Prediction of the Long-Term Performance and Durability of GFRP Bars under the Combined Effect of a Sustained Load and Severe Environments

With the continuous development of production technology, the performance of glass-fiber-reinforced polymer (GFRP) bars is also changing, and some design codes are no longer applicable to new materials based on previous research results. In this study, a series of durability tests were carried out on a new generation of GFRP bars in laboratory-simulated seawater and a concrete environment under different temperatures and sustained loads. The durability performance of GFRP bars was investigated by analysing the residual tensile properties. The degradation mechanism of GFRP bars was also analysed by scanning electronic microscopy (SEM). Furthermore, the long-term performance of GFRP bars exposed to concrete pore solution under different stress levels was predicted using Arrhenius theory. The research results show that the degradation rate of GFRP bars was increased significantly at a 40% stress level. By comparing the test results, design limits, and other scholars’ research results, it is demonstrated that the GFRP bars used in this test have a good durability performance. It is found that the main degradation mechanism of the GFRP bars is the debonding at the fiber-matrix interface. In the range test, the effects of a 20% stress level on the degradation of GFRP bars were not obvious. However, the long-term performance prediction results show that when the exposure time was long enough, the degradation processes were accelerated by a 20% stress level.


Introduction
The steel bars in traditional reinforced concrete (RC) structures are subject to increasing corrosion due to the continuous intrusion of chloride ions in seawater, which reduces the durability of the RC structures. Costly remedial measures, such as coating the surface of the steel bar with epoxy or galvanized steel using stainless steel bars, have failed to improve the long-term durability performance of RC structures [1]. In recent years, fiber-reinforced polymer (FRP) bars have been increasingly used in civil engineering and are expected to serve as an ideal alternative to steel bars because of their resistance to corrosion. Due to the lower cost of glass-FRP (GFRP) bars, they are more widely used than aramid-FRP (AFRP) and carbon-FRP (GFRP) bars. However, glass fibers and commonly used thermosetting resin matrices are known to be susceptible to erosion by moisture, a high temperature, and alkali conditions. GFRP bars may undergo polymer degradation, fiber-matrix debonding, and a.
The residual tensile properties of stressed GFRP bars in a corrosive environment were evaluated and the obtained results were compared with the literature and the design limits given by ACI 440.1R-15 [22]; b.
The experimental phenomena and SEM analysis results were analysed to investigate the degradation mechanism of GFRP bars exposed to simulated seawater solution and concrete pore solution; c.
The long-term performance of GFRP bars exposed to concrete pore solution under different stress levels was predicted based on Arrhenius theory.

The GFRP Bars Used in this Study Were Prepared
The GFRP bars used in this study were prepared by the pultrusion process with E44 epoxy resin and fibers. The detailed production parameters are shown in Table 1 (Harbin FRP Institute, Harbin, Heilongjiang, China) was wrapped on the surface of the GFRP bar, which resulted in varied diameters of the GFRP bars at different locations. The nominal diameter of the GFRP bar was 9.2 mm, as shown in Figure 1. The tensile and physical properties of the GFRP bars are listed in Table 2. The tensile properties of the bar were determined by ACI 440.1R-15 [22] and the physical properties were provided by the manufacturer. physical properties were provided by the manufacturer.

Test Parameters
This study mainly includes three parameters: a. Sustained tensile stresses: 20% and 40% of the ultimate tensile strength were selected in this test.
The corresponding strains were 5200 and 10,400 με, respectively. These levels of strain are about 1.68-3.35 times the values recommended by ACI 440.1R-15 for creep rupture strain (Table 2). This was done to explore the material's potential and evaluate how conservative the current codes and guidelines are; b. Surrounding environment: Two different corrosive mediums were used to simulate the seawater and concrete environment. The compositions of the two simulated solutions are shown in Table  3, and the pH of the two solutions meets the requirements of ACI 440.3R-04 [23]; c. Temperature: Both ambient and elevated temperatures were used, with the ambient temperature being 23 °C and elevated temperatures being 40 and 60 °C. Table 3. Compositions of simulated seawater and concrete pore solutions.   Table 3, and the pH of the two solutions meets the requirements of ACI 440.3R-04 [23]; c.
Temperature: Both ambient and elevated temperatures were used, with the ambient temperature being 23 • C and elevated temperatures being 40 and 60 • C.

Specimen Design and Test Procedure
Test specimen: The total length of the GFRP bar used for testing was 1000 mm, and each end was anchored with 275 mm-long and 18 mm-inner diameter steel tubes for applying axial tension. The middle of the specimens was instrumented with two strain gauges to monitor longitudinal strains. In order to expose the GFRP to a specific solution, a plastic pipe (Linyi Dongli Plastic building materials CO., LTD, Linyi, Shandong, China) with a diameter of 63 mm and a length of 250 mm was inserted into the middle of the bar, which served as a reservoir. A PVC pipe was inserted into the gap between the two ends of the plastic pipe and the GFRP bar, and then glued them together. An opening with a diameter of 15 mm was left in the PVC pipe at one end of the plastic pipe, to allow for frequent replacement of the solution when the pH of the solution was lower than the set value. The unstressed specimen is shown in Figure 2a.

Specimen Numbering
In this research, three specimens were tested for each condition, and a total of 162 specimens were tested. The specimen numbering was defined as follows.
The numbering of conditioned specimens consists of five parts: The first letter G stands for GFRP, and the second letter stands for the solution type. For example, S stands for seawater solutions, and C stands for concrete pore solutions. Then, the first number represents the exposure temperature, the second number represents the sustained stress level, and the third number represents the exposure time. For a typical conditioned specimen, G/S/40/20/30 indicates that a GFRP bar with a stress level of 20% is exposed to a simulated seawater solution at 40 °C for 30 days. Loading device: The loading device was composed of a reaction force frame and a spring. The reaction frame was connected by two 18 mm thick steel plates and four steel rods (diameter = 8 mm) through bolts, and the stiffness of the spring was 790 kN/mm. The loading procedures mainly consisted of two steps: the first step was to fix the specimen on the reaction frame, and the second step was to compress the spring by constantly tightening the nut (2) to provide a reaction force. Then, the force was applied to the test specimen through the reaction frame until the compression of the spring reached the target length and the strain gauge reached the corresponding strain. Finally, the nut (3) was tightened so that the GFRP bar was under sustained stress. During the test, when the compression length of the spring dropped below the calculated length, it was compressed to the target length. In the subsequent test process, the strains of the GFRP bar were checked every 5 days to observe the creep process. The stressed specimen is shown in Figure 2b.

Scanning Electron Microscope (SEM)
Temperature control: A glass fiber heating belt (Beijun Group, Yanchen, Jiangsu, China) was used to wrap the plastic tube and heat the solution in the plastic tube. A temperature sensor between the plastic tube and the heating belt was installed, with a US-controlled TN-99 temperature controller (Tianjin, China), to monitor the temperature. During the test, the temperature error was controlled within ±2 • C, as shown in Figure 3.

Specimen Numbering
In this research, three specimens were tested for each condition, and a total of 162 specimens were tested. The specimen numbering was defined as follows.
The numbering of conditioned specimens consists of five parts: The first letter G stands for GFRP, and the second letter stands for the solution type. For example, S stands for seawater solutions, and C stands for concrete pore solutions. Then, the first number represents the exposure temperature, the second number represents the sustained stress level, and the third number represents the exposure time. For a typical conditioned specimen, G/S/40/20/30 indicates that a GFRP bar with a stress level of 20% is exposed to a simulated seawater solution at 40 °C for 30 days.

Scanning Electron Microscope (SEM)
A JEOL 7001F field emission scanning electron microscope (SEM, Osaka, Japan) was used to observe the cross section of the GFRP bar samples to understand the degradation mechanism of GFRP bars. The samples observed with the SEM were reference specimens and specimens conditioned in 60 °C solutions for the longest duration, which was 90 days. To prepare the samples, the GFRP bars were embedded in epoxy resin, cut with a low-speed saw, and polished with sandpaper and fluffy cloth.  Test procedure: After reaching a defined duration under the controlled testing condition, the preload on specimens was released and the PVC container was removed. Then, the specimen was subjected to the tensile test according to ACI 440.3R B.2-04 [23].

Specimen Numbering
In this research, three specimens were tested for each condition, and a total of 162 specimens were tested. The specimen numbering was defined as follows.
The numbering of conditioned specimens consists of five parts: The first letter G stands for GFRP, and the second letter stands for the solution type. For example, S stands for seawater solutions, and C stands for concrete pore solutions. Then, the first number represents the exposure temperature, the second number represents the sustained stress level, and the third number represents the exposure time. For a typical conditioned specimen, G/S/40/20/30 indicates that a GFRP bar with a stress level of 20% is exposed to a simulated seawater solution at 40 • C for 30 days.

Scanning Electron Microscope (SEM)
A JEOL 7001F field emission scanning electron microscope (SEM, Osaka, Japan) was used to observe the cross section of the GFRP bar samples to understand the degradation mechanism of GFRP bars. The samples observed with the SEM were reference specimens and specimens conditioned in 60 • C solutions for the longest duration, which was 90 days. To prepare the samples, the GFRP bars were embedded in epoxy resin, cut with a low-speed saw, and polished with sandpaper and fluffy cloth.

Results and Discussion
Tables 4 and 5 list the results obtained for the GFRP bars tested in the laboratory. The results include the residual tensile strength vs. strain, tensile strength vs. strain retention, and the residual elastic modulus.   Tables 4 and 5 present the experimental residual tensile strength and corresponding retention for the tested specimens. It can be observed that the tensile strength degradation behavior of GFRP bars exposed to seawater and concrete pore solutions is almost similar. That is, the higher the temperature and stress level, the lower the tensile strength retention.

Columns 3 and 4 in
The tensile strength retentions of GFRP bars are summarized in Figure 4. It can be concluded from Figure 4 that a higher temperature accelerates the degradation rate of GFRP bars. When the temperature increased from 0 to 60 • C, the tensile strength retention of unstressed GFRP bars decreased by 16.4% and 25.0%, respectively, after being exposed to seawater and concrete pore water solution for 60 days. This indicates that the diffusion rate of the corrosive mediums is accelerated by higher temperatures. The curves of the GFRP bars under stress levels of 0% and 20% were close to each other, as shown in Figure 4. However, when the stress level increased to 40% of the ultimate tensile strength, the degradation rate of GFRP bars was significantly increased, as shown in Figure 4. This indicates that there were no micro-cracks in the matrix at a 20% stress level, and the corrosive mediums passing through the matrix were dominated by diffusion. However, a 40% stress level represents the "medium stress level" defined by Benmokrane et al. [10] This stress level is able to cause micro-cracks in the matrix, and the corrosive mediums can travel through the matrix faster and easier, accelerating the degradation of GFRP bars. The synergistic effect of stress and temperature causes the tensile strength degradation of GFRP bars to become more obvious when exposed to concrete pore water solution. After being exposed to concrete pore solution at 60 • C for 90 days, the tensile strength retention of GFRP bars tested under a 40% stress level is 24.5% lower than those of unstressed GFRP bars, but only 9.4% lower than those exposed to seawater solution. Columns 7 and 8 in Tables 4 and 5 present the experimental residual tensile strain and corresponding retention for the specimens. It can be observed that the degradation characteristics of tensile strain are similar to those of tensile strength. The tensile strength retentions of GFRP bars are summarized in Figure 4. It can be concluded from Figure 4 that a higher temperature accelerates the degradation rate of GFRP bars. When the temperature increased from 0 to 60 °C, the tensile strength retention of unstressed GFRP bars decreased by 16.4% and 25.0%, respectively, after being exposed to seawater and concrete pore water solution for 60 days. This indicates that the diffusion rate of the corrosive mediums is accelerated by higher temperatures. The curves of the GFRP bars under stress levels of 0% and 20% were close to each other, as shown in Figure 4. However, when the stress level increased to 40% of the ultimate tensile strength, the degradation rate of GFRP bars was significantly increased, as shown in Figure 4. This indicates that there were no micro-cracks in the matrix at a 20% stress level, and the corrosive mediums passing through the matrix were dominated by diffusion. However, a 40% stress level represents the "medium stress level" defined by Benmokrane et al. [10] This stress level is able to cause micro-cracks in the matrix, and the corrosive mediums can travel through the matrix faster and easier, accelerating the degradation of GFRP bars. The synergistic effect of stress and temperature causes the tensile strength degradation of GFRP bars to become more obvious when exposed to concrete pore water solution. After being exposed to concrete pore solution at 60 °C for 90 days, the tensile strength retention of GFRP bars tested under a 40% stress level is 24.5% lower than those of unstressed GFRP bars, but only 9.4% lower than those exposed to seawater solution. Columns 7 and 8 in Tables 4 and 5 present the experimental residual tensile strain and corresponding retention for the specimens. It can be observed that the degradation characteristics of tensile strain are similar to those of tensile strength.

Comparing Obtained Results with Literature
In order to verify the applicability of the GFRP bars investigated in this study, this paper compares the test data with results presented in literature. The selected literature includes similar test methods, compositions of solutions, and pH levels, to ensure reliability. Figure 5 compares the tensile strength retention of GFRP bars exposed to seawater solutions (G1 and G2 represent different types of GFRP bars in the literature). From Figure 5, it can be seen that except for the improved GFRP bars (G2) used by Kim et al. [24], the GFRP bars investigated in this study had a better residual tensile strength compared to other GFRP bars reported in the literature, indicating their good resistance to seawater. Figure 6 compares the tensile strength retention of GFRP bars exposed to concrete pore water solution. It can be seen from Figure 6 that, with or without stress, the tensile strength retention of GFRP bars Materials 2020, 13, 2341 8 of 18 used in this study is higher than that of the GFRP bars used in the literature for the same test duration. The degradation rate of GFRP bars used by Sen et al. [20] is very fast, and the degradation rate has already been increased significantly under a stress level of 15%, as shown in Figure 6. The reason for this situation is that Sen et al. [20] used GFRP bars with smaller diameters, so the corrosion mediums could penetrate into the GFRP bars more rapidly. Another reason is that the GFRP bars used were an early FRP product with low manufacturing maturity.
test methods, compositions of solutions, and pH levels, to ensure reliability. Figure 5 compares the tensile strength retention of GFRP bars exposed to seawater solutions (G1 and G2 represent different types of GFRP bars in the literature). From Figure 5, it can be seen that except for the improved GFRP bars (G2) used by Kim et al. [24], the GFRP bars investigated in this study had a better residual tensile strength compared to other GFRP bars reported in the literature, indicating their good resistance to seawater. Figure 6 compares the tensile strength retention of GFRP bars exposed to concrete pore water solution. It can be seen from Figure 6 that, with or without stress, the tensile strength retention of GFRP bars used in this study is higher than that of the GFRP bars used in the literature for the same test duration. The degradation rate of GFRP bars used by Sen et al. [20] is very fast, and the degradation rate has already been increased significantly under a stress level of 15%, as shown in Figure 6. The reason for this situation is that Sen et al. [20] used GFRP bars with smaller diameters, so the corrosion mediums could penetrate into the GFRP bars more rapidly. Another reason is that the GFRP bars used were an early FRP product with low manufacturing maturity.  tensile strength retention of GFRP bars exposed to seawater solutions (G1 and G2 represent different types of GFRP bars in the literature). From Figure 5, it can be seen that except for the improved GFRP bars (G2) used by Kim et al. [24], the GFRP bars investigated in this study had a better residual tensile strength compared to other GFRP bars reported in the literature, indicating their good resistance to seawater. Figure 6 compares the tensile strength retention of GFRP bars exposed to concrete pore water solution. It can be seen from Figure 6 that, with or without stress, the tensile strength retention of GFRP bars used in this study is higher than that of the GFRP bars used in the literature for the same test duration. The degradation rate of GFRP bars used by Sen et al. [20] is very fast, and the degradation rate has already been increased significantly under a stress level of 15%, as shown in Figure 6. The reason for this situation is that Sen et al. [20] used GFRP bars with smaller diameters, so the corrosion mediums could penetrate into the GFRP bars more rapidly. Another reason is that the GFRP bars used were an early FRP product with low manufacturing maturity.

Residual Modulus of Elasticity
Due to the low elastic modulus of GFRP bars, under the same reinforcement ratio, GFRP-RC members tend to exhibit larger deformation and result in wider cracks compared to steel-RC members [25][26][27]. The design of GFRP-RC members is usually governed by the serviceability limit state (deformation and cracking). Therefore, the reduction in the elastic modulus of GFRP bars during

Residual Modulus of Elasticity
Due to the low elastic modulus of GFRP bars, under the same reinforcement ratio, GFRP-RC members tend to exhibit larger deformation and result in wider cracks compared to steel-RC members [25][26][27]. The design of GFRP-RC members is usually governed by the serviceability limit state (deformation and cracking). Therefore, the reduction in the elastic modulus of GFRP bars during the service life is also an important factor. Figure 7 compares the elastic modulus of GFRP bars exposed to seawater and concrete pore water solutions at 23, 40, and 60 • C for 90 days at different stress levels. As shown in Figure 7, the elastic modulus of GFRP bars exposed to seawater solution and concrete pore aqueous solution is 0.95-1.04 times and 0.94-1.03 times that of unstressed GFRP bars, respectively. This result indicates that the sustained loads (20% and 40% stress levels) had almost no effect on the elastic modulus of GFRP bars, and degradation of the tensile strength and strain of GFRP bars remained at the same level, which agrees with previous literature [9,16,28].
(b) Figure 6. Comparison of experimental data (GFRP bars exposed to concrete pore solutions solution) from this study and previous literature: (a) unstressed bars; (b)stressed bars.

Residual Modulus of Elasticity
Due to the low elastic modulus of GFRP bars, under the same reinforcement ratio, GFRP-RC members tend to exhibit larger deformation and result in wider cracks compared to steel-RC members [25][26][27]. The design of GFRP-RC members is usually governed by the serviceability limit state (deformation and cracking). Therefore, the reduction in the elastic modulus of GFRP bars during the service life is also an important factor. Figure 7 compares the elastic modulus of GFRP bars exposed to seawater and concrete pore water solutions at 23, 40, and 60 °C for 90 days at different stress levels. As shown in Figure 7, the elastic modulus of GFRP bars exposed to seawater solution and concrete pore aqueous solution is 0.95-1.04 times and 0.94-1.03 times that of unstressed GFRP bars, respectively. This result indicates that the sustained loads (20% and 40% stress levels) had almost no effect on the elastic modulus of GFRP bars, and degradation of the tensile strength and strain of GFRP bars remained at the same level, which agrees with previous literature [9,16,28].

Comparison with ACI 440.1R-15
Creep tensile strain: According to the changes in the strain of GFRP bars over time at sustained stress, the creep characteristics can be analysed. Figure 8 shows this change for GFRP bars exposed to seawater and concrete pore solution at 60 °C. It can be observed that almost no significant changes in the strain occurred after 90 days. At the 20% and 40% stress level, the maximum increase in strain was only 3% and 7% of the initial strain, respectively. Nkurunziza et al. [16] and Robert et al. [6] forecasted the creep strain over the service life of concrete structures (75 years). It was found from their test results that the strain of GFRP bars only experienced an 8% increase in strain at the 38% stress level. This shows that the ACI 440.1R-15 [22] is very conservative in setting the creep failure stress of GFRP bars to 0.2 ffu (ffu is the design tensile strength of GFRP).
Tensile strength and strain: Figure 9 shows the ratios between the residual tensile strength/strain of conditioned GFRP bars and the design limits specified by ACI 440.1R-15 [22]. It can be seen from Figure 9 that, except for the GFRP bars at the 40% stress level of 90-day exposure to concrete pore solution, the residual tensile strength and tensile strain are slightly lower than the design limits, and the other conditioned GFRP bars have higher values that exceed the design limits. In fact, in the design of a GFRP-RC structure, the GFRP bars will remain below a stress level of 40%, which also proves the applicability of the GFRP bars used in this test.

Comparison with ACI 440.1R-15
Creep tensile strain: According to the changes in the strain of GFRP bars over time at sustained stress, the creep characteristics can be analysed. Figure 8 shows this change for GFRP bars exposed to seawater and concrete pore solution at 60 • C. It can be observed that almost no significant changes in the strain occurred after 90 days. At the 20% and 40% stress level, the maximum increase in strain was only 3% and 7% of the initial strain, respectively. Nkurunziza et al. [16] and Robert et al. [6] forecasted the creep strain over the service life of concrete structures (75 years). It was found from their test results that the strain of GFRP bars only experienced an 8% increase in strain at the 38% stress level. This shows that the ACI 440.1R-15 [22] is very conservative in setting the creep failure stress of GFRP bars to 0.2 f fu (f fu is the design tensile strength of GFRP).
Tensile strength and strain: Figure 9 shows the ratios between the residual tensile strength/strain of conditioned GFRP bars and the design limits specified by ACI 440.1R-15 [22]. It can be seen from Figure 9 that, except for the GFRP bars at the 40% stress level of 90-day exposure to concrete pore solution, the residual tensile strength and tensile strain are slightly lower than the design limits, and the other conditioned GFRP bars have higher values that exceed the design limits. In fact, in the design of a GFRP-RC structure, the GFRP bars will remain below a stress level of 40%, which also proves the applicability of the GFRP bars used in this test.  Tensile strength and strain: Figure 9 shows the ratios between the residual tensile strength/strain of conditioned GFRP bars and the design limits specified by ACI 440.1R-15 [22]. It can be seen from Figure 9 that, except for the GFRP bars at the 40% stress level of 90-day exposure to concrete pore solution, the residual tensile strength and tensile strain are slightly lower than the design limits, and the other conditioned GFRP bars have higher values that exceed the design limits. In fact, in the design of a GFRP-RC structure, the GFRP bars will remain below a stress level of 40%, which also proves the applicability of the GFRP bars used in this test.

Microstructure Analysis
The degradation rate of GFRP bars is relatively more rapid when exposed to a solution at 60 ° C. Therefore, only the microstructure analysis of GFRP bars after 90 days of exposure to a solution at 60 ° C was performed to understand the degradation mechanism of GFRP bars. Figure 10 shows the

Microstructure Analysis
The degradation rate of GFRP bars is relatively more rapid when exposed to a solution at 60 • C. Therefore, only the microstructure analysis of GFRP bars after 90 days of exposure to a solution at 60 • C was performed to understand the degradation mechanism of GFRP bars. Figure 10 shows the cross-sections of unconditioned GFRP bars at three magnifications (500×, 1000×, and 2000×). It can be seen from Figure 10 that there are no visible voids (debonding) between the fibers and the resin. For the conditioned GFRP bars, as shown in Figure 11, there is a distinct increase in the number of voids between the fibers and the resin, and the width and distribution range of the voids increase with the stress level. Compared with Figure 10, it can be found that Figure 11 shows that the glass fibers of the conditioned GFRP bars have not been significantly damaged, which indicates that the degradation of the GFRP bars in this study mainly occurred at the fiber/matrix interface.

Prediction Model Selection
The long-term performance prediction models for commonly used FRP composites are based on Arrheneius acceleration theory [4,18], and several prediction models have been proposed in the literature. After a careful analysis and comparison of the existing models [18,20,21], the prediction model proposed by Phani and Bose [29] could most accurately reflect the degradation process of FRP bars. This model was first used to predict the degradation of flexural strength of inter-laminar plates in wet and high-temperature environments. Moreover, the degradation mechanism for the model is assumed to be debonding at the fiber/matrix interface. This has also been confirmed by the SEM test results presented in this paper. The model is shown in Equation (1): where Y is the tensile strength retention (%), t is the exposure time, τ is the fitted parameter, and Y∞ is the tensile strength retention (%) at the exposure time of infinity. As the GFRP bars were exposed to the solution for only a short time in this study, the tensile strength of the GFRP bars was still decreasing after exposure, so it is impossible to know the tensile strength retention Y∞ after an infinite exposure time. Therefore, Equation (2) was used to predict the long-term performance of GFRP bars. Previous research (Chen et al. [4] and Wu et al. [19]) has confirmed the applicability and reliability of Equation (2) for unstressed FRP bars.

Arrheneius Theory
According to Arrheneius theory, the relationship between the degradation rate and the temperature can be expressed as where k is the degradation rate (1/time), A is a constant related to the material degradation process, Ea is the activation energy, R is the universal gas constant, and T is the temperature in Kelvin.
It should be noted that the primary assumption of Arrheneius theory is that the single dominant degradation mechanism of the material does not change with time and temperature during the

Prediction Model Selection
The long-term performance prediction models for commonly used FRP composites are based on Arrheneius acceleration theory [4,18], and several prediction models have been proposed in the literature. After a careful analysis and comparison of the existing models [18,20,21], the prediction model proposed by Phani and Bose [29] could most accurately reflect the degradation process of FRP bars. This model was first used to predict the degradation of flexural strength of inter-laminar plates in wet and high-temperature environments. Moreover, the degradation mechanism for the model is assumed to be debonding at the fiber/matrix interface. This has also been confirmed by the SEM test results presented in this paper. The model is shown in Equation (1): where Y is the tensile strength retention (%), t is the exposure time, τ is the fitted parameter, and Y ∞ is the tensile strength retention (%) at the exposure time of infinity. As the GFRP bars were exposed to the solution for only a short time in this study, the tensile strength of the GFRP bars was still decreasing after exposure, so it is impossible to know the tensile strength retention Y ∞ after an infinite exposure time. Therefore, Equation (2) was used to predict the long-term performance of GFRP bars. Previous research (Chen et al. [4] and Wu et al. [19]) has confirmed the applicability and reliability of Equation (2) for unstressed FRP bars.

Arrheneius Theory
According to Arrheneius theory, the relationship between the degradation rate and the temperature can be expressed as where k is the degradation rate (1/time), A is a constant related to the material degradation process, E a is the activation energy, R is the universal gas constant, and T is the temperature in Kelvin.
It should be noted that the primary assumption of Arrheneius theory is that the single dominant degradation mechanism of the material does not change with time and temperature during the exposure, but the rate of degradation is accelerated with temperature [30]. Therefore, Equation (3) can be rewritten as From Equation (4), the logarithm of time needed for a material property to reach a given value is a linear function of 1/T with the slope of the E a /R value.
According to Equation (3), the acceleration coefficient (AF) for a range of high temperature (T 1 ) to low temperature (T 0 ) can be calculated as where AF is the acceleration coefficient; t 1 and t 0 are the times required to reach a given tensile strength retention value at temperatures of T 1 and T 0 , respectively; c is a constant; and k 1 and k 0 are the degradation rates at temperatures of T 1 and T 0 , respectively. Therefore, the long-term tensile performance of GFRP bars at any given temperature can be predicted using Equation (2).

Prediction Procedure
The long-term performance prediction of GFRP bars consists of four steps.
Step 1: As shown in Figure 12, the obtained experimental data (Figure 4b) could be fit using Equation (2) and the linear regression coefficient τ and correlation coefficient R 2 were as shown in Table 6.
Materials 2020, 13, x FOR PEER REVIEW 14 of 19 exposure, but the rate of degradation is accelerated with temperature [30]. Therefore, Equation (3) can be rewritten as From Equation (4), the logarithm of time needed for a material property to reach a given value is a linear function of 1/T with the slope of the Ea/R value.
According to Equation (3), the acceleration coefficient (AF) for a range of high temperature (T1) to low temperature (T0) can be calculated as where AF is the acceleration coefficient; t1 and t0 are the times required to reach a given tensile strength retention value at temperatures of T1 and T0, respectively; c is a constant; and k1 and k0 are the degradation rates at temperatures of T1 and T0, respectively. Therefore, the long-term tensile performance of GFRP bars at any given temperature can be predicted using Equation (2).

Prediction Procedure
The long-term performance prediction of GFRP bars consists of four steps.
Step 1: As shown in Figure 12, the obtained experimental data (Figure 4b) could be fit using Equation (2) and the linear regression coefficient τ and correlation coefficient R 2 were as shown in Table 6.  Step 2: The parameter τ (Table 6) could be substituted in Equation (2) to obtain the time required to achieve strength retention of 50%, 60%, 70%, and 80% at 23, 40, and 60 • C under different stress levels. Then, the Arrhenius relationships were obtained by plotting the natural logarithm of time (t) against the inverse of temperature (1/T), and the data were fit using Equation (4), as shown in Figure 13. The fitted Ea/R and correlation coefficient R 2 are presented in Table 7. It was found that the correlation coefficients of all regression lines were close to 1, and the straight lines at different temperatures were parallel to each other. This indicates that the Arrhenius relationship can be used to describe the degradation rate of GFRP bars, as the degradation mechanism may not change with temperature and time during exposure in the range test.  Step 2: The parameter τ (Table 6) could be substituted in Equation (2) to obtain the time required to achieve strength retention of 50%, 60%, 70%, and 80% at 23, 40, and 60 °C under different stress levels. Then, the Arrhenius relationships were obtained by plotting the natural logarithm of time (t) against the inverse of temperature (1/T), and the data were fit using Equation (4), as shown in Figure 13. The fitted Ea/R and correlation coefficient R 2 are presented in Table 7. It was found that the correlation coefficients of all regression lines were close to 1, and the straight lines at different temperatures were parallel to each other. This indicates that the Arrhenius relationship can be used to describe the degradation rate of GFRP bars, as the degradation mechanism may not change with temperature and time during exposure in the range test.    Step 3: 20 • C was taken as a reference temperature and Equation (5) was used to calculate the AF values for other temperatures (23,40, and 60 • C) based on the reference temperature of 20 • C. The obtained AF values are listed in Table 8. Step 4: Once the AFs for 23, 40, and 60 • C were obtained, Figure 4b was transformed into Figure 14 by multiplying the exposure time at 23, 40, and 60 • C with the corresponding AFs. Then, Equation (2) was used to fit Figure 9 to obtain the master curves for tensile strength retention versus exposure time at 20 • C under different stress levels. The fitting parameters and correlation coefficients are listed in Table 9. It is worth noting that the τ values in Table 9 are very close to the τ values obtained at 23 • C presented in Table 6. This close correlation also confirms the validity of this procedure for predicting the long-term performance of GFRP bars. Therefore, the tensile strength retention of the unstressed and stressed GFRP bars at any exposure time could be predicted using Equation (4) and the τ values listed in Table 9.
Materials 2020, 13 Step 3: 20 °C was taken as a reference temperature and Equation (5) Table 8. Step 4: Once the AFs for 23, 40, and 60 °C were obtained, Figure 4b was transformed into Figure 14 by multiplying the exposure time at 23, 40, and 60 °C with the corresponding AFs. Then, Equation (2) was used to fit Figure 9 to obtain the master curves for tensile strength retention versus exposure time at 20 °C under different stress levels. The fitting parameters and correlation coefficients are listed in Table 9. It is worth noting that the τ values in Table 9 are very close to the τ values obtained at 23 °C presented in Table 6. This close correlation also confirms the validity of this procedure for predicting the long-term performance of GFRP bars. Therefore, the tensile strength retention of the unstressed and stressed GFRP bars at any exposure time could be predicted using Equation (4) and the τ values listed in Table 9. Master curves for GFRP bars exposed to concrete pore solutions at 20 °C.

Discussion on Predicted Results
As shown in Figure 14, it can be can that the time required to achieve 50% tensile strength retention under three stresses (0%, 20%, and 40%) at 20 °C in concrete pore solution is 1.5, 1.2, and 0.65 years, respectively. Mufti et al. [31] found that there is no degradation for GFRP bars in a GFRP-RC bridge with 5-8 years of service by using analytical methods (e.g., SEM). It can be seen that  Master curves for GFRP bars exposed to concrete pore solutions at 20 • C.

Discussion on Predicted Results
As shown in Figure 14, it can be can that the time required to achieve 50% tensile strength retention under three stresses (0%, 20%, and 40%) at 20 • C in concrete pore solution is 1.5, 1.2, and 0.65 years, respectively. Mufti et al. [31] found that there is no degradation for GFRP bars in a GFRP-RC bridge with 5-8 years of service by using analytical methods (e.g., SEM). It can be seen that the long-term performance results of GFRP bars directly exposed to corrosive solution can not truly reflect the degradation of GFRP bars in field concrete structures. Therefore, long-term data should be collected in field applications, and the correlation between the degradation of GFRP bars in accelerated tests and field applications needs to be investigated.
From the test results in Table 5 and Figure 4, it can be seen that a stress level of 20% does not seem to affect the degradation rate of GFRP bars. However, the degradation rates significantly increase with time, as shown in Figure 11. For strength retention of 50%, the required time under a 20% stress level is 25% shorter than that of unstressed GFRP bars. Davaloset al. [18] also found a similar phenomenon in their study on the durability of GFRP bars embedded in concrete. This may be due to the short exposure time of GFRP bars.
The activation energy (E a ) is defined as the energy required for a chemical reaction. The lower the required activation energy of a GFRP bar, the faster the GFRP bar degradation rate. For example, the activation energy of GFRP bars at a 40% stress level is 28.6% and 14% lower compared to lower stress levels of 0% and 20%, respectively, as shown in Table 7. The required activation energy of GFRP bars exposed to alkaline solution ranges from 12 to 41 KJ/mol [4,20,21,28], while the required activation energy of GFRP bars embedded in concrete ranges from 36 to 91 KJ/mol [7,18,32]. Therefore, the degradation rate of GFRP bars in concrete is much lower than those directly exposed to corrosive solution. It also shows that the long-term performance prediction results are conservative based on laboratory-simulated testing.

Conclusions
This experimental research has investigated the durability of new GFRP bars exposed to laboratory-simulated seawater and concrete environments. The main conclusions are as follows: (1) The degradation rate of GFRP bars in seawater solution is significantly lower than that in concrete pore solution, and the difference in degradation rate becomes more obvious as the temperature and stress level increase; (2) A stress level of 40% is able to cause micro-cracks in the resin matrix of the GFRP bars, accelerating the degradation rate of the GFRP bar. In the design of GFRP-RC structures, the stress level in the GFRP bars should be limited to under 40% of the ultimate tensile strength; (3) It has been found that a lower stress level of 20% does not affect the degradation rate of GFRP bars. However, based on the prediction analysis of the service life of GFRP bars, it has been found that for the same tensile strength retention, the time required for a stress level of 20% is significantly reduced compared to a 0% stress level. With the increase of the exposure time of GFRP bars, the damage of GFRP bars increases continuously and the degradation rate of damage GFRP bars is accelerated by a 20% stress level; (4) The elastic modulus of GFRP bars is not affected by the corrosive environment and stress level, and the degradation rate for tensile strength and tensile strain tends to remain at the same level; (5) The residual tensile properties of all GFRP bars can meet the requirements of ACI 440.1R-15, except for those tested under a 40% stress level in 60 • C concrete pore solution for 90 days; (6) Since the GFRP bars were directly exposed to the corrosion solution, the long-term performance prediction results of the GFRP bars obtained using Arrheneius theory are considered to be conservative. Future research is needed to establish the correlation between the degradation of GFRP bars in laboratory-simulated and field conditions, in order to provide practical guidelines for GFRP-RC structures.