Professional Documents
Culture Documents
This paper develops reliability-based design provisions for eccen- et al. 2010; Afifi et al. 2014) and eccentric loading (Hadhood
trically loaded rectangular concrete short columns reinforced with et al. 2017, 2018; Elchalakani and Ma 2017; Guérin et al.
fiber-reinforced polymer (FRP) bars. First, closed-form solutions 2018a,b; Xue et al. 2018). It was concluded that the compres-
for the axial force-moment strength interaction diagram were sion behavior of FRP-RC columns was similar to that of
presented. Subsequently, the statistical parameters of the resistance
conventional steel-RC columns, but with less contribution of
model were estimated based on available experimental data and
FRP longitudinal bars to load capacity. The FRP-RC rectan-
Monte Carlo simulations. The first-order second-moment (FOSM)
method was then applied to calibrate strength reduction factors gular columns under eccentric loading were not triggered by
to meet the uniform target reliability level βT = 4.0. To verify the rupture of the FRP bars in the tension side, but rather
results obtained, a comparison between reliability indexes obtained attributed to concrete crushing on the compression side,
from Monte Carlo simulations and those from the FOSM was even for the column with a low longitudinal reinforcement
conducted. Finally, sensitivity analyses were performed to evaluate ratio (1%) and large eccentricity (e/h ≥ 0.7) (Gong and
the influence of various design parameters on the reliability index, Zhang 2009; Sun et al. 2017; Xue et al. 2018; Guérin et al.
indicating that the reliability of FRP-reinforced concrete columns 2018a,b). Recently, valuable research works have been
is dependent on the axial reinforcement stiffness Efρf. As a result, conducted to evaluate seismic performance of FRP-RC
this study recommended strength reduction factors of 0.60 for the columns. Test results indicated that properly designed and
columns with Efρf ≤ 2 GPa (290 ksi), 0.65 for those with Efρf >
detailed FRP-RC columns could reach high deformation
4 GPa (580 ksi), and a linear variation between the two.
levels with no strength degradation (Elshamandy et al.
Keywords: eccentricity; fiber-reinforced polymer (FRP); rectangular 2018). The drift capacity of the columns exceeded the limita-
column; reinforced concrete; reliability; strength reduction factor. tions of North American building codes (Tavassoli et al.
2015; Ali and El-Salakawy 2016; Elshamandy et al. 2018).
INTRODUCTION In addition to the experimental studies, valuable theoret-
Corrosion of steel bars is a major concern in reinforced ical approaches have been developed to calculate the nominal
concrete (RC) structures, particularly in harsh, corrosive, axial force-bending moment (P-M) interaction diagrams of
coastal environments (Nanni et al. 2014). As a material with FRP-RC rectangular short columns (Zadeh and Nanni 2013a;
excellent corrosion resistance, fiber-reinforced polymer Choo et al. 2006; Xue et al. 2018; Guérin et al. 2018b). These
(FRP) has emerged as a viable alternative to steel in RC approaches are based on the fundamentals of equilibrium of
structures. In addition to corrosion resistance, FRP materials forces, compatibility of strains, and constitutive equations,
possess characteristics that include a high strength-weight except the difference in the manner that they account for the
ratio, outstanding fatigue resistance, nonmagnetic conduc- contribution of FRP longitudinal bar in compression. Choo
tance, lower elastic modulus compared to steel, and a linear et al. (2006) developed an analytical approach to examine
stress-strain relationship. In practice, widely used fibers the P-M interaction behavior of FRP-RC rectangular
include glass, carbon, aramid and basalt, and bars using columns by using different ratios of compressive to tension
them are termed glass fiber-reinforced polymer (GFRP), modulus. According to their analysis, FRP-RC column
carbon fiber-reinforced polymer (CFRP), aramid fiber-rein- cross section interaction strength diagrams do not exhibit
forced polymer (AFRP), and basalt fiber-reinforced polymer balance points for the columns bound by the reinforcement
(BFRP) bars, respectively. Over the last three decades, the limits [(ρmin = 1%) ≤ ρ ≤ (ρmax = 8%)]. Zadeh and Nanni
flexural and shear behavior of FRP-RC members have been (2013a) developed an ultimate strength approach to calculate
extensively investigated, and the corresponding design the P-M interaction diagrams, neglecting the contribution of
guidelines have been well established to assist engineers in the FRP bars in compression to the load capacity. Xue et al.
the design and application of FRP bars for flexural members. (2018) presented an analytical approach for FRP-RC cross
As a result, FRP bars are widely used in bridge decks and section strength, in which FRP bars in compression were
beam elements (Benmokrane et al. 2016). Nonetheless, there
is little experience in the use of FRP bars in RC columns due ACI Structural Journal, V. 116, No. 4, July 2019.
to a lack of relevant design guidelines. MS No. S-2018-331, doi: 10.14359/5171563, was received August 3, 2018, and
reviewed under Institute publication policies. Copyright © 2019, American Concrete
In recent years, a considerable number of experimental Institute. All rights reserved, including the making of copies unless permission is
research efforts have focused on the behavior of FRP-RC obtained from the copyright proprietors. Pertinent discussion including author’s
closure, if any, will be published ten months from this journal’s date if the discussion
columns under concentric (Tobbi et al. 2012, 2014; De Luca is received within four months of the paper’s print publication.
considered assuming the FRP tensile and the compressive RESEARCH SIGNIFICANCE
elastic modulus were equal. Several researchers (Hadhood Currently, ACI 440.1R-15 (ACI Committee 440 2015)
et al. 2017; Guérin et al. 2018a,b) compared experimental does not provide guidance for designing concrete sections
and analytical resistances and concluded that considering reinforced with FRP bars under combined axial load and
the compression contribution of FRP bars provided accurate bending moment. To the authors’ knowledge, the strength
predictions of the experimental P-M interaction diagrams, reduction factors for the FRP-RC columns have not yet
while neglecting their contribution added to the level of been addressed in available technical literature. This paper
conservativeness. It should be noted that the aforementioned will conduct a rigorous reliability analysis and calibrate the
approaches proposed for the design of FRP-RC columns are strength reduction factors for eccentrically loaded FRP-RC
based on a deterministic point of view. A reliability-based short columns. The presented research will assist engineers
design provision of FRP-RC columns, however, is not in the design and application of FRP bars in RC columns.
available.
Currently, the load and resistance factor design (LRFD) LOAD MODEL
methodology is widely accepted among researchers and The load model specified in ACI 318-14 (ACI Committee
practicing engineers. The underlying principle of LRFD is 318 2014) was used in the reliability analysis. The following
to achieve a certain target reliability level by calibrating factored load U was considered as the basic load combina-
load and resistance factors, which accounts for probable tion of the ultimate limit state
overloading and understrength of RC structural members.
Because FRP possesses different mechanical properties from 1.2 D + 1.6 L
steel, the design philosophy developed for RC members U = max (1)
1.4 D
with steel bars is not necessarily applicable to those with
FRP bars (Pilakoutas et al. 2002). In the last two decades, where D is the dead load; and L is the live load.
valuable research efforts have been made to calibrate the It should be noted that various loads can exist in the
resistance factors for flexural members with FRP (Pilakoutas load combinations accounted for in ACI 318. In this study,
et al. 2002; Shield et al. 2011; Zadeh and Nanni 2013b; Kim however, only the loads in Eq. (1) were considered to reduce
and Nickle 2016; Peng and Xue 2019). Nowadays, several complexity of the reliability analyses. According to Nowak
LRFD-based guidelines and codes have been drafted to and Collins (2012), dead load follows normal distribution with
assist engineers in the design and application of FRP bars for a bias factor and COV of 1.05 and 0.10, respectively, while
flexural members (ACI Committee 440 2015; CSA 2012; fib the live load follows Extreme Type I distribution with a bias
Task Group 9.3 2007). However, there is no research to cali- factor and COV of 1.00 and 0.18, respectively. Table 1 lists
brate the resistance factors for FRP-RC columns, and design the selection of bias, COV, and type of distribution function.
provisions of the columns under combinations of axial
compression and bending moment have not been addressed RESISTANCE MODEL
in current design guidelines and codes. Because information on the variability of the resistance of
This study, therefore, presents a reliability analysis for FRP-RC columns is not available, it is necessary to develop
FRP-RC rectangular short columns under eccentric loading. resistance models using available test data and numerical
First, the uncertainty in the resistance model is evaluated simulation. Three sources of uncertainty affect the vari-
based on a large experimental database and Monte-Carlo ability of resistance—namely, material variability, M; vari-
simulation. Subsequently, two reliability approaches— ability in fabrication tolerances, F; and analysis factor, P. In
namely, first-order second-moment method and Monte- reliability analysis, the random variable R for the resistance
Carlo simulation—are used to calculate the reliability can be considered as a product of the nominal resistance Rn
indexes. Sensitivity analyses are then performed to examine and three parameters that account for the sources of uncer-
the influence of various design parameters on the reliability tainty mentioned previously (Nowak and Collins 2012)
index. Finally, design provisions for eccentrically loaded RC
columns with FRP bars are proposed. R = ψMψFψPRN (2)
Fig. 1—Experimental versus predicted resistance: (a) including contribution of FRP bars in compression; and (b) neglecting
contribution of FRP bars in compression.
The proposed methodology, however, gave more accurate λ f ' = −2.47 × 10−5 ( f c' )3 + 3.17 × 10−3 ( f c' )2 − 1.35 × 10−1 f c' + 3.0649 ≥ 1.15
predictions of the experimental P-M interaction diagrams in
c
terms of model bias and COV of the ratio between the exper- (fc′ in MPa) (10)
imental and the predicted values. The bias λP and COV VP
for the ratio of measured-to-predicted resistances obtained As given in Eq. (8), the variation in the ultimate compres-
from the proposed methodology are 1.03 and 0.125, respec- sive strain in concrete, εcu, may have a significant effect on
tively, whereas the corresponding value from Zadeh and the stress in FRP and consequently the prediction of resis-
Nanni methodology are 1.17 and 0.128, respectively. tance of the columns. In this study, therefore, the εcu was
treated as a random variable. Based on an extensive exper-
Statistical parameters of material properties and imental database, Baji and Ronagh (2016) found that the
geometry uncertainty lognormal distribution was the best-fit probability density
The statistical properties of concrete compressive function for representing the probabilistic distribution of
strength, fc′, are based on the model proposed by Nowak and εcu. The mean and COV of the εcu for the best-fit lognormal
Szerszen (2003). This model covers a wide range of concrete distribution were 0.0034 and 0.15, respectively.
compressive strengths. In this model, the normal distribu- The statistical characterization of FRP properties has
tion was used for representing the probability distribution of been reported in literature (Pilakoutas et al. 2002; Shield
concrete compressive strength. A COV of 10% was adopted et al. 2011; Kim and Nickle 2016). Generally, the ultimate
for concrete compressive strength, fc′, and the bias factor, strength of FRP bar (ffu) can be modeled by Weibull distri-
λfc′, was evaluated as follows bution. In this study, a bias factor of 1.18 and a COV of
0.12 were adopted for the ultimate strength of the FRP bar.
The modulus of elasticity can be modeled using lognormal
β = μZ/σZ (12)
Fig. 3—Bias and COV of resistance as function of normalized eccentricity, e/h: (a) fc′ =30 MPa; (b) fc′ = 65 MPa.