PhD thesis. Canada:
Element Analysis of GFRP Reinforced Geopolymer Concrete Rectangular Columns Department of Civil Engineering, University of Toronto, Toronto, ON; 2000.
Subjected to Concentric and Eccentric Axial Loading. Structures, 14; 2018: pp. 273- [37] Farrow, B., Frigui, I., Klingner, R., Tensile Capacity of Single Anchors in Concrete:
289. doi: 10.1016/j.istruc.2018.04.001. Evaluation of Existing Formulas on an LRFD Basis. ACI Struct J 1996; 93: 1, p. 128-
[19] Hadhood A, Mohamed HM, Benmokrane B. Experimental study of circular high- 137.
strength concrete columns reinforced with GFRP bars and spirals under concentric [38] Szerszen MM, Nowak AS. Calibration of Design Code for Buildings (ACI 318): Part
and eccentric loading. J Compos Constr 2017;21(2). 2—Reliability Analysis and Resistance Factors. ACI Struct J 2003;100(3):383–91.
[20] Hadi MN, Youssef J. Experimental investigation of GFRP-reinforced and GFRP- [39] Kolmogorov A. Sulla determinazione empirica di una lgge di distribuzione. Inst Ital
encased square concrete specimens under axial and eccentric load, and four-point Attuari Giorn 1933;4. pp. 83–91.42.
bending test. J Compos Constr 2016;20(5). [40] Szerszen MM, Szwed A, Nowak AS. Reliability Analysis for Eccentrically Loaded
Columns. ACI Struct J 2005;102(5):676–88.

1588

You might also like