Construction and Building Materials
Construction and Building Materials
Construction and Building Materials
h i g h l i g h t s g r a p h i c a l a b s t r a c t
a r t i c l e i n f o a b s t r a c t
Article history: In this research, the mechanical properties of glass and carbon fiber reinforced polymer (FRP) bars with
Received 19 May 2018 epoxy resin matrices embedded in concrete were investigated under an extensive range of elevated tem-
Received in revised form 12 March 2019 peratures (i.e., 25–800 °C). Embedded FRP bars with various bar diameters were studied in order to deter-
Accepted 18 March 2019
mine bar diameter influence on the results. In addition, analysis of variance (ANOVA) was performed on
the experimental results to investigate the contribution of exposure temperature and bar diameter to the
tensile behavior of embedded in concrete FRP bars at elevated temperatures. The results show that the
Keywords:
tensile strength of embedded FRP bars generally decreases with increasing temperature; however, the
FRP bars
Fiber reinforced polymers
rate of decrease of the tensile strength varies within different temperature ranges due to different failure
Elevated temperatures modes and states of the FRP material. The influence of the bar diameter was not significant in the tensile
Concrete sleeve behavior of embedded FRP bars at various elevated temperatures. However, the influence of the bar
Concrete cover diameter increased within the temperature range of 150–500 °C compared to the temperatures lower
GFRP than 150 °C and above 500 °C. The results also show that the concrete cover prevented direct heat and
CFRP oxygen from reaching the bars; as a result, embedded FRP exhibited improved tensile performance at ele-
vated temperatures compared to bare bars that were directly exposed to heat. Two Bayesian regression
models were developed for predicting the tensile performance of embedded GFRP and CFRP bars at ele-
vated temperatures. The models have been shown to be in good agreement with the experimental results.
Ó 2019 Elsevier Ltd. All rights reserved.
⇑ Corresponding author.
E-mail address: vatani@sru.ac.ir (A.V. Oskouei).
https://doi.org/10.1016/j.conbuildmat.2019.03.239
0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1139
1. Introduction As it is evident from the above summary, FRP bars are vulnera-
ble to fire conditions, and they lose their mechanical properties in
The advent of reinforced concrete (RC) structures incorporating such conditions to a significant degree. There are several promising
FRP bars has been a significant advancement in the construction solution for reducing fire vulnerability of exposed FRP materials,
industry. However, concrete is known to be very weak in tension such as covering FRP materials in shotcrete, mortar, fire resistant
and cracks when subjected to local tensile stresses [1]. Therefore, surface coatings [31–35], adding nanoparticles to resin [36], mod-
to overcome this critical issue concrete is often reinforced with ification of resin matrix using micronized fire retardant additives
reinforcing bars. Steel reinforcing bars are the traditional reinforc- [37,38], adding ceramic thermal isolators on the FRP laminates
ing material used in RC structures. However, corrosion of conven- [39]. In this regard, in real life applications, FRP bars are not
tional steel bars is one of the main reasons for the strength exposed directly to open flames and abundant oxygen during fires
degradation and loss of RC structures and their reduced service life as they are embedded in concrete covers. However, in most of the
[2–5]. Owing to their high corrosion resistance, the use of fiber studies available in the literature on the behavior of FRP bars at
reinforced polymer (FRP) bars instead of steel bars offers an attrac- elevated temperatures, the bars are placed in furnaces, and they
tive solution to this problem [4,6–10]. FRP materials possess desir- were burned to char at high elevated temperatures due to open
able properties such high stiffness and high strength-to-weight flames and abundant oxygen [5,25,29,40]. This ignition produces
ratio, and ease of transportation and handling. The broad applica- some extra heat, which magnified the reduction in the mechanical
tions of FRP materials in civil engineering lead to several building properties of the FRP material. Hence, to accurately assess the ten-
codes (CAN/CSA 806-02 [11]; ACI 440.1R-06 [12]; CNR-DT203 sile behavior of FRP bars at elevated temperatures as will be expe-
[13]). In this regard, few provisions, design guidelines, and formu- rienced in real life conditions, it is essential to expose the bars to
lations are available to design or assess the behavior of concrete elevated temperatures when they are embedded in concrete cov-
structures reinforced with FRP bars when RC members reinforced ers. Currently, there are no studies available on the behavior of
with them are exposed to open flames or elevated temperatures. FRP bars embedded in concrete sleeves at elevated temperatures.
Some studies have been conducted to date to assess the behavior To study the FRP bars in such condition the concrete sleeves that
of concrete structures reinforced with FRP bars at elevated temper- were previously used by Robert et al. [41] and Chen et al. [6] can
atures [14–17]. be used.
The available literature on the fire behavior of concrete struc- In this research, the tensile performance of FRP bars embedded
tures reinforced with FRP bars can be divided into three main cat- in concrete sleeves that were subjected at elevated temperatures
egories. First, the studies on the behavior of concrete slabs, beams, was assessed for the first time. This testing procedure allows the
and columns reinforced with FRP bars which were tested at fire bars to be tested at high temperatures up to 800 °C because the
exposure conditions [1,15,18,19]. Second, the investigations on covers prevent the ignition of the epoxy resin. Thus, the designated
the temperature profiles and temperature distribution models of experimental temperatures for the current study laid in the exten-
concrete members that are exposed to standard fire conditions, sive range of 25–800 °C. In addition to experimental studies, an
such as ISO834 or elevated temperatures [20–23]. Based on these ANOVA analysis was performed on the tensile test results to assess
studies, the strength of concrete and FRP bars are calculated the effects of different variables on the tensile behavior of bars.
according to the defined temperature profiles and models in the Finally, a Bayesian regression was applied to the experimental
concrete section; thus, the strength of the concrete member can results, and two predictive models for the behavior of GFRP and
be calculated when it is exposed to fire. Third, the studies on the CFRP bars at elevated temperatures were conducted.
tensile strength of FRP bars at different elevated temperatures
[5,24–27]. This study fits into the last category; therefore, the
2. Experimental program
existing studies on the behavior of FRP bars at elevated tempera-
tures are summarized in this section.
To perform tensile tests on FRP bars and sheets, they were
Hamad et al. [28] reported 55% and 30% loss in tensile strength,
gripped in steel pipes and plates at both ends [5,42]. The tensile
and elastic modulus of glass FRP (GFRP) bars tested at 325 °C,
forces were transferred to the bars using steel bars or plates to
respectively. Wang et al. [24] conducted a study on GFRP and car-
compensate for local failure, slippage, and low gripping capacity
bon FRP (CFRP) bars at elevated temperatures and reported 45%
of the bars. The details of the experimental procedure including
and 35% loss in their initial tensile strength, respectively, when
materials, specimens, conditioning of specimens (i.e., exposure to
they were tested in a temperature of 350 °C. Robert and Ben-
elevated temperatures), and tensile test procedure are explained
mokrane [29] studied the effects of different elevated temperature
in this section in detail.
ranges on the mechanical properties of FRP bars. They reported a
severe strength degradation at glass transition temperature, Tg
(i.e. a temperature where the resin changes from a glassy state to 2.1. The properties of materials
a rubbery one) and resin decomposition temperature, Td (i.e. a tem-
perature where the chemical bonds, modular chains of the resin, 2.1.1. FRP bars
and the bond between resin/fibers may break). Eventually, they In this study two types of FRP bars were used, namely: the sand
reported that at temperatures higher than Td, the resin ignition coated GFRP bars with nominal diameters of 4, 6, 8 and 10 mm
would occur if sufficient ambient oxygen is available. When resin (Type A) and helically wrapped CFRP bars with nominal diameters
ignites it produces toxic fumes heat and no longer bears any stress. of 4 and 5 mm (Type B) (Fig. 1). The mechanical properties of these
However, it is stated in this study that after resin ignition, fibers bars, as provided by the manufacturers, are listed in Table 1.
can still carry the load in their longitudinal direction. Fibers lose
their longitudinal direction load bearing capacity when they reach 2.1.2. Concrete
their melting temperatures (e.g., about 1000 °C for GFRP and about The concrete cover of the specimens were manufactured with
2000 °C for CFRP [30]). Ashrafi et al. [5] investigated the behavior 730, 1430, 270 kg/m3 of sand (0–4 mm), gravel (4–10 mm) and
of GFRP and CFRP bars at elevated temperatures and reported Portland I cement, respectively, using water to cement ratio of
71% and 65% loss in the ultimate stress of GFRP and CFRP bars at 0.4. The sieve analysis of the concrete is listed in Table 2. Three
450 °C, respectively. identical 150 300 mm cylindrical specimens were manufactured
1140 E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152
Table 1
Mechanical properties of bars.
2.3. Conditioning of specimens
Property Bar Type The specimens were tested at the room temperature of 25 °C
GFRP CFRP (control specimens) and elevated temperatures of 80, 150, 300,
Resin type Epoxy Epoxy 400, 500, 600, 700, and 800 °C. The temperatures are chosen in a
Fiber volume (%) 75 75 way to simulate short-term fire conditions of FRP bars embedded
Characteristic tensile strength (MPa) 1200 2100 in concrete from a low to moderate level of elevated temperatures
Modulus of elasticity by manufacturer (GPa) 50 150
(150–400 °C) to high elevated temperatures (400–800 °C). A
Modulus of elasticity by measurements (GPa) 53.1 147.4
Strain at failure (%) 2.4 1.4 steady-state temperature regime was used in this study. The spec-
Density (g/cm3) 2.2 1.65 imens were tested 20 min after the time when the temperature at
Glass transition temperature (Tg) 110 110 the bar/concrete interface reached the target temperature to obtain
a uniform temperature distribution across the bar cross-section
[5].
Table 2
Sieve analysis.
2.4. Tensile test methodology
ASTM E11 Sieves Remained Weight (kg) Remained Weight (%)
size (mm) Alternate The tests were conducted with a 150 kN capacity servo electric
11.2 7/16 in 0 0 universal testing machine that comprises a standard three-zone
9.5 3/8 in 2.71 27.1 furnace in accordance with ASTM D7205M recommendations
8 5/16 in 1.45 14.5 [43] (Fig. 4). The thickness of concrete cover was not chosen as a
6.3 1/4 in 1.96 19.6 test parameter because the specimens were tested when the tem-
4 No.5 0.5 5
2.8 No.7 2.6 26
perature of the bars reached the furnace temperature. The concrete
2 No.10 0.32 3.2 covers only prevented direct heat and oxygen from reaching the
1 No.18 0.2 2 bars. To obtain the tensile elastic modulus of bars, the bare bars
Tray Tray 0.26 2.6 were tested using an extensometer (Fig. 5). The tensile tests were
performed in a displacement control manner at a low rate of
2 mm/min to obtain the post-peak tensile behavior of the bars
and tested at 28 days resulting, which gave an average concrete [44,45]. Fig. 6 illustrates the placement procedure of specimens
compressive strength of 21.2 MPa or splitting tensile strength of in the test device, as well as the placement of embedded thermo-
1.92 MPa. The elastic modulus of this concrete was measured to couple inside the specimens at the bar/concrete interface.
be 20 GPa.
3. Test results and discussions
2.1.3. High-strength adhesive
The steel pipes with grooved inside surfaces were anchored to This section presents the results of the mechanical tests at ele-
both ends of the bars using a high strength adhesive with vated temperatures. The results include the ultimate tensile
36 MPa shear strength (Fig. 2). These pipes were anchored to FRP strength of FRP bars, different failure modes of FRP bars, changes
bars to compensate for the low gripping capacity of these bars in the appearance of bars inside the concrete sleeve, and the
and to prevent slippage and local failure of bars during tensile changes in the appearance of concrete sleeve under elevated
tests. temperatures.
FRP bars of 800 mm length were prepared for the tensile tests. The results of the tensile tests conducted on GFRP bars are listed
The bars were anchored to 220 mm grooved steel pipes in both in Table 3. As expected, the results show that the average tensile
end, and they were placed in a 30 30 330 mm concrete sleeve strength of GFRP bars decreased with increasing test temperature.
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1141
The results also show that the failure mode of the bars changed
(five different failure modes) with a change in the exposure tem- temperatures. The first region is dedicated to the specimens tested
perature. The force–displacement curves of a selected set of GFRP at temperatures lower than 80 °C. In this region, the tensile
bars are presented in Fig. 7. strength loss of bars was lower than 4% due to unchanged molec-
The force–displacement curve of FRP bars typically comprised a ular chain mobility of the resin matrix [29]. When the temperature
long linear deformation phase followed by a very short nonlinear exceeded 80 °C, the bars reached their resin transition temperature
phase before the complete failure (Fig. 7). It is worth noting, how- (Tg) where the resin changes from a glassy state to a rubbery state
ever, that for bars that were tested at temperatures higher than and load transfer capacity of fiber/resin bonds decreases. Thus, in
500 °C the linear deformation phase was not entirely linear and the second region ranging from 80 °C to 150 °C, a sudden drop
it exhibited a curve-like trend. This deviation from the linear trend was observed in the tensile load bearing capacity of GFRP bars.
was because the fibers were reaching their melting point (i.e., The third region comprises the specimens tested at temperatures
around 880 °C for glass fibers [46]). ranging from 150 °C to 500 °C. According to the studies on FRP bars
The average tensile strength of the GFRP bars and its retention at elevated temperatures, bars that were tested at this temperature
are depicted in Fig. 8. The strength retention curves of GFRP bars range produced a large amount of toxic gas, heat, and fumes due to
can be separated into five main regions based on in the exposure oxidation and charring of the resin matrix [5,20]. Consequently, the
1142 E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152
Table 3
Tensile test results of GFRP bars.
T: Temperature, ruave: Average Ultimate Strength, R: Strength retention, COV: Coefficient of Variation, F: Failure mode.
generated heat due to resin ignition adds to the furnace heat and position of the bars was observed in by absorbing some heat. It is
reduces the tensile performance of the bars to a greater degree. worth mentioning that this is not, in fact, the increase of decompo-
In this region, the bars reached their decomposition temperature sition temperature. This is just the increase in the temperature that
(Td) at temperatures higher than 300 °C [5]. However, based on we observed in the decomposition of the FRP bars. As an example, a
the observations, the bars tested in this study did not produce 330 °C decomposition temperature was observed for CFRP bars in
any visible or smellable toxic gases, heat, and fumes due to the the study by Ashrafi et al. while a 400 °C decomposition for the
concrete sleeve that prevented direct flames and oxygen reaching same bar was observed in this study. The decomposition of the
the bars. Consequently, the FRP bars embedded in concrete showed resin matrix is dependent on the resin type and is not related to
better tensile performances in the absence of ignition and charring. water vapor or other similar factors. In the third region, the aver-
In addition, some water steam was produced of the concrete age tensile strength of the GFRP bars decreased less in comparison
sleeve; water vapor may increase the temperature that the decom- to the second region. The fourth region includes the specimens
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1143
Fig. 8. Tensile strength and strength retention versus test temperatures of GFRP bars.
tested at temperatures between 500 and 700 °C. In this region, the of the fibers turned to glossy white (Fig. 10). In this phase, the color
load carrying capacity of the resin was totally lost, and the forces of the bars and concrete sleeves became molten red during the
were carried solely by the fibers. In this region, the strength degra- tests due to their exposure to an extremely high temperature of
dation was almost negligible owing to the excellent thermal resis- 800 °C (Fig. 10), which led to melting of the glass fibers.
tance of the fibers. However, in the fifth region, at temperatures In general, FRP bars with larger diameters possess lower ulti-
higher than 700 °C, the longitudinal load-bearing capacity of the mate tensile strength than the bars with smaller diameters
fibers was lost entirely because they melted. As a result, a total because the resin of the larger bars is not able to transfer shear
strength degradation of GFRP bars happened in the last region. loads to their core fibers [5]. However, the difference between
These findings on the tensile strength degradation trend of FRP the tensile strength of larger and smaller FRP bars decreases at ele-
materials with increasing temperature from ambient temperature vated temperatures. The reason to that is the strength of larger
to high elevated temperature are in agreement with previous stud- bars decreases less than smaller bars at elevated temperatures
ies [5,24,25,29]. Ashrafi et al. [5] tested GFRP bars with the same because their core is less affected by heat.
mechanical properties to those used in the current study and
reported that the strength retention curves of these bars comprised 3.2. CFRP bars
four different zones when the test temperatures ranged from 25 °C
to 450 °C. They reported strength retention of 98%, 79%, 63%, and The tensile test results of CFRP bars tested in this study are tab-
40% for a 10-mm GFRP bar tested at 90, 150, 300, and 450 °C, ulated in Table 4. The typical tensile load–displacement curves for
respectively. CFRP bars are shown in Fig. 11. As it was the case for GFRP bars, the
As can be seen in Fig. 9, the bars experienced concentrated fail- ultimate tensile strength of CFRP bars decreased with increasing
ure with the formation of transverse cracks in concrete sleeves. The temperature. The tensile force–displacement curves of CFRP bars
test was continued after the failure of the bars to expose the fibers consisted of a long linear and a short non-linear phase similar to
by widening the crack width. The failure of GFRP bars can be sep- those of GFRP bars. However, the linear phase exhibited a curve
arated into five different modes, as shown in Fig. 9a-e. These failure similar to a trend for the bars tested at temperatures over 500 °C
modes are outlined in Table 3 (i.e., Type a-e). The first failure mode (Fig. 11). This trend was less than that seen in GFRP bars because
(Type a) account for the brittle failure of bars that were tested at of the higher melting point of the carbon fibers (i.e., approximately
temperatures below 150 °C, in which the fibers are not separated. 1600 °C [46]), which is significantly higher than the maximum test
The second failure mode (Type b) happened in the specimens temperature applied in this study.
tested at temperatures in the range of 150–300 °C. The fibers were The tensile strength and strength retention curves of CFRP bars
separated entirely in this failure mode because the bars were tested at different temperatures are shown in Fig. 12. Similar to the
tested at a temperature in which the resin was in a rubbery state GFRP bars, the tensile strength retention curves of CFRP bars can be
(i.e., exceeding Tg). When the bars were tested at a temperature separated into different regions based on the behavior of bars. In
of 300–500 °C, the failure Mode c occurred. In this failure mode, the case of CFRP bars, there were four distinct regions. The first
the fibers were separated, and their color changed from a milky region is dedicated to the specimens tested at temperatures lower
color to a light brown color, indicating that the fibers started to than 80 °C. In this region the molecular chain mobility of the resin
oxide in this [47]. The failure mode Type d occurred in the speci- matrix does not change; thus, no significant tensile strength degra-
mens that were tested at temperatures between 500 °C and dation was observed. In the second region ranging from 80 °C to
700 °C. In this failure mode, the tensile strength degradation was 150 °C sudden tensile strength degradation was observed because
due to the fiber decomposition only, and the color of the fibers the temperature exceeded the transition temperature of the resin
changed to a very dark brown color. The final failure mode, Type matrix causing the fiber/resin bond to degrade (Fig. 13-b). In the
e, occurred in specimens that were tested at 800 °C. In this failure third region (150–500 °C) the rate of strength degradation was less
mode the fibers melted and the axial load bearing capacity of all than that in the second region. In this region, the tensile strength of
the GFRP bars became almost zero. However, the bars with larger the bars decreases due to the increase in fiber/resin debonding
diameters retained about 8% of their tensile strength due to the while the longitudinal load-bearing capacity of the fibers remained
partial survival of their inner fibers that were less exposed to high unchanged. The final region includes the CFRP bars tested at a tem-
temperatures. The color of the fibers changed to very dark black perature above 500 °C. In this region, only the fibers carried the
due to high decomposition, while in the melting zone the color tensile loads, and hence the changes in the tensile behavior of CFRP
1144 E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152
Table 4
The tensile test results of CFRP bars.
Temperature
(C) CFRP 4 mm CFRP 5 mm F
ruave (MPa) COV (%) R (%) ruave (MPa) COV (%) R (%)
25 2015 1.0 100 1945 0.8 100 a
80 2012 0.8 99 1915 0.5 98 a
150 1569 2.5 77 1495 2.4 76 a-b
300 1280 1.1 63 1284 1.6 66 b
400 986 5.4 48 959 3.3 49 b-c
500 650 3.1 32 702 4.1 36 c
600 503 6.0 24 681 3.5 35 c
700 485 1.4 24 630 6.5 32 d-c
800 459 4.3 22 553 7.6 28 d
Fig. 12. Tensile strength and strength retention versus test temperatures of CFRP bars.
elevated temperatures. However, as can be seen in Fig. 15, the transverse cracks that occurred in the concrete sleeve at extremely
delamination of the resin and outer sand coating of the FRP bars high temperatures (above 700 °C), a plethora of small alligator
is inevitable at temperatures over 150 °C and occurs even if no cracks was formed due to thermal loading on the surface of con-
cracks form in the concrete cover (Fig. 16). As noted previously, crete sleeves in specimens tested at temperatures above 500 °C
the concrete sleeve prevented direct heat and oxygen reaching (Fig. 18). However, these cracks were superficial and not deep
the FRP bars. However, due to the high thermal transverse expan- enough to transfer direct heat and oxygen to the FRP bars embed-
sion of CFRP and GFRP bars, wide longitudinal cracks formed in the ded in concrete. Therefore, they do not affect the results.
concrete sleeve of three of the 10-mm GFRP and one of the 5-mm
CFRP bars, which resulted in the bars burning to char (Fig. 17). This 4. Comparison with previous studies
phenomenon is attributable to the thin concrete cover (15 mm)
used in this study, which would not occur in FRP bars embedded In this section the results of the tensile tests performed on the
in thicker concrete covers. In addition to large longitudinal and FRP bars embedded in concrete are compared with the results of
1146 E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152
Fig. 15. Bars inside concrete: a) GFRP 4 mm 300 °C, b) GFRP-8 mm 500 °C, c) GFRP-
8 mm 800 °C, d) CFRP-4 mm 600 °C.
the study performed by Ashrafi et al. [5] on the same GFRP bars
tested in this study and the predictive model developed by Wang
et al. [27] that predicts the tensile behavior of FRP materials at ele-
Fig. 13. Failure modes of 4 mm CFRP bar at different temperatures: a) 25 °C, b) vated temperatures.
300 °C, c) 600 °C, d) 800 °C. The average ultimate tensile strengths of the GFRP bars recorded
in this study and those recorded by Ashrafi et al. are depicted in
Fig. 19. In the current research, the bars were embedded in concrete
sleeves whereas Ashrafi et al. tested the same bars with no concrete
cover. The comparison shows that the bars embedded in concrete
showed a better tensile performance at elevated temperatures,
especially at temperatures above 300 °C (more than 20% increase
in tensile capacity), because the concrete covers prevented direct
heat and oxygen reaching the FRP bars. Ashrafi et al. continued their
tests only up to a temperature of 450 °C since the bare FRP bars
ignite at temperatures above 450 °C that causes severe damage to
the test apparatus and furnace.
Ashrafi et al. [5] reported critical temperatures (i.e., where the
bar loses half of its initial tensile strength) of 300, 375, 377, and
450 °C for GFRP bars of diameters 4, 6, 8, and 10 mm, respectively.
In this study, the critical temperatures of GFRP bars were observed
to be about 325, 425, 425, and 475 °C for bars of diameters 4, 6, 8,
and 10 mm, respectively. Likewise, Ashrafi et al. reported the crit-
ical temperature of the same CFRP bar tested in this study as
330 °C, whereas this temperature was obtained as 400 °C in the
current research. As can be seen in Figs. 8 and 12, the critical tem-
perature of GFRP bars with larger diameters was higher than those
of bars with smaller diameters.
Ashrafi et al. [5] reported the temperature in which the GFRP
bars lost half of their tensile strength to be approximately 300 °C,
Fig. 14. The 5-mm CFRP bar tested at 800 °C right after failure. whereas this temperature was established as approximately
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1147
Fig. 16. Delaminated resin and sand coating remained on concrete cover: a) GFRP 6 mm at 500 °C, b) CFRP4 mm at 300 °C.
5. ANOVA analysis
Fig. 19. Comparison of the tensile behavior of FRP bars embedded in concrete (this study) and bare bars (Ashrafi et al.)
f u;T 1
¼ h1 þ h2 þ h3 lnðDÞ þ e ð3Þ
fu lnðT Þ
Fig. 20. The comparison of results of current study on GFRP bars with the A stepwise mode parameter reduction [52] was applied to the
predictions of the model by Wang et al.
initial model form due to the high correlation between parameters
(Table 6). The two parameters that possess correlations higher than
bars at elevated temperatures. In the Bayesian regression 0.5 were combined using Eq. (4) [52].
^hi ¼ l þ q rhi hj l
approach, the model’s uncertainty is explicitly considered in the
ð4Þ
Bayesian inference (Table 9). The general model form is presented hi hi hj
rhj hj
in Eq. (2).
where lhi and rhi are the posterior mean and standard deviation of
f u;T the hi, and qhihj is the correlation between hi and hj. After the reduc-
¼ h1 f 1 ðxÞ þ h2 f 2 ðxÞ þ þ hk f k ðxÞ þ e ð2Þ
fu tion, the Bayesian regression was applied once again, and the pro-
cess continued if correlations higher than 0.5 were available. In
where fu,T, fu, fk, hk, X, and e are the tensile strength at temperature T, this model, as a high correlation was observed between parameters
tensile strength at ambient temperature, kth explanatory function, the reduction led to a model (Eq. (5)) with only one model
kth model parameter, independent variable vectors (regressors), parameter (h2).
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1149
Table 5
ANOVA analysis results for GFRP bars tested at 25–800 °C.
SS: the sum of squares of the deviations of all the observations from their mean.
df: the number of degrees of freedom associated with the sample variance.
MS: the mean square, which is obtained by dividing the sum of squares by the respective degrees of freedom.
F: the variation between samples means (Mean Square Between) to the variation within the samples (Mean Squared Error).
P-value: the probability that an equal amount of variation in the dependent variable would be observed in the case that the independent variable does not affect the
dependent variable.
F crit: an indicator that corresponds to the p value (F crit < F shows that the variable has significant effect on the results).
Table 6
ANOVA analysis results for GFRP bars tested at 150–500 °C.
Table 7
ANOVA analysis results for CFRP bars tested at 25–800 °C.
Table 8
ANOVA analysis results for CFRP bars tested at 150–700 °C.
f u;T 1
¼ h2 B þ þ B0 lnðDÞ þ A þ A0 lnðDÞ þ e ð5Þ
fu lnðTÞ
Table 9
Explanatory functions of the models. In which A, A0 , B, and B0 are the coefficients incorporated into the
Explanatory Function Description Symbol Unite model during the model reduction procedure. These parameters
are tabulated in Table 11. The average of the remaining model
f1 Intercept – –
f2 Test temperature T Kelvin
parameters and their coefficient of variation, the mean of model
f3 Bar Diameter D mm error and its coefficient of variation, and the R-Factor of the model
are presented in Table 12.
Table 10
Results of Bayesian analysis of the first step based on the model in Eq. (2).
Table 11 These models show the parameters affecting the tensile perfor-
Model coefficients.
mance of embedded FRP bars at elevated temperatures. The model
Coefficient GFRP CFRP forms indicate that the tensile performance of FRP bars, as defined
A 8.11 7.7 by fu,t/fu, is in reverse relation with ln(T) and direct relation with ln
B 0.156 0.156 (D) where T and D are the ambient temperature and diameter of
A0 4.44 5.53 FRP bars. These models and their variables can be further modified
B0 0.16 0.21
through future experimental studies and are the basic pathway for
a complete study towards finalized formulation of the strength of
Table 12 GFRP and CFRP profiles at elevated temperatures.
The variable of the proposed model form.
Fig. 21. Model diagnosis for GFRP bars: (a) model prediction versus observation, (b) model predication to observation ratio, (c) model error normality, (d) residual versus
regressor.
E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152 1151
Fig. 22. Model diagnosis for CFRP bars: (a) model prediction versus observation, (b) model predication to observation ratio, (c) model error normality, (d) residual versus
regressor.
reinforced concrete) is used to prevent wide transverse cracks, 8. The developed Bayesian regression models are accurate and can
b) thicker concrete covers are provided over FRP bars. be used to predict the tensile performance of FRP bars embed-
3. The tensile performance of FRP bars experiences a sudden drop ded in concrete covers at elevated temperatures. The model
when the ambient temperature exceeds the glass transition forms indicate that the tensile performance of FRP bars at ele-
temperature (Tg). The presence of a concrete sleeve does not vated temperatures decrease with an increase in temperature
significantly help this issue. and decrease in bar diameter.
4. Resin ignition does occur inside concrete cover at high elevated
temperatures due to lack of oxygen inside concrete. This phe- Additional experimental studies are recommended on FRP bars
nomenon increases the tensile performance of FRP bars since manufactured with different resin matrices and fibers to expand
resin ignition (an exothermic reaction) produces a high amount the database of available test results to enable the development
of heat and accelerates fiber and resin oxidation. This difference of reliable predictive models.
is less than 20% for temperatures below 300 °C but exceeds 20%
for temperatures over 300 °C.
5. At temperatures higher than 500 °C, the axial force is carried Conflict of interest
only by the fibers without contribution from the resin.
Therefore, at temperatures higher than 500 °C the tensile None declared.
performance of FRP bars decreases slowly up to the
temperature where the fibers reach their melting point, beyond Ethical statement
which a sudden drop occurs in the tensile performance of the
bars. Authors state that the research was conducted according to eth-
6. Investigation of the cut sections of concrete sleeves showed that ical standards.
the portion of the bars outside the transverse crack region that
lead to tensile failure of bars was significantly less damaged
than the exposed segment of the bar. Funding body
7. Based on the results of ANOVA analyses, the contribution of bar
diameter has been found to be significant between the temper- This research did not receive any specific grant from funding
atures of 150 °C–500 °C. agencies in the public, commercial, or not-for-profit sectors.
1152 E.P. Najafabadi et al. / Construction and Building Materials 211 (2019) 1138–1152
Acknowledgment [25] X.L. Wang, X.X. Zha, Experimental research on mechanical behavior of GFRP
bars under high temperature, Appl. Mech. Mater. (2011). Trans Tech Publ.
[26] A. Nadjai, D. Talamona, F. Ali, Fire performance of concrete beams reinforced
The support of Vatan Composite Company in supplying materi- with FRP bars, in: Proceeding of the Int. Sympsium on Bond Behaviour of FRP
als and the support of Mr. Hamed Gholipoor in the experiments are in Structures, 2005, 401–410.
[27] K. Wang, B. Young, S.T. Smith, Mechanical properties of pultruded carbon
greatly acknowledged.
fibre-reinforced polymer (CFRP) plates at elevated temperatures, Eng. Struct.
33 (7) (2011) 2154–2161.
[28] R.J. Hamad, M.M. Johari, R.H. Haddad, Mechanical properties and bond
References characteristics of different fiber reinforced polymer rebars at elevated
temperatures, Constr. Build. Mater. 142 (2017) 521–535.
[1] M.M. Rafi, A. Nadjai, Behavior of hybrid (steel-CFRP) and CFRP bar-reinforced [29] M. Robert, B. Benmokrane, Behavior of GFRP reinforcing bars subjected to
concrete beams in fire, J. Compos. Mater. 45 (15) (2011) 1573–1584. extreme temperatures, J. Compos. Constr. 14 (4) (2009) 353–360.
[2] H.M. Mohamed, B. Benmokrane, Design and performance of reinforced [30] L.C. Hollaway, J.-G. Teng, Strengthening and Rehabilitation of Civil
concrete water chlorination tank totally reinforced with GFRP bars: case Infrastructures Using Fibre-reinforced Polymer (FRP) Composites, Elsevier,
study, J. Compos. Constr. 18 (1) (2013) 05013001. 2008.
[3] J.F. Davalos, Y. Chen, I. Ray, Effect of FRP bar degradation on interface bond [31] J. Milke, V. Kodur, C. Marrion, A overview of fire protection in buildings.
with high strength concrete, Cem. Concr. Compos. 30 (8) (2008) 722–730. Appendix A, World Trade Center Building Performance Study, Federal
[4] Z. Wang et al., Long-term durability of basalt-and glass-fibre reinforced Emergency Management Agency, Washington, DC, available at: www.
polymer (BFRP/GFRP) bars in seawater and sea sand concrete environment, fema.gov/library/viewRecord.do, 2002.
Constr. Build. Mater. 139 (2017) 467–489. [32] B. Kandola, E. Kandare, Composites having improved fire resistance, Adv. Fire
[5] H. Ashrafi et al., The effect of mechanical and thermal properties of FRP bars on Retardant Mater. (2008) 398–442.
their tensile performance under elevated temperatures, Constr. Build. Mater. [33] E. Kandare et al., Fire structural modelling of fibre–polymer laminates
157 (2017) 1001–1010. protected with an intumescent coating, Compos. A Appl. Sci. Manuf. 43 (5)
[6] Y. Chen et al., Accelerated aging tests for evaluations of durability performance (2012) 793–802.
of FRP reinforcing bars for concrete structures, Compos. Struct. 78 (1) (2007) [34] B.K. Kandola, W. Bhatti, E. Kandare, A comparative study on the efficacy of
101–111. varied surface coatings in fireproofing glass/epoxy composites, Polym. Degrad.
[7] F.E. Tannous, H. Saadatmanesh, Durability of AR glass fiber reinforced plastic Stab. 97 (11) (2012) 2418–2427.
bars, J. Compos. Constr. 3 (1) (1999) 12–19. [35] M.H. Khaneghahi et al., Effect of intumescent paint coating on mechanical
[8] Y.A. Al-Salloum et al., Effect of harsh environmental conditions on the tensile properties of FRP bars at elevated temperature, Polym. Test. 71 (2018) 72–86.
properties of GFRP bars, Compos. B Eng. 45 (1) (2013) 835–844. [36] C. Katsoulis et al., Post-fire flexural performance of epoxy-nanocomposite
[9] E.P. Najafabadi et al., Effect of applied stress and bar characteristics on the matrix glass fibre composites containing conventional flame retardants,
short-term creep behavior of FRP bars, Constr. Build. Mater. 171 (2018) 960– Compos. A Appl. Sci. Manuf. 43 (8) (2012) 1389–1399.
968. [37] H. Li et al., Micromechanical finite element analyses of fire-retarded woven
[10] M. Bazli, H. Ashrafi, A.V. Oskouei, Effect of harsh environments on mechanical fabric composites at elevated temperatures using unit cells at multiple length
properties of GFRP pultruded profiles, Compos. B Eng. 99 (2016) 203–215. scales, Comput. Mater. Sci. 55 (2012) 23–33.
[11] CAN, C.S.CSA-S806-02. Design and Construction of Building Components with [38] E. Kandare et al., Fiber-reinforced epoxy composites exposed to high
Fiber-Reinforced Polymers, Canadian Standards Association, Mississauga, temperature environments. Part II: modeling mechanical property
Ontario, Canada, 2002. degradation, J. Compos. Mater. 45 (14) (2011) 1511–1521.
[12] 440, A.C.Guide for the Design and Construction of Structural Concrete [39] S. Feih et al., Fire structural modeling of polymer composites with passive
Reinforced with FRP Bars, American Concrete Institute, 2007. thermal barrier, J. Fire Sci. 28 (2) (2010) 141–160.
[13] DT203, C.Guide for the Design and Construction of Concrete Structures [40] S. Alsayed et al., Performance of glass fiber reinforced polymer bars under
Reinforced with Fiber-reinforced Polymer Bars, National Research Council, elevated temperatures, Compos. B Eng. 43 (5) (2012) 2265–2271.
Rome, Italy, 2006. [41] M. Robert, P. Cousin, B. Benmokrane, Durability of GFRP reinforcing bars
[14] B. Williams et al., Fire insulation schemes for FRP-strengthened concrete slabs, embedded in moist concrete, J. Compos. Constr. 13 (2) (2009) 66–73.
Compos. A Appl. Sci. Manuf. 37 (8) (2006) 1151–1160. [42] M. Jarrah et al., The effect of elevated temperatures on the tensile performance
[15] E. Nigro et al., Fire resistance of concrete slabs reinforced with FRP bars. Part I: of GFRP and CFRP sheets, Constr. Build. Mater. 190 (2018) 38–52.
experimental investigations on the mechanical behavior, Compos. B Eng. 42 [43] ASTM International, W.C., PA, USA, Standard Test Method for Tensile
(6) (2011) 1739–1750. Properties of Fiber Reinforced Polymer Matrix Composite Bars ASTM: /
[16] E. Nigro, et al., Tests at high temperatures on concrete slabs reinforced with D7205M-06, 2011.
bent FRP bars. Special Publication, 2011, Vol. 275, pp. 1–20. [44] H.A. Amiri, E.P. Najafabadi, H.E. Estekanchi, Experimental and analytical study
[17] H. Ashrafi et al., Effect of sequential exposure to UV radiation and water vapor of Block Slit Damper, J. Constr. Steel Res. 141 (2018) 167–178.
condensation and extreme temperatures on the mechanical properties of GFRP [45] M. Eskandari, E.P. Najafabadi, Experimental and analytical study of telescopic
bars, J. Compos. Constr. 22 (1) (2017) 04017047. lead yielding damper, J. Constr. Steel Res. 150 (2018) 371–383.
[18] M. Saafi, Effect of fire on FRP reinforced concrete members, Compos. Struct. 58 [46] Institute, A.C.Guide for Design and Construction Structural Concrete
(1) (2002) 11–20. Reinforced with Fiber-Reinforced Polymer (FRP) Bars, 38800 Country Club
[19] L.A. Bisby, M.F. Green, V.K. Kodur, Response to fire of concrete structures that Drive, Farmington Hills, MI, 2015, p. 48331.
incorporate FRP, Prog. Struct. Mat. Eng. 7 (3) (2005) 136–149. [47] J.M. Davies, Y.C. Wang, P.M. Wong, Polymer composites in fire, Compos. A
[20] A. Abbasi, P.J. Hogg, A model for predicting the properties of the constituents of Appl. Sci. Manuf. 37 (8) (2006) 1131–1141.
a glass fibre rebar reinforced concrete beam at elevated temperatures [48] F. Nardone et al., Tensile behavior of epoxy based FRP composites under
simulating a fire test, Compos. B Eng. 36 (5) (2005) 384–393. extreme service conditions, Compos. B Eng. 43 (3) (2012) 1468–1474.
[21] H. Wang, X. Zha, J. Ye, Fire resistance performance of FRP rebar reinforced [49] M.H. Khaneghahi, M. Alembagheri, N. Soltani, Reliability and variance-based
concrete columns, Int. J. Concr. Struct. Mater. 3 (2) (2009) 111–117. sensitivity analysis of arch dams during construction and reservoir
[22] M. Rafi, A. Nadjai, Analytical Method of Temperature Prediction in Reinforced impoundment, Front. Struct. Civ. Eng. (2018) 1–16.
Concrete Beams, J. Struct. Fire Eng. 5 (4) (2014) 367–380. [50] G.E. Box, G.C. Tiao, Bayesian Inference in Statistical Analysis, Vol. 40, John
[23] W. Gao, J. Dai, J. Teng, Simple method for predicting temperatures in Wiley & Sons, 2011.
reinforced concrete beams exposed to a standard fire, Adv. Struct. Eng. 17 [51] M. Mahsuli, T. Haukaas, Computer program for multimodel reliability and
(4) (2014) 573–589. optimization analysis, J. Comput. Civil Eng. 27 (1) (2012) 87–98.
[24] Y.C. Wang, P. Wong, V. Kodur, An experimental study of the mechanical [52] P. Gardoni, A. Der Kiureghian, K.M. Mosalam, Probabilistic capacity models and
properties of fibre reinforced polymer (FRP) and steel reinforcing bars at fragility estimates for reinforced concrete columns based on experimental
elevated temperatures, Compos. Struct. 80 (1) (2007) 131–140. observations, J. Eng. Mech. 128 (10) (2002) 1024–1038.