Composites Part B 141 (2018) 123–136

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Effect of the environment on the performance of GFRP reinforcing bars T

∗
T. D'Antino , M.A. Pisani, C. Poggi
Department of Architecture, Built Environment and Construction Engineering, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133, Milan, Italy

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

Keywords: Fiber reinforced Polymer (FRP) composites are currently employed in the civil engineering industry as externally
Glass fibers bonded reinforcement (EBR) of existing reinforced concrete (RC) and masonry structures and as internal re-
Environmental degradation inforcement of concrete elements as an alternative to steel reinforcing bars. Carbon FRP (CFRP) composites are
Mechanical properties mainly used for EBR applications whereas glass FRP (GFRP) bars are employed as internal reinforcement of
GFRP bars
concrete elements. This paper sheds light into the effect of different aggressive environments on the tensile
Environmental reduction factor
behavior of reinforcing GFRP bars. 356 results of tensile tests of GFRP bars subjected to hot dry and humid air,
different alkali environments, salt solutions with various concentrations, and plain and distillated water were
collected from the literature. According to the “design by testing” procedure provided by EN 1990, a statistical
analysis of the results was carried out to calibrate environmental reduction factors able to provide reliable
estimations of the long-term behavior of GFRP bars subjected to different exposure conditions. For a given
aggressive environment, a clear and unique degradation trend could not be identified, which points out the need
of a standard testing procedure able to provide reliable and repeatable results.

1. Introduction bars are usually comprised of E-Glass fiber reinforcement, which is

known to be degraded by marked alkaline environment. Although al-
In cold regions de-icing is traditionally done with salt along with kali-resistant (AR) glass fibers showed good durability to alkaline en-
sand and gravel. Salt is the most efficient de-icing agent if the pavement vironment such as cementitious matrixes [22], their use in GFRP bars is
temperature is not below −18 °C. Moreover, it is cheap, easy to spread still limited and the bar alkali-resistance is assigned to the matrix that
and store, and readily available in large quantities. Nevertheless, the should prevent the diffusion of deleterious ions [23]. The absorption of
use of de-icing salt on bridge decks leads to corrosion of steel reinfor- moisture depends on the type of organic matrix employed. Moisture,
cing bars and therefore affects their durability. The corrosion of steel which penetrates within the matrix through a diffusion process at the
reinforcing bars causes cracking and spalling of concrete bridge decks molecular level, mainly affects the matrix causing plasticization,
with consequent elevated costs for rehabilitation and traffic disruption. strength decrease, and reduction of the glass transition temperature
To overcome these issues, from the early nineties Glass Fiber Reinforced value.
Polymer (GFRP) bars have been adopted in some regions like Canada as The elastic modulus and strength of GFRP bars decrease with in-
internal reinforcement of concrete slabs, increasing their service life in creasing temperature. This behavior is mainly caused by the marked
these unfavorable conditions (see e.g. Refs. [1–3]). GFRP bars, which change of mechanical properties of the organic matrix when tempera-
are comprised of glass fibers embedded within organic matrices, have a ture attains values close or above the glass transition temperature of the
high strength-to-weight ratio and a good durability with respect to matrix, whereas the glass fibers do not significantly change their
certain exposure conditions. Moreover, GFRP bars are particularly properties (see for instance [24]).
suitable as internal reinforcement of concrete bridge decks because they Design codes and guidelines (see Refs. [23,25–27]) generally take
show good resistance to fatigue loads [4]. Because of these properties, into account the possible degradation of GFRP bars by means of a
fiber reinforced composites have been increasingly employed to strength reduction factor ηa (known also as environmental reduction
strengthen existing concrete elements [5–16]. factor), whose value should depend on fiber/resin type and exposure
However, GFRP bars are sensitive to some environmental condi- conditions, and should be calibrated using a sufficiently large number
tions. In particular, alkaline environment, moisture, and extreme tem- of experimental tests.
peratures degrade their mechanical properties (e.g., tensile strength, In this paper, after a brief description of GFRP bars and their me-
ultimate strain, and modulus of elasticity) [17–21]. GFRP reinforcing chanical properties, 356 results of conditioned GFRP bars subjected to

∗
Corresponding author.
E-mail addresses: [email protected] (T. D'Antino), [email protected] (M.A. Pisani), [email protected] (C. Poggi).

https://doi.org/10.1016/j.compositesb.2017.12.037
Received 13 October 2017; Received in revised form 18 December 2017; Accepted 18 December 2017
Available online 21 December 2017
1359-8368/ © 2017 Elsevier Ltd. All rights reserved.
T. D'Antino et al. Composites Part B 141 (2018) 123–136

tensile tests carried out by 20 different research groups are collected

from the literature and compared to assess the effect of different ex-
posure conditions. The effect on the bar tensile strength of prolonged
exposition to hot dry and humid air, alkali environment with different
pH, salt solutions with different concentrations, and plain and dis-
tillated water were investigated, also considering the bar fiber volume
fraction (i.e. ratio between the fiber volume and volume of the bar) and
matrix and resin type. Results were divided in groups depending on the
exposure time and temperature. According to the “design by testing” Fig. 2. Slippage at the bar-anchor head interface.
procedure provided by EN 1990, a statistical analysis of the results was
carried out to calibrate environmental reduction factors able to provide
reliable estimations of the long-term behavior of GFRP bars subjected to
different exposure conditions. The results obtained help to shed light on
the durability of GFRP bars with respect to different exposure and ac-
celerated aging conditions. However, it is opinion of the authors that
further studies are needed to obtain reliable estimations of the long-
term behavior of GFRP bars.

2. GFRP reinforcing bars Fig. 3. Failure due to fiber trimming on the collar of the anchor area.

GFRP reinforcing bars are comprised of longitudinal glass fibers

[28] (Fig. 3);
impregnated with an organic resin. The tensile strength and modulus of
elasticity of GFRP bars, which are affected by several factors (e.g. fiber • Longitudinal delamination. Non-uniform radial pressure distribu-
tion causes detachment of portions of the bar at the bar-anchor head
volume fraction, typically ranging between 0.5 and 0.7, and manu-
interface, where the contact pressure is low. As a consequence, bar
facturing and curing conditions), were reported to vary between
delamination in the direction parallel to the fibers occurs (Fig. 4).
483 MPa and 1600 MPa and between 35 GPa and 51 GPa, respectively
[23]. To improve the bond behavior between the GFRP reinforcing bar
To obtain the complete tensile rupture of the bar, ASTM D7205 [29]
and the embedding concrete, different bar surfaces can be employed.
recommends anchorages comprised of steel tubes filled with either a
Although smooth bars are available (usually with small diameters), the
polymer resin or an expansive cement grout able to prevent excessive
bar surface may be molded-deformed with resin ribs, ribbed with glass
slip of the bar prior to failure. The recommended steel tube length
fiber wrapped around the bar longitudinal axis, sand-coated, or may
varies from 300 mm to 460 mm for GFRP bars with diameter that varies
present a combination of these treatments.
between 6.4 mm and 9.5 mm.

2.1. Tensile tests

2.2. Long-term behavior tests
The experimental evaluation of the tensile strength of GFRP bars is
To investigate the long-term behavior of GFRP bars subjected do
not an easy task due to the difficulty to anchor the specimen to the
different exposure conditions, accelerated aging tests are usually
testing machine. The reason of this difficulty is the marked orthotropy
adopted. No specific standard or guideline for accelerated tests of GFRP
of this unidirectionally-reinforced composite material. The GFRP bar
bars is currently available and only a standard general framework for
weakness in radial direction requires a soft gripping of the specimen
developing accelerated tests for building materials and components is
but, at the same time, a strong gripping is needed to allow for attaining
available from the literature [30]. Different models for estimating the
rupture of the glass fibers. Anchor heads are often used to provide the
long-term behavior of GFRP bars can be found in the literature [31].
proper stress-transfer from the testing machine to the specimen without
However, GFRP accelerated aging tests are generally based on the Ar-
damaging the bar [28]. GFRP bars tested in tension can present dif-
rhenius equation, which provides the relation between the analyzed
ferent failure modes:
reaction rate (e.g. diffusion of humidity within the matrix resin) and the

• Tensile rupture. This failure mode is associated with the highest temperature:
stress that can be attained. The failure occurs with a sudden con- k = Ae−Ea/ RT (1)
temporary collapse of all fibers and an extensive interfacial de-
where k is the reaction rate constant, A is the pre-exponential factor
bonding while the broken fibers are fan-shaped (Fig. 1);
•
that depends on the type of reaction, Ea is the activation energy of the
Slippage at the bar-anchor head interface. The anchorage is not
reaction, and R and T are the universal gas constant and absolute
effective and the bar slips with respect to the machine wedge or to
temperature, respectively. Coefficients of Eq. (1) can be calibrated
the anchor head (Fig. 2);
•
following the procedures provided by ASTM E 2041 [32] and ASTM E
Fiber trimming on the collar of the anchor area. The edges of the
698 [33]. Arrhenius equation assumes that there is a single dominant
anchor head (or the machine wedges) cut the peripherals fibers of
degradation process that does not change with time. Moreover, the
the bar at the loaded end, where the interface pressure has a peak
process degradation rate increases with increasing temperature [34].
With these assumptions, Arrhenius equation can be employed to obtain

Fig. 1. Fiber tensile rupture at the end of a tensile test. Fig. 4. Longitudinal delamination failure.

124
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Fig. 5. Frequency distribution of a) matrix type, b) fiber volume fraction vf, c) diameter ϕ, and d) unconditioned tensile strength σfu of bars included in the database.

the relationship between the behavior of a GFRP bar exposed to a terephthalate (PT) matrix. The fiber volume fraction vf ranged between
specific aggressive condition at temperature T1 for a time period t1 and 45.0 and 84.2 (according to the authors), whereas the bar diameter ϕ
that of a GFRP bar exposed to the same condition at temperature ranged between 6.0 mm and 19.5 mm.
T2 > T1 and for a time period t2 < t1. Therefore, the long-term be- The initial (unconditioned) tensile strength σfu of the bar varied
havior and service-life of GFRP bars can be investigated by increasing between 362 MPa and 1478 MPa. The frequency distribution of speci-
the temperature for a given exposure condition. The main limit of this mens included in the database is depicted in Fig. 5a, b, c, and d for
approach lies in the assumption that the reaction process does not different matrix types and different ranges of vf, ϕ, and σfu, respectively.
change with time. In fact, it was observed that reaction processes tend Details of the specimens included in the database and subjected to hot
to change during the exposure period, which may affect the provisions dry and humid air, alkali environment, salt solutions, plain and dis-
obtained with the Arrhenius equation [35]. tillated water, and embedded within concrete beams immersed in plain
water and salt solution (sea water) are provided in the appendix in
Tables A1, A2, A3, A4, and A5, respectively.
3. Experimental database The bar surface treatment, indicated in Tables A1-A5 for each spe-
cimen, was assumed to have no influence on the bar tensile strength
The experimental database employed in this paper is comprised of and was not included among the parameters studied.
356 results of tensile tests on conditioned GFRP bars collected from the
literature. Results were obtained by 20 research groups [31,34,36–53].
In this section, the database is presented and analyzed while a detailed 4. Analysis and comparison of conditioned GFRP bar residual
discussion of the results is reported in Section 5. Bars included in the strength
database are mainly made of E-glass fibers (340 specimens, 95.5% of
total specimens) although some specimens comprising alkali-resistant The database collected was employed to evaluate the effect on GFRP
(AR) glass fibers (16 specimens, 4.5% of total specimens) were con- bar tensile strength of prolonged exposition to:
sidered.
Specimens presented different matrix type, fiber volume fraction vf, I) hot dry and humid air,
diameter ϕ, tensile strength of the unconditioned (control) specimen II) alkali environment with different pH,
σfu, and surface treatment. Different matrix types were used, namely III) salt solutions with different concentrations,
thermoplastic (TP), urethane modified vinylester (UM-VE), vinylester IV) plain and distillated water.
(VE), polyester (PE), modified vinylester (M-VE), epoxy vinylester (E-
VE), mixed vinylester and polyester (VE + PE), and polyethylene Results are provided in term of residual strength ratio σf/σfu, which

125
T. D'Antino et al. Composites Part B 141 (2018) 123–136

obtained with bare bars, which indicated that the well-known concrete
alkalinity did not affect the bar residual strength ratio [50]. This ob-
servation is not always confirmed by tests on bars subjected to different
alkaline environment, as discussed below.

4.2. Bars immersed in alkaline solutions with different pH

202 specimens were tested by 18 research groups to investigate the

effect of different alkaline environments on the bar tensile strength.
Bars were mainly comprised of E-glass fibers with different volume
fraction and vinylester matrix, although some bars presented different
Fig. 6. Residual strength ratio of bars exposed to air with various temperatures and RH resin matrix (e.g. thermoplastic, urethane, polyester). Eight bars com-
for different exposure times. prised of AR-glass and different inorganic matrices [37] were con-
sidered for comparison. Details of each bar, including the diameter and
is defined as the ratio between the tensile strength of the conditioned surface treatment, are provided in Table A2. To simulate a harsh al-
specimen σf and the tensile strength σfu of the corresponding uncondi- kaline environment, specimens were immersed in alkaline solutions
tioned specimen. that were heated to accelerate the degradation process. The alkaline
solutions adopted were grouped in four pH ranges, namely
pH = 12–12.15, pH = 12.5–12.7, pH = 13, and pH = 13.4–13.6. Si-
4.1. Bars exposed to hot air with different relative humidity milarly, the solution temperatures were divided in three ranges, namely
range a (11–25 °C), b (26–53 °C), and c (57–80 °C). The residual strength
Two research groups investigated the residual strength of GFRP bars ratio of each conditioned bar is shown in Fig. 7 for the different tem-
exposed to air for different exposure times. Twelve specimens [51] were perature ranges analyzed. The results reported in Ref. [38] were not
comprised of bare GFRP bars, whereas 9 specimens were comprised of considered because specimens were embedded in concrete and the pH
GFRP bars embedded within a concrete prism in their central part [50]. was not clearly measured. Robert et al. [49] and Robert and Benmok-
All bars employed were comprised of E-glass fiber and vinylester matrix rane [52] conditioned bars embedded in concrete and measured the pH
(Table A1). Two exposure temperatures were considered, namely 23 °C of the interstitial solution extracted from concrete after aging. Simi-
(lab conditions) and 50 °C. Some specimens were exposed to two dif- larly, Almusallam et al. [50] conditioned bars embedded within con-
ferent field conditions with temperatures in the range 9–45 °C and crete with pH between 12.5 and 13 (pH = 12.7 was assumed). Speci-
average relative humidity RH = 0.26 and RH = 0.52. mens subjected to freeze and thaw and wet and dry cycles were not
The residual strength ratio σf/σfu obtained for each specimen is considered for comparison.
shown in Fig. 6. The results show that exposure to air did not sig- Bars subjected to different alkaline environments in the temperature
nificantly affect the bar strength (minimum residual strength ratio σf/ range 11–25 °C (66 specimens, Fig. 7a), show a decreasing residual
σfu = 97.5% for a bar exposed to 50 °C air for 12960 h). Since a small strength ratio with increasing the exposure time. Data with pH = 13.0
scatter between results is observed (coefficient of variation and pH = 13.4–13.6 show a decrease rate much higher than those with
CoV = 0.90%), the average residual strength ratio σf/σfu = 99.24% pH = 12–12.15 and pH = 12.5–12.7. These results, however, are af-
could be computed without distinguishing between different exposure fected by the failure mode of the bar. Considering Almusallam et al.
times. Furthermore, a clear influence of the RH cannot be observed. [50] and Vijay [31], which attained complete bar failure, it is possible
Specimens conditioned within concrete provided results similar to those to observe a minimum residual strength ratio of 78.3%. Results from Al-

Fig. 7. Residual strength ratio of bars exposed to alkaline solutions at a) 11–25 °C, b) 26–53 °C, and c) 57–80 °C for different exposure times.

126
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Zahrani et al. [40], which tested bars similar to those employed by significant decrease of residual strength ratio with increasing exposure
Vijay [31] under a more aggressive exposure condition, did not show a time. Various authors reported different behaviors and different re-
significant strength reduction. This difference may be attributed to the sidual strength ratio decrease rates with time t, d(σf/σfu)/dt. Robert and
slippage of the bars within the anchorage, which prevented the fiber Benmokrane [52], which reported complete bar failure, observed a
failure of specimens by Al-Zahrani et al. [40]. limited strength decrease, namely σf/σfu = 94.4%, for a bar conditioned
Although the behavior of results by Chen et al. [45] and Chen [46] for 2880 h. Chen et al. [45] reported a clear decreasing behavior, with
is in agreement with the behavior generally observed for increasing increasing d(σf/σfu)/dt with increasing pH. Won et al. [47] reported a
value of pH, a minimum residual strength ratio (σf/σfu = 42.0% for lower d(σf/σfu)/dt respect to that observed by Chen et al. [45] for the
5760 h exposure and pH = 13.6) lower than that reported for the same same pH and exposure time. Micelli and Nanni [43] tested different
conditioning time and similar pH values (σf/σfu = 90.9% for 5760 h diameter bars and observed that a significant strength decrease oc-
exposure and pH = 13.0, Vijay [31]) was observed. This difference curred for bars with diameter 6.3 mm, whereas no decrease was ob-
could be attributed to the non-complete bar failure observed in Refs. served for bars with diameter 12 mm. Surprisingly, the bar tested by
[45,46], which reported bar delamination with partial fiber tensile Sawpan et al. [53] after 17520 h exposure with pH = 13.0 reported a
failure. residual strength ratio higher than that reported by other authors for
Considering specimens that were subjected to different alkaline lower pH and exposure time.
environments at temperatures in the range 26–53 °C (59 specimens), it Specimens comprising AR-glass provided σf/σfu values similar to
is possible to observe a general decrease of the bar residual strength those of specimens made of E-glass conditioned for the same exposure
ratio for increasing exposure time and pH (Fig. 7b). For authors that time and temperature. However, no final conclusions can be drawn
documented a complete bar failure, i.e. tensile rupture of all fibers since this result was attributed to the specific AR-glass bar manu-
within the cross-section [49–52], a minimum residual strength ratio of facturing process and tests with different bars were suggested by the
76.0% was observed, which is similar to that observed for specimens authors to confirm the results [37].
subject to temperatures in the range 11–25 °C. This suggests that the Results of specimens comprising different fiber volume fractions vf
temperature range considered did not significantly affect the bar tensile and conditioned for different exposure times are shown in Fig. 8 for the
strength. Analogously to specimens conditioned at 11–25 °C, the results three temperature ranges considered. For bars in the range 11–25 °C
from Refs. [45,46] presented a larger strength decrease rate for speci- (Fig. 8a), a clear influence of vf on the residual strength ratio cannot be
mens subjected to pH = 13.5 than those subjected to pH = 12.5–12.7, observed. High residual strength ratios were obtained with fiber volume
with a minimum residual strength ratio of 32.5% for 5760 h exposure fractions higher or equal to 60% and lower than 50%, whereas scat-
and pH = 13.5. tered results were obtained for vf = 50%.
Boris and Porter [36] tested bars with different matrices (two bars The fiber volume fraction vf appears to have a limited effect on the
with a mixed vinylester-polyester matrix and one bar with a vinylester residual strength ratio for bars conditioned at 26–53 °C. Similar residual
matrix) subjected to the same environment (pH = 12.7) for 2904 h. strength ratio values were obtained for bars with vf higher than 70%
Their results showed that the vinylester matrix provided a better pro- conditioned for the same number of hours [49]. However, a slight re-
tection to the bar, which had a residual strength ratio of 83.1%, duction of the bar residual strength ratio for high values of vf (namely
whereas bars with the mixed vinylester-polyester matrix showed an 84.2%), was observed (Fig. 8b).
average residual strength ratio of 74.1%. However, this consideration For bars with 49.1% ≤ vf ≤ 60%, the fiber volume fraction ap-
cannot be generalized because the number of specimens tested appears peared to affect the bar tensile strength. A reduction of residual strength
too limited. ratio from σf/σfu = 83.1% to σf/σfu = 69.9% was reported by Boris and
Considering the temperature range 57–80 °C (Fig. 7c), 53 specimens Porter [36] for bars with vf = 54.3% and vf = 58.3%, respectively,
were exposed to different alkali environments and presented a which were conditioned for 2904 h. Similarly, Kim et al. [48] observed

Fig. 8. Residual strength ratio of bars exposed to alkaline solutions at a) 11–25 °C, b) 26–53 °C, and c) 57–80 °C for different fiber volume fractions.

127
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Fig. 9. Residual strength ratio of bars comprising different matrix types exposed to alkaline solutions for different exposure times.

a strength reduction from 80.5% to 67.3% for bars conditioned for solutions heated to accelerate the degradation process. Three tem-
1800 h and with vf = 49.9% and vf = 54.2%, respectively. perature ranges were identified, namely range a (11–25 °C), b
Comparing the residual strength ratio obtained by bars with dif- (40–50 °C), and c (80 °C). Furthermore, different salt concentrations
ferent fiber volume fractions conditioned at temperatures in the range were considered, namely 3% by weight, 3.5% by weight, and ≥4% by
57–80 °C, the ratio σf/σfu appears to increase with increasing vf weight. To obtain the last salt concentration, sea water of the Arabian
(Fig. 8c). This behavior, which is in contrast with the behavior observed Gulf-Eastern Province of Saudi Arabia, which can have a salt con-
for specimens conditioned at temperatures between 26 °C and 53 °C that centration even higher than 4% by weight (see for instance [54–56]),
showed an opposite trend, could be explained by the matrix severe was employed.
degradation at temperatures close to the glass transition temperature, Twenty-one bars were immersed in salt solutions at 11–25 °C. The
which in turn could have affected the residual strength ratio of bars degradation obtained was not pronounced and the residual strength
with low fiber volume fraction. ratio was higher than 80% except for one bar conditioned for 7200 h
Since bars comprising AR-glass had the same fiber volume fraction that provided a residual strength ratio of 74.8% (Fig. 10a). For this
vf = 72.0%, the influence of vf on the residual strength ratio obtained temperature range, the salt concentration did not appear to influence
could not be investigated. the results, which do not present a clear trend.
All specimens subjected to alkaline environment are shown in Fig. 9 Results from 14 specimens conditioned at 40–50 °C (Fig. 10b) show
for different exposure times and with the indication of the matrix type a limited reduction of tensile strength, with a minimum residual
employed. The scatter between the results obtained do not allow for strength ratio of 77.2% associated with 1440 h conditioning. Four
identifying the effect of each matrix type. It should be noted, however, specimens were also subjected to wet and dry cycles (WD in Fig. 10),
that most specimens (73%) were made of vinylester resin. which did not affect the residual strength ratio obtained.
The average of residual strength ratios for each temperature range Only one research group applied a conditioning temperature higher
considered, together with the corresponding coefficient of variation and than 50 °C, namely Kim et al. [48], which conditioned 8 specimens at
number of specimens considered, are reported in Table 1. A decrease of 80 °C (Fig. 10c). The results obtained show two separate groups, the
the residual strength ratio with increasing temperature can be ob- former with residual strength ratios between 80.7% and 89.7% and the
served, with a minimum average residual strength ratio of 73.2% for latter with residual strength ratios between 42.6% and 65.3%. The
the temperature range 57–80 °C. If related to the temperature of ex- presence of two distinct groups can be attributed to the different matrix
posure, this result shows a relatively limited bar degradation. type and fiber volume fraction of the bars, as discussed below.
The computation of the average of residual strength ratio made in The fiber volume fraction vf does not seem to affect the bar residual
Table 1 are based on the outcomes of Fig. 9, which do not show a de- strength ratio for specimens conditioned at 11–25 °C (Fig. 11a),
pendence of the residual strength ratio on the exposure time, although whereas σf/σfu shows a slight increase with increasing vf for specimens
such relationship should be reasonably expected. conditioned at 40–50 °C (Fig. 11b). A significant increase of residual
strength ratio was observed for bars conditioned at 80 °C with in-
creasing vf from 49.1% to 54.2% (Fig. 11c).
4.3. Bars immersed in solutions with different salt concentrations
All specimens subjected to saline environment are shown in Fig. 12
for different exposure times and with the indication of the matrix type
Results of 47 GFRP bars immersed in solutions with different salt
employed. Vinylester, urethane modified vinylester and polyester ma-
concentrations were collected from the literature. Four bars were sub-
trix did not affect the obtained residual strength ratio. Specimens with
jected to freeze and thaw cycles and were not considered for compar-
modified vinylester matrix (vf = 49.1%) conditioned at 80 °C showed a
ison [31]. Most of the bars were comprised of E-glass fiber and viny-
significant reduction of residual strength ratio respect to bars with vi-
lester matrix and presented a ribbed surface with sand coating. Details
nylester matrix (vf = 54.2%) subjected to the same conditioning.
of each bar are provided in Table A3. The bars were immersed in saline
However, it should be noted that this difference shall not be entirely
attributed to the matrix type but also to the different fiber volume
Table 1
Average residual strength ratio of bars immersed in alkaline solutions for different tem- fractions.
perature ranges. The average of residual strength ratios for each temperature range
considered, together with the corresponding coefficient of variation and
Range Temperature [°C] Number of specimens Average σf/σfu [%] (CoV) number of specimens, is reported in Table 2. The average residual
a 11–25 66 88.9 (0.139) strength ratio of specimens conditioned at 11–25 °C and 40–50 °C is
b 26–53 59 84.3 (0.145) very similar. Although a significant reduction is observed for specimens
c 57–80 53 73.2 (0.205) conditioned at 80 °C, the average σf/σfu = 68.2% cannot be confirmed

128
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Fig. 10. Residual strength ratio of bars exposed to saline solution at a) 11–25 °C, b) 40–50 °C, and c) 80 °C for different exposure times.

Fig. 11. Residual strength ratio of bars exposed to saline solution at a) 11–25 °C, b) 40–50 °C, and c) 80 °C for different fiber volume fractions.

Fig. 12. Residual strength ratio of bars comprising different

matrix types exposed to saline solutions for different ex-
posure times.

129
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Table 2 reported a residual strength ratio between 15% and 20% lower than
Average residual strength ratio of bars immersed in salt solutions for different tempera- that obtained by the same authors at 11–25 °C for the same exposure
ture ranges.
times. This behavior can be also observed by comparing the residual
Range Temperature [°C] Number of specimens Average σf/σfu [%] (CoV) strength ratio for varying fiber volume fractions in the range 26–50 °C,
where σf/σfu increases with increasing vf except for specimens by Al-
a 11–25 21 88.2 (0.069) Salloum et al. [51] (Fig. 14b).
b 40–50 14 87.0 (0.050)
Nine specimens were immersed in water heated at temperatures
c 80 8 68.2 (0.249)
between 60 °C and 80 °C (Fig. 13c). The residual strength ratio obtained
from these specimens is generally lower than that obtained at tem-
due to the large scatter between results conditioned at 80 °C perature in the range 11–50 °C, with a minimum residual strength ratio
(CoV = 0.249), presence of differences between bars tested (i.e. matrix σf/σfu = 43.9% for 3168 h exposure time. As for lower temperatures,
type and fiber volume fraction), and absence of results from different σf/σfu appeared to increase with increasing vf for specimens conditioned
authors. at 60–80 °C (Fig. 14c).
Comparing results obtained with bars made of different matrix
4.4. Bars immersed in water types, a clear influence of the matrix was not observed in the range
11–50 °C (Fig. 13a and b). However, only specimens made with poly-
Fifty-seven specimens, tested by five different research groups, were ethylene terephthalate matrix presented an increase of tensile strength
immersed in tap or distillated water for different exposure times to after conditioning [42]. For conditioning at 60–80 °C, specimens with
assess the effect of possible moisture absorption on the bar tensile modified vinylester matrix (vf = 49.1%) were associated with residual
strength. Bars with different matrix types and fiber volume fractions strength ratios lower than those obtained with the vinylester matrix
were employed (Table A4). In one case, which is not considered for (vf = 54.2%) for the same exposure times. This behavior, which was
comparison, wet and dry cycles were applied [46]. To accelerate the observed for specimen immersed in salt solution and conditioned at
moisture absorption, the conditioning water was heated at different 80 °C by the same research group [48], should be attributed both to the
temperatures. Results were divided into three groups depending on the matrix type and fiber volume fraction but was not confirmed by other
water temperature, namely range a (11–25 °C), b (26–50 °C), and c authors.
(60–80 °C). The average of residual strength ratios for each temperature range
Residual strength ratios of bars immersed in water at 11–25 °C, considered, together with the corresponding coefficient of variation and
26–50 °C, and 60–80 °C for different exposure times are reported in number of specimens, is reported in Table 3. Specimens in ranges
Fig. 13a, b, and c, respectively. For low temperatures (11–25 °C), the 11–25 °C and 26–50 °C provided similar results, although slightly lower
bar tensile strength was not significantly affected by the water con- values were obtained in the range 26–50 °C. Similarly to the case of bars
ditioning. Results by Al-Salloum et al. [51], which reported a complete immersed in salt solutions, high temperatures (60–80 °C) were asso-
bar failure, showed a minimum residual strength ratio of 95.1% for ciated with significant strength reduction (average σf/σfu = 67.2%).
12960 h exposure time. Results by Kim et al. [48] are more scattered, This behavior was observed by a single research group and could not be
with a minimum residual strength ratio of 75.7% for 1440 h exposure confirmed by other data.
time. For some specimens, Gaona [42] reported an increase of tensile
strength after conditioning. Differences between results by Al-Salloum 4.5. Bars embedded within concrete beams immersed in plain water and salt
et al. [51], Gaona [42], and Kim et al. [48] may be attributed to the solution (sea water)
different fiber volume fraction adopted (Fig. 14a). Indeed, the residual
strength ratio appears to increase with increasing vf in the range To study the durability of GFRP bars under real serviceability con-
11–25 °C. ditions, some authors immersed concrete beams reinforced with GFRP
Eighteen specimens were immersed in water at 26–50 °C and pre- bars in plain water and salt solutions. After the conditioning period, the
sented residual strength ratios like those obtained at 11–25 °C bars were extracted from the beams and subjected to tensile test to
(Fig. 13b). However, specimens tested by Al-Salloum et al. [51] assess their strength.

Fig. 13. Residual strength ratio of bars immersed in water at a) 11–25 °C, b) 26–50 °C, and c) 60–80 °C for different exposure times.

130
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Fig. 14. Residual strength ratio of bars immersed in water at a) 11–25 °C, b) 26–50 °C, and c) 60–80 °C for different fiber volume fractions.

Table 3 those directly immersed in water at similar temperatures (Fig. 13b).

Average residual strength ratio of bars immersed in water for different temperature This difference may be attributed to the alkalinity of the embedding
ranges. concrete that contributed to the bar degradation. Indeed, a similar re-
Range Temperature [°C] Number of specimens Average σf/σfu [%] (CoV)
sidual strength ratio was obtained by bars embedded within concrete
and bars directly immersed in alkaline solutions at temperatures be-
a 11–25 29 93.5 (0.076) tween 11 °C and 53 °C (Fig. 7a and b). However, it should be noted that
b 26–50 18 87.3 (0.083) specimens embedded within concrete and exposed to air did not show a
c 60–80 9 67.2 (0.217)
significant strength reduction ([50], Section 4.1).
When conditioned at 50–60 °C, specimens embedded within con-
Twenty-nine GFRP bars comprising E-glass fiber and vinylester crete showed a significant strength decrease with increasing exposure
matrix were used as reinforcement of concrete beams that were im- time. A similar decrease was observed for some specimens directly
mersed in plain water and in a salt solution for different exposure times immersed in water at temperature in the range 60–80 °C (Fig. 13c),
(Table A5). For the salt solution, sea water of the Arabian Gulf-Eastern whereas higher residual strength ratios were provided by bars directly
Province of Saudi Arabia, which can have a salt concentration higher immersed in alkaline solutions with pH ≤ 13 at 57–80 °C (Fig. 7c).
than 4% by weight, was employed [39,50]. To accelerate the possible In Fig. 15 the residual strength ratio obtained by bars embedded
bar degradation, the conditioning water was heated at 40 °C, 50 °C, and within concrete and conditioned with a solution with NaCl ≥ 4% by
60 °C for some specimens. weight (≥4% NaCl in Fig. 15) and subjected to wet and dry cycles with
The residual strength ratio σf/σfu obtained by bars conditioned at the same solution (≥4% NaCl WD in Fig. 15) is also reported. The σf/σfu
various temperatures is plotted in Fig. 15 for different exposure times. obtained showed that specimens conditioned with salt solution were
Specimens by Davalos et al. [34] showed a decrease of σf/σfu with in- less degraded than those conditioned with plain water at the same
creasing the exposure time. Furthermore, the rate of decrease of σf/σfu temperature (i.e. 40 °C). Specimens subjected to wet and dry cycles
increased with increasing the conditioning temperature. If compared provided σf/σfu values lower than those of specimens simply immersed
with specimens directly immersed in water (Fig. 13a), bars embedded in the same salt solution but higher than those of corresponding bars
within concrete and conditioned at 20 °C showed a similar residual conditioned with plain water.
strength ratio, whereas lower residual strength ratios were observed for Residual strength ratios are plotted in Fig. 16 for specimens tested at
bars embedded within concrete and conditioned at 40 °C respect to 20–40 °C with different fiber volume fractions vf, which did not have

Fig. 15. Residual strength ratio of bars embedded within

concrete and immersed in plain water and salt solution at
20 °C, 40 °C, 50 °C, and 60 °C for different exposure times.

131
T. D'Antino et al. Composites Part B 141 (2018) 123–136

Fig. 16. Residual strength ratio of bars embedded within

concrete and immersed in plain water and salt solution at
20–40 °C for different fiber volume fractions.

Table 4 solutions present in the concrete fill its pores, which cover a small
Average residual strength ratio of bars embedded within concrete for different tem- fraction of the concrete-bar contact surface, whereas in accelerated
perature ranges.
aging conditions bars are completely immersed in the alkaline solution
Range Temperature [°C] Number of specimens Average σf/σfu [%] (CoV)
that wets all the bar external surface. As in the case of elements im-
mersed in water, the exposure to high temperatures to accelerate the
a 20–40 19 83.6 (0.108) alkali degradation process might affect the results. The thermal re-
b 50–60 10 57.6 (0.231) sponse of a structural member is a transient phenomenon influenced by
many parameters, such as time-dependent solar radiation, temperature,
wind speed fluctuations, material properties, surface characteristics,
any influence on the results. The influence of fiber volume fraction was
and section geometry [59]. Thermal analyses indicate that the design
not studied for specimens exposed to temperature higher than 40 °C,
distributions are not likely to greatly exceed those provided by the New
which had the same vf. Furthermore, since all bars were comprised of E-
Zealand code [60] for any latitude between 45°S and 45°N, as the peak
glass fiber and vinylester matrix, the influence of matrix and fiber type
solar radiation levels on horizontal surfaces between these latitudes do
could not be investigated.
not vary significantly. According to [60], the maximum temperature of
The average residual strength ratio provided by specimens em-
reinforcing bars embedded within concrete never exceeds 45 °C. This
bedded within concrete at 20–40 °C (range a in Table 4) is similar to
thermal distribution is similar to that adopted by EN 1991-1-5 [61].
that obtained from bars directly immersed in alkaline solutions at
However, these considerations do not explain the dispersion of ex-
26–53 °C (Table 1). However, higher conditioning temperature pro-
perimental results observed in Section 4. A possible reason may be the
vided an average residual strength ratio lower for bars embedded
tensile test procedure adopted and, specifically, the anchorage system
within concrete (range b in Table 4) than for bars directly immersed in
that should guarantee complete fiber tensile rupture (Section 2.1).
alkaline solutions, salt solutions, and plain water (Tables 1–3, respec-
Pictures of some failure modes were reported in the analyzed papers.
tively). The average results are affected by the overall number of spe-
Only few authors well-documented the failure mode observed pro-
cimens and by the number of specimens tested for each exposure time
viding photos and descriptions, whereas most of the papers provided
considered.
only limited information.
Results included in the database were analyzed in terms of residual
5. Discussion tensile strength. However, the tensile strength of the unconditioned
control specimens varied significantly (see Section 3). Nowadays, good
The comparison of the residual tensile strength obtained from GFRP quality GFRP bars (with diameter between 10 mm and 28 mm) gen-
reinforcing bars exposed to severe environmental conditions and tested erally provide a tensile strength between 800 MPa and 1100 MPa (e.g.
by 20 different research groups showed in many cases contradictory Ref. [28]). Two of the research groups declare strengths much lower
results and discordant trends. than these values. The lowest strength is reported in Ref. [43] and refers
The first issue is represented by the temperature of water solutions to a group of specimens which strength ranges between 295 MPa and
employed for conditioning some specimens. It is well-known that 407 MPa. The authors attributed this small strength “to a low fiber
temperatures above 50 °C, although often lower than the glass transi- content as seen by SEM investigations”. The low unconditioned tensile
tion temperature of the organic matrix, affect the physical properties of strength values reported by Abbasi and Hogg [44], namely 362 MPa
the bars by increasing the thermal expansion coefficient [57]. There- and 366 MPa, were attributed to the damage of fibers during the
fore, the adoption of such high temperatures to accelerate the de- manufacturing process. The bad alignment of the fibers was excluded
gradation process can alter the bar and should be adopted only in the because straight glass fibers were observed after resin burning-off. Fiber
case of real severe environmental conditions, since usual serviceability damage, due to poor manufacturing, leads to inefficiency in the stress
conditions never reach these high temperatures. In fact, the highest transfer mechanisms between fiber filaments, which causes a high
temperature of the sea surface is measured in the Persian Gulf, where strength reduction.
37 °C can be attained (see for instance [56,58]). Consequently, results The tensile testing of GFRP bars is not an easy task with uncondi-
of experimental tests made on bars conditioned in water solutions he- tioned specimens and becomes even more complex with conditioned
ated at temperatures above 40 °C are not realistic and may be affected bars. For example, in Ref. [62] the authors commented on the results of
by the modification of the GFRP properties. their tests by observing that “the conditioned GFRP bars usually failed
The accelerated alkali environment conditions do not reflect the real at the end section of the conditioning zone (edges of the plastic re-
exposure condition of reinforcing bars. The pH values selected for the servoir). A sudden change in bar properties occurs here, which led to
accelerated alkaline environment conditions, which vary between 12 failure initiation. Therefore, the resulting failure stresses stand as a
and 13.6 for the specimens included in the database, should replicate lower limit and would have been higher had the entire bar been sub-
that of concrete, whose value is not unique. Moreover, alkaline jected to the conditioning as in field situations”.

132
T. D'Antino et al. Composites Part B 141 (2018) 123–136

To overcome the issues associated with tensile testing, bending Table 5

tests, usually with three-point bending test set-ups, were employed by Results of the linear regression analysis for the different environmental conditions and
ranges studied.
some authors [35,63,64]. However, bending tests of GFRP bars do not
allow for controlling the failure mode obtained. Specimens can fail due Range Alkaline Salt Water
to fiber tensile rupture in the tension zone or to bar crushing in the
compression zone. a b a b a b
The mechanism leading to failure is affected by the mechanical
n 54 48 21 14 29 18
properties in compression and tension of the bar, which depend on the m 0.0002 −0.0007 0.0000 0.0005 0.0009 −0.0006
type of resin matrix and on the fiber volume fraction. Another im- q 90.0083 90.2535 88.1377 84.8079 89.3492 89.9811
portant limit of bar bending tests is that the fibers of the outer (sleeve) r2 0.0058 0.0891 0.0000 0.2050 0.1275 0.0795
layer, which are more degraded than inner (core) fibers in conditioned
specimens, are the most stressed, whereas the core fibers, which retain
f (t ) = mt + q (3)
their mechanical properties, provide a limited contribution to the
bending strength. Local defects due to manufacture or handling can A linear regression analysis was performed on residual strength
cause similar problems. On the other hand, a GFRP bar employed as ratios in ranges a and b of specimens subjected to alkaline, salt, and
internal reinforcement of a concrete slab subjected to bending can be water conditioning to obtain the parameters m and q. The results ob-
assumed to be stressed in longitudinal direction homogeneously across tained are reported in Table 5, where r2 is the coefficient of determi-
its entire cross-section, which implies that core fibers play a primary nation and n is the number of specimen of the specific range considered.
role in the bar strength. Therefore, bending tests of reinforcing bars can It should be noted that specimens from Chen et al. [45] and Chen [46]
provide strength values lower than those provided by the same bar were not included in the regression analysis of alkaline environment
subjected to real serviceability conditions. ranges because their results significantly differ from those of other
Bending tests of concrete beams reinforced with GFRP bars, even if authors for the same exposure condition (see Fig. 7a and b). Therefore,
expensive, appear ideal for providing reliable results on the bar tensile the number of specimens in ranges a and b of alkaline environment
strength since they do not require the development of a specific pro- reduced from 66 to 54 and from 59 to 48, respectively. Although
cedure or testing set-up and are the most realistic. However, results of functions with more complex shapes (e.g. polynomial, exponential,
these tests are affected by the mechanical properties of concrete, which etc.) provided results slightly more accurate than those obtained with
vary over time and depend on the type of conditioning. This dis- Eq. (3), their behavior was not monotonical with time and therefore
advantage can be solved by subjecting a series of concrete specimens, they were not considered.
obtained from the same concrete batch used to cast the reinforced In general, the slope m of the lines obtained from the linear re-
beam, to the same conditioning of the beam. These specimens should be gression analysis is very limited (Table 5). Furthermore, only ranges b
then tested contemporarily to each bending test in order to measure the of alkaline and water environments provided a negative slope, whereas
mechanical properties of concrete at that specific time. Nevertheless, all other ranges provided positive values of m, which would entail for
the need to handle, condition, and test heavy and bulky specimens an increase of the residual strength ratio with time. Therefore, it was
unfortunately make these tests much more expensive than bar tensile or assumed that the slope of the regression lines be negligible (i.e. m ≈ 0)
bending tests. and q was computed as:
n

6. Calibration of environmental reduction factors q = (σf /σfu)avg = 1/ n⋅ ∑ (σf /σfu)i (4)

i=1

In this section, the results collected in the database are statistically where (σf/σfu)i is the residual strength ratio of the i-th specimen.
analyzed to calibrate environmental reduction factors that allow for Once the function f(t) is known, γ can be obtained through the “design
obtaining the residual tensile strength of GFRP bars subjected to alka- by testing” procedure proposed by EN 1990 [65]. To verify that γ is a
line environment, salt solution, and immersed in water. Since bars ex- unit-mean normally-distributed function for all ranges, the first four
posed to hot air did not show a reduction of the tensile strength (Section statistics moments, namely the average (μ), variance (s2), skewness (γ1),
4.1), it was assumed that no environmental reduction factor is needed and kurtosis (γ2), were computed for the γ functions associated with
for this treatment. Furthermore, bars embedded within concrete beams each range. In addition, the Kolgomorov-Smirnov (K-S) test was exe-
were not considered in this section because limited results by only two cuted to verify the maximum distance between the γ function and a
research groups were found in the literature (Section 4.5). normal distribution with the same mean and standard deviation of the γ
To calibrate the environmental reduction factors, the “design by considered. The CoV of each γ function was computed too. The results
testing” procedure provided by EN 1990 [65] was followed. According obtained are reported in Table 6 for all ranges and exposure conditions
to this procedure, the residual strength ratio σf/σfu can be expressed as: considered.
σf Table 6 shows that all functions γ has mean values very close to unit
= γ⋅f (t ,ξ) (2) and can be considered normally distributed. For comparison, the
σfu

where f(t,ξ) is a deterministic function that relates the mean residual Table 6
Results of the statistical analysis of the distribution of functions γ.
strength ratio (σf/σfu)avg to the exposure time t and to relevant en-
vironmental parameters ξ, and γ is a normally-distributed unit-mean Range Alkaline Salt Water
aleatoric function. As a first attempt, each exposure condition was
analyzed separately for ranges a and b. Range c was not considered a b a b a b
because degradation processes different from those of the exposure n 54 48 21 14 29 18
condition considered may arise with high temperatures, which in turn μ 1.009 1.010 1.005 1.002 1.006 1.006
would affect the results obtained (see Section 5). Since exposure con- s2 0.011 0.011 0.005 0.003 0.006 0.007
ditions and temperatures ranges were analyzed separately, the depen- γ1 1.845 0.846 0.534 1.143 1.012 −0.051
γ2 3.943 0.301 0.131 0.722 1.467 −0.440
dence of f to the environmental parameters ξ can be neglected, which
CoV 0.104 0.103 0.070 0.053 0.080 0.082
provided f(t,ξ) = f(t). As a first step, a linear shape of the function f(t) K-S 0.182 0.125 0.133 0.199 0.188 0.111
was assumed:

133
T. D'Antino et al. Composites Part B 141 (2018) 123–136

equal to 3.04 for an infinite number of measures. Mean (γm), char-

acteristic (γk), and design (γd) values of γ obtained are reported in
Table 7.
In addition, values of γ associated with a 5% (characteristic value)
and 0.12% (design value) probability of obtaining a lower value [65]
were computed based on the Student's t-distribution. The results ob-
tained, which are provided in Table 7, are similar to those based on the
procedure by EN 1990 [65] for ranges with large number of specimens
and are more conservative when n decreases.
The characteristic and design values of the residual strength ratio
computed by Eqs. (5) and (6), respectively, are reported in Table 8 for
each range studied. Residual strength ratios provided in Table 8 are
depicted in Figs. 7, 10 and 13 for bars exposed to alkaline environment,
salt solution, and immersed in water, respectively.
The residual tensile strength of a GFRP bar exposed to an alkaline
environment, salt solution, or immersed in water at a temperature
Fig. 17. Comparison between cumulative distribution frequencies of the experimental
within ranges a and b can be obtained as:
results of alkaline solution range b and corresponding normal distribution.

Table 7 σf , p = σfu ⎡γp (σf /σfu)avg ⎤ (7)

⎢ ⎥
Values of γ for the different environmental conditions and ranges studied. ⎣ ⎦

Range n γm γk γd Method where p = m, k, and d for mean, characteristic, and design values, re-
spectively. Since environmental reduction factors ηa are generally
a b a b a b a b
multiplied by the unconditioned bar tensile strength σfu to obtain the
Alkaline 54 48 1.00 1.00 0.852 0.794 0.726 0.666 EN 1990 [65] reduced tensile strength (see e.g. Ref. [23]), results reported in Table 8
0.849 0.791 0.713 0.650 t-distribution represent environmental reduction factors ηa.
Salt 21 14 1.00 1.00 0.884 0.902 0.783 0.828 EN 1990 [65]
Environmental reduction factors provided by ACI 440.1R-15 [23]
0.882 0.899 0.763 0.803 t-distribution
Water 29 18 1.00 1.00 0.873 0.802 0.738 0.781 EN 1990 [65] and CNR-DT 203/2006 [26] are equal to 0.8 for concrete elements not
0.871 0.799 0.720 0.758 t-distribution exposed to earth and weather and to 0.7 for concrete elements exposed
to earth and weather. These factors, which include the effect of tem-
perature on the rebar (provided that the service temperature is lower
cumulative distribution function (CDF) of range b of alkaline environ- than the matrix resin glass transition temperature), represent con-
ment, which is representative of the CDFs obtained, is compared with servative estimates based on the consensus of the guideline committees.
the CDF of the corresponding normal distribution in Fig. 17. The red The most conservative values of ηa obtained in this paper following EN
cross mark reported in Fig. 17 represents the maximum difference be- 1990 [65] are equal to 0.61 and 0.67 for specimens in alkaline en-
tween the two CDFs, as indicated by the K-S test of this range. vironment and conditioned at temperatures in ranges a and b, respec-
Since unit-mean can be assumed, a measure of the γ function ac- tively (Table 8). However, results of specimens exposed to alkaline
curacy in estimating the corresponding residual strength ratio can be environments discussed in Section 4.2 were mostly obtained by direct
obtained by the coefficients of variation (CoV). Table 6 reports rela- exposure of the GFRP bars to alkaline solutions, which are likely to
tively limited CoV values, which confirm the accuracy of the γ function provide a higher degradation than that of bars embedded in concrete
provisions. and exposed to alkaline environments (see Section 5). Values of ηa
The characteristic (i.e. 5% percentile) value of the residual strength obtained for each exposure condition analyzed at environmental tem-
ratio (σf/σfu)k can be obtained by Eq. (5) [65]: perature (range a) are all close to 0.8, which is consistent with values of
(σf /σfu)k = (1 − kn⋅CoV )(σf /σfu)avg = γk (σf /σfu)avg ηa provided by Refs. [23,26] for concrete elements not exposed to earth
(5)
and weather.
where kn is the characteristic fractile factor, which depends on the The reliability of the ηa calibrated in this paper is strictly related to
number of test results n and is equal to 1.64 for an infinite number of the number of tests considered and to the homogeneity of results in
measures. EN 1990 [65] provides also a procedure for direct assessing each range [65]. However, in some cases significant discrepancies be-
the design value (σf/σfu)d for ultimate limit state verifications: tween results with similar environmental conditions by different re-
search groups were observed (Sections 4 and 5). Further studies in-
(σf /σfu)d = (1 − kd, n⋅CoV )(σf /σfu)avg = γd (σf /σfu)avg (6) volving concrete elements reinforced with GFRP bars and exposed to
where kd,n is the design fractile factor, which is associated with a real environmental conditions are needed to provide more reliable es-
probability of 0.12%, depends on the number of test results n, and is timation of the composite durability.

Table 8
Residual strength ratios [%] obtained from the statistical analysis for the different environmental conditions and ranges studied.

Range n γm(σf/σfu)avg γk(σf/σfu)avg γd(σf/σfu)avg Method

a b a b a b a b

Alkaline 54 48 91.65 86.56 78.09 72.77 66.51 61.00 EN 1990 [65]

77.81 72.46 65.31 59.60 t-distribution
Salt 21 14 88.18 87.02 77.98 79.57 69.00 72.99 EN 1990 [65]
77.74 79.30 67.24 70.82 t-distribution
Water 29 18 93.45 87.32 81.56 74.96 69.00 72.99 EN 1990 [65]
81.43 74.71 67.24 70.82 t-distribution

134
T. D'Antino et al. Composites Part B 141 (2018) 123–136

7. Conclusions Appendix A. Supplementary data

In this paper, the residual strength ratio of GFRP bars subjected to Supplementary data related to this article can be found at http://dx.
different environment conditioning was investigated in an attempt to doi.org/10.1016/j.compositesb.2017.12.037.
shed light on the durability of this reinforcement that can promisingly
increase the durability of reinforced concrete structures. 356 tensile References
tests on conditioned GFRP bars were collected from the literature and
their results were analyzed and compared. A statistical analysis of the [1] Klowak C, Memon A, Mufti A. Static and fatigue investigation of second generation
results was carried out following the “design by testing” procedure steel-free bridge decks. Cement Concr Compos 2006;28:890–7.
[2] Mufti A, Neale K. State-of-the-art of FRP and SHM applications in bridge structures
provided by EN 1990. Characteristics and design environmental re- in Canada. Compos Res J 2008;2(2):60–9.
duction factors were provided for specimens subjected to alkaline en- [3] Rostàsy F. FRP tensile elements for prestressed concrete: state of the art, potentials
vironment, salt solution, and immersed in water. Based on the results and limits. In: Nanni A, Dolan C, editors. Fiber-reinforced plastic reinforcement for
concrete structures - SP-138. Farmington Hills, Mich: American Concrete Institute;
obtained, the following conclusions can be drawn: 1993. p. 347–65.
[4] Carvelli V, Pisani M, Poggi C. Fatigue behaviour of concrete bridge deck slabs re-
• significant differences of methodologies, procedures, and result in- inforced with GFRP bars. Compos Part B 2010;41:560–7.
[5] D'Antino T, Carloni C, Sneed LH, Pellegrino C. Fatigue and post-fatigue behavior of
terpretation between different experimental studies with similar
PBO FRCM-concrete joints. Int J Fatig 2015;81:91–104.
environmental conditions were observed. The differences in testing [6] D'Antino T, Triantafillou TC. Accuracy of design-oriented formulation for evaluating
procedures can explain the dispersion of the results. the flexural and shear capacities of FRP-strengthened RC beams. Struct Concr

• environmental reduction factors obtained are very similar to those 2016;17(3):425–42.

[7] D'Antino T, Pisani MA. Evaluation of the effectiveness of current guidelines in de-
provided by ACI 440.1R-15 and CNR-DT 203/2006 for all exposure termining the strength of RC beams retrofitted by means of NSM reinforcement.
conditions studied at environmental temperature. Lower values of ηa Compos Struct 2017;167:166–77.
were obtained for specimens exposed to temperature higher than [8] Del Zoppo M, Di Ludovico M, Balsamo A, Prota A, Manfredi G. FRP for seismic
strengthening of shear controlled RC columns: experience from earthquakes and
25 °C. experimental analysis. Compo Part B 2017;129:47–57.
• a shared standard procedure is urgently needed. Such procedure [9] Rousakis TC, Saridaki ME, Mavrothalassitou SA, Hui D. Utilization of hybrid ap-
proach towards advanced database of concrete beams strengthened in shear with
should guarantee the effectiveness and reliability of the test set-up -
FRPs. Compos Part B 2016;85:315–35.
with particular attention to anchoring and possible damage of the [10] Seo S-Y, Sung Lee M, Feo L. Flexural analysis of RC beam strengthened by partially
bar - and the application of environmental conditions that correctly de-bonded NSM FRP strip. Compos Part B 2016;101:21–30.
reproduce real field conditions without introducing further dete- [11] Charalambidi BG, Rousakis TC, Karabinis AI. Analysis of the fatigue behavior of
reinforced concrete beams strengthened in flexure with fiber reinforced polymer
rioration sources. Namely, durability tensile tests on GFRP bars laminates. Compos Part B 2016;96:69–78.
should comply with the following rules: i) tests in the alkaline en- [12] Di Luccio G, Michel L, Ferrier E, Martinelli E. Seismic retrofitting of RC walls ex-
vironment should be carried out on specimens comprised by GFRP ternally strengthened by flax–FRP strips. Compos Part B 2017;127:133–49.
[13] Triantafillou GG, Rousakis TC, Karabinis AI. Analytical assessment of the bearing
bars embedded within concrete (dry or wet) whereas direct bar
capacity of RC beams with corroded steel bars beyond concrete cover cracking.
exposure to alkaline solutions should be avoided; ii) the exposure Compos Part B 2017;119:132–40.
temperature should not exceed 45 °C and 40 °C for alkaline en- [14] Kalfat R, Al-Mahaidi R. Mitigation of premature failure of FRP bonded to concrete
using mechanical substrate strengthening and FRP spike anchors. Compos Part B
vironments and water solutions, respectively; iii) fiber tensile rup-
2016;94:209–17.
ture, proved by photos of the failed specimens, should be obtained. [15] Carozzi FG, Bellini A, D'Antino T, de Felice G, Focacci F, Hojdys L, et al.
Experimental investigation of tensile and bond properties of Carbon-FRCM com-
Although the large number of collected test results, the significant posites for strengthening masonry elements. Compos Part B 2017;128:100–19.
[16] D'Antino T, Papanicolaou C. Mechanical characterization of textile reinforced in-
discrepancies observed between the results suggests the need of further organic-matrix composites. Compos Part B 2017;127:78–91.
studies involving concrete elements reinforced with GFRP bars and [17] Daniel I, Ishai O. Engineering mechanics of composite materials. second ed. New
exposed to real environmental conditions to provide more reliable es- York: Oxford University Press; 2006. p. 411.
[18] Karbhari VM, Murphy K, Zhang S. Effect of concrete based alkali solutions on short-
timation of the composite durability. term durability of e-glass/vinyl ester composites. J Compos Mater
2002;36(17):2101–21.
List of symbols [19] fib (International Federation for StrucConcrete). FRP reinforcement in RC struc-
tures. Lausanne: fib; 2007. p. 120.
[20] Yan F, Lin Z, Zhang D, Gao Z, Li M. Experimental study on bond durability of glass
* bars embedded within concrete or mortar. fiber reinforced polymer bars in concrete exposed to harsh environmental agents:
AR AR-glass freeze-thaw cycles and alkaline-saline solution. Compos Part B 2017;116:406–21.
[21] Benmokrane B, Ali AH, Mohamed HM, ElSafty A. Laboratory assessment and dur-
E E-glass ability performance of vinyl-ester, polyester, and epoxy glass-FRP bars for concrete
E-CR corrosion resistant E-glass structures. Compos Part B 2017;114:163–74.
E-VE epoxy vinylester [22] ACI 549.4R-13. Guide to design and construction of externally bonded fabric-re-
inforced cementitious matrix (FRCM) systems for repair and strengthening concrete
FT freeze and thaw cycles
and masonry structures. Farmington Hills, MI: American Concrete Institute; 2013.
M molded p. 71.
M-VE modified vinylester [23] ACI 440.1R-15. Guide for the design and construction of concrete reinforced with
PE polyester FRP bars. Farmington Hills, MI: American Concrete Institute; 2015. p. 88.
[24] Carvelli V, Pisani M, Poggi C. High temperature effects on concrete members re-
PT polyethylene terephthalate inforced with GFRP rebars. Compos Part B 2013;54:125–32.
R ribbed with glass fiber wrapped around the bar longitudinal [25] Japan Society of Civil Engineers (JSCE). Recommendation for design and con-
axis struction of concrete structures using continuous fiber reinforcing materials. 1997.
Tokyo.
Rsa R + Sa [26] CNR-DT 203/2006. Guide for the design and construction of concrete structures
S smooth reinforced with fiber-reinforced polymer bars. Rome: National Research Council;
Sa sand-coated 2006. p. 35.
[27] CAN/CSA-S806-02. Design and construction of building components with fibre-
TP thermoplastic reinforced polymers. Rexdale: Canadian Standards Association; 2012. p. 198.
UM-VE urethane modified vinylester [28] Carvelli V, Fava G, Pisani M. Anchor system for tension testing of large diameter
VE + PE mixed vinylester and polyester GFRP bars. ASCE J Compos Construct 2009;13(5):344–9.
[29] ASTM D7205/D7205M-6. Standard test method for tensile properties of fiber re-
VE vinylester inforced polymer matrix composite bars. West Conshohocken, PA: ASTM
WD wet and dry cycles International; 2016. p. 13.
[30] ASTM E 632-82. Standard practice for developing accelerated tests to aid prediction

135
T. D'Antino et al. Composites Part B 141 (2018) 123–136

of the service life of building components and materials. West Conshohocken, PA: 2008;39:764–72.
ASTM International; 1996. p. 6. [48] Kim H, Park Y, You Y, Moon C. Short-term durability test for GFRP rods under
[31] Vijay P. Aging and design of concrete members reinforced with GFRP bars PhD various environmental conditions. Compos Struct 2008;83:37–47.
Thesis Morgantown, West Virginia, USA: Department of Civil Engineering, West [49] Robert M, Cousin P, Benmokrane B. Durability of GFRP reinforcing bars embedded
Virginia University; 1999 p. 205. in moist concrete. J Compos Construct 2009;13(2):66–73.
[32] ASTM E 2041-03. Standard method for estimating kinetic parameters by differential [50] Almusallam TH, Al-Salloum YA, Alsayed SH, El-Gamal S, Aqel M. Tensile properties
scanning calorimeter using the Borchardt and Daniels method. West Conshohocken, degradation of glass fiber-reinforced polymer bars embedded in concrete under
PA: ASTM International; 2003. p. 9. severe laboratory and field environmental conditions. J Compos Mater
[33] ASTM E 698-01. Standard test method for Arrhenius kinetic constants for thermally 2012;47(4):393–407.
unstable materials using differential scanning calorimetry and the Fkynn/Wall/ [51] Al-Salloum YA, El-Gamal S, Almusallam TH, Alsayed SH, Aqel M. Effect of harsh
Ozawa method. West Conshohocken, PA: ASTM International; 2005. p. 8. environmental conditions on the tensile properties of GFRP bars. Compos Part B
[34] Davalos J, Chen Y, Ray I. Long-term durability prediction models for GFRP bars in 2013;45:835–44.
concrete environment. J Compos Mater 2012;46(16):1899–914. [52] Robert M, Benmokrane B. Combined effects of saline solution and moist concrete on
[35] Gonenc O. Durability and service life prediction of concrete reinforcing materials long-term durability of GFRP reinforcing bars. Construct Build Mater
PhD Thesis Madison, WI, USA: Department of Civil and Environmental 2013;38:274–84.
Engineering,University od Wisconsin-Madison; 2003 [53] Sawpan MA, Mamun AA, Holdsworth PG. Long term durability of pultruded
[36] Boris T, Porter M. Advancements in GFRP materials provide improved durability for polymer composite rebar in concrete environment. Mater Des 2014;57:616–24.
reinforced concrete. Convention, Marketing/technical sessions of the Composites [54] Smith R, Purnama A, Al-Barwani HH. Sensitivity of hypersaline Arabian Gulf to
Institute's international composites EXPO '99, Cincinnati, OH, USA. 1999. seawater desalination plants. Appl Math Model 2007;31:2347–54.
[37] Tannous FE, Saadatmanesh H. Durability of AR glass fiber reinforced plastic bars. J [55] Taher MM, Mohamed ARM, Al-Ali AKH. Some ecological characteristics and ich-
Compos Construct 1999;3(1):12–9. thyofauna of surrounding Sammaliah Island, Abu Dhabi, UAE. Bahamas J Sci
[38] Bhise VS. Strength degradation of GFRP bars PhD Thesis Blacksburg, Virginia, USA: 2012;30(2):31–49.
Department of Civil and Environmental Engineering, Virginia Polytechnic and State [56] John VC, Coles SL, Abozed A i. Seasonal cycles of temperature,salinity and water
University; 2002 p. 105. masses of the Western Arabian Gulf. Oceanol Acta 1990;13(3):273–82.
[39] Almusallam T, Al-Salloum Y, Alsayed S, Alhozaimy A. Durability of GFRP rebars in [57] Robert M, Wang P, Cousin P, Benmokrane B. Temperature as an accelerating factor
stressed concrete beams at different environments. The 6th Saudi engineering for long-term durability testing of FRPs: should there be any limitations? ASCE J
conference. Dhahran, Saudi Arabia: KFUPM; 2002. Compos Construct August 2010;14(4):361–7.
[40] Al-Zahrani MM, Al-Dulaijan SU, Sharif A, Maslehuddin M. Durability performance [58] Kabiri K, Pradhan b, Rezai H, Ghobadi Y, MM. Fluctuation of sea surface tem-
of glass fiber reinforced plastic reinforcement in harsh environments. The 6th Saudi perature in the Persian Gulf and its impact on coral reef communities around Kish
engineering conference. Dhahran, Saudi Arabia: KFUPM; 2002. Island. IEEE colloquium on humanities, science & engineering research (CHUSER
[41] Stone DK, Koenigsfeld D, Myers J, Nanni A. Durability of GFRP rods, laminates and 2012), Kota Kinabalu, Sabah, Malaysia. 2012.
sandwich panels subjected to various environmental conditioning. Second inter- [59] Priestley M. Chapter 5: the thermal response of concrete bridges. In: Cope RJ,
national conference on durability of fiber reinforced polymer (FRP) composites for editor. Concrete bridges engineering: performance and advances. London (UK):
construction, Montreal, Quebec, Canada. 2002. Elsevier Applied Science; 1978. p. 143–88.
[42] Gaona FA. Characterization of design parameters for fiber reinforced polymer [60] N. Z. Ministry of Work and Development. Highway bridge design brief, CDP701/A.
composite reinforced concrete systems PhD Thesis College Station, TX, USA: Zachry 1978.
Department of Civil Engineering, Texas A&M University; 2003 p. 270. [61] European Committee for Standardization (CEN). Eurocode 1: actions on structures -
[43] Micelli F, Nanni A. Durability of FRP rods for concrete structures. Construct Build Part 1-5: general actions - thermal actions. EN 1991-1-5. Brussels: CEN; 2009.
Mater 2004;18:491–503. [62] Debaiky A, Nkurunziza G, Benmokrane B, Cousin P. Residual tensile properties of
[44] Abbasi A, Hogg PJ. Temperature and environmental effects on glass fibre rebar: GFRP reinforcing bars after loading in severe environments. ASCE J Compos
modulus, strength and interfacial bond strength with concrete. Compos Part B Construct 2006;10(5):370–80.
2005;36:394–404. [63] Gentry TR, Bank LC, Barkatt A, Prian L. Accelerated test methods to determine the
[45] Chen Y, Davalos JF, Ray I. Durability prediction for GFRP reinforcing bars using long-term behavior of composite highway structures subject to environmental
short-term data of accelerated aging tests. J Compos Construct 2006;10(4):279–86. loading. J Compos Technol Res 1998;20(1):38–50.
[46] Chen Y. Accelerated ageing tests and long-term prediction models for durability of [64] Robert M, Benmokrane B. Behavior of GFRP reinforcing bars subjected to extreme
FRP bars in concrete PhD Thesis Morgantown, West Virginia, USA: Department of temperatures. J Compos Construct 2010;14(4):353–60.
Civil and Environmental Engineering, West Virginia University; 2007 p. 214. [65] European Committee for Standardization (CEN). Eurocode - basis of structural de-
[47] Won J, Lee S, Kim Y, Jang C, Lee S. The effect of exposure to alkaline solution and sign. EN 1990. Brussels: CEN; 2002.
water on the strength–porosity relationship of GFRP rebar. Compos Part B

136