Effect of Temperature on Material Properties of Carbon Fiber Reinforced Polymer (CFRP) Tendons: Experiments and Model Assessment

Material properties at elevated temperatures are important factors in the fire safety design and numerical analysis of concrete members strengthened with fiber reinforced polymer (FRP) composites. Most of the previous research mainly focused on tensile strength and elastic modulus in conventional steady state temperature tests. However, the transient state test method is more realistic for strengthening concrete structures. At the same time, the coefficient of thermal expansion of FRP composites is also one of the important factors affecting concrete members at elevated temperatures. This paper presents a detailed experimental investigation on the longitudinal thermal expansion deformation, and the mechanical properties of carbon FRP (CFRP) tendons with 8 mm diameter in the steady state and transient state. The results indicate that longitudinal deformation of CFRP tendons is negative at high temperature; in addition, the transient state test results of CFRP tendons are slightly higher than the steady state test results. The final part of this paper assesses the accuracy of different empirical models. Furthermore, a new equation calculating the properties of CFRP composites at elevated temperatures is presented with the numerical fitting technique, which is in good agreement with the experimental results.


Introduction
In recent years, the application of fiber reinforced polymer (FRP) composites has increased significantly in civil engineering structures around the world. It is mainly attributed to the fact that FRP composites have many advantages compared with steel, such as high strength-to-weight ratio, corrosion resistance, fatigue resistance, and non-magnetic properties. Although numerous investigations have shown that FRP composites are effective materials for concrete members [1,2], many challenges still exist in the increasing application of FRP composites. One of these challenges is to fully understand the material properties of FRP composites at elevated temperatures. Due to the fact that FRP composites are typically produced from the thermosetting polymer matrix, their mechanical properties deteriorate severely when exposed to fire. Even though FRP composites embedded in the concrete do not burn due to lack of oxygen, the resin will soften and decompose when the temperature exceeds the glass transition temperature (T g ) and the decomposition temperature (T d ). The softening and decomposition of the resin not only reduces the strength of the resin itself, but also weakens the bonding effect of fibers, resulting in the rapid reduction of the strength of FRP composites.

Material Characteristics and Specimen Details
In the current research, CFRP tendons with 8 mm diameter and smooth surface from Jiangsu Hengshen Co. Ltd. (Zhenjiang, China) were tested. They were made of T700 continuous carbon fibers impregnated in thermosetting epoxy resin with a fiber content of 65%. The average tensile strength and modulus of CFRP tendons were 2070 MPa and 156 GPa, respectively, at room temperatures. According to E1356-08 [18], The T g value is usually the midpoint temperature of resin glass transition. The T d value can be considered as the temperature point where the weight loss of CFRP composites reaches the peak (the peak point of DTG curve) [19]. Based on the previous experimental study, the T g and T d of epoxy resins from the same batches were 126 • C and 405 • C, respectively [20]. Other properties of CFRP tendons used for testing are listed in Table 1. In the tensile test, special anchorage was needed at both ends of the specimen due to the smooth surface of the tendons. So far, the clip anchorage is the most reliable anchorage for CFRP tendons, which can effectively prevent the slippage between the tendon and clip by pre-pressing ( Figure 1). Details of the anchorage system for CFRP specimen were presented in Zhou [20].
Materials 2018, 11, x FOR PEER REVIEW 3 of 12 In the current research, CFRP tendons with 8 mm diameter and smooth surface from Jiangsu Hengshen Co. Ltd. (Zhenjiang, China) were tested. They were made of T700 continuous carbon fibers impregnated in thermosetting epoxy resin with a fiber content of 65%. The average tensile strength and modulus of CFRP tendons were 2070 MPa and 156 GPa, respectively, at room temperatures. According to E1356-08 [18], The Tg value is usually the midpoint temperature of resin glass transition. The Td value can be considered as the temperature point where the weight loss of CFRP composites reaches the peak (the peak point of DTG curve) [19]. Based on the previous experimental study, the Tg and Td of epoxy resins from the same batches were 126 °C and 405 °C, respectively [20]. Other properties of CFRP tendons used for testing are listed in Table 1. In the tensile test, special anchorage was needed at both ends of the specimen due to the smooth surface of the tendons. So far, the clip anchorage is the most reliable anchorage for CFRP tendons, which can effectively prevent the slippage between the tendon and clip by pre-pressing ( Figure 1). Details of the anchorage system for CFRP specimen were presented in Zhou [20].

Testing Setups
The tests were performed at the Testing Laboratory of Civil Engineering at Southeast University. A UTM5305 Material Test System with a loading capacity of 300 kN was used in this study, as shown in Figure 2. The tendons used in the tests were all 800 mm long, of which 340 mm was exposed to heat inside the furnace. The elongation of CFRP tendons was measured by a high temperature extensometer with a gauge length of 50 mm. In order to obtain more accurate results, the temperature distribution on the surface of the tendons was measured by six thermocouples fixed on the surface of the tendon.

Testing Setups
The tests were performed at the Testing Laboratory of Civil Engineering at Southeast University. A UTM5305 Material Test System with a loading capacity of 300 kN was used in this study, as shown in Figure 2. The tendons used in the tests were all 800 mm long, of which 340 mm was exposed to heat inside the furnace. The elongation of CFRP tendons was measured by a high temperature extensometer with a gauge length of 50 mm. In order to obtain more accurate results, the temperature distribution on the surface of the tendons was measured by six thermocouples fixed on the surface of the tendon.

Testing Procedure
The tests of the material properties of CFRP tendons at elevated temperatures were divided into three series, namely thermal expansion test, steady state test, and transient state test. A total of 42 specimens were performed in this study, as shown in Table 2.  Thermal expansion test  3  Steady state test  24  Transient state test  15  Total  42 (1) Thermal expansion tests

Type of Tests Number of Specimens
In the thermal expansion test, one end of the specimen was gripped by the upper grip, and the other end was relaxed. In order to make the outer and inner temperatures of the tendons to be consistent, the furnace temperature (target temperature) was held constant for about half an hour after it was raised to a target temperature ( Figure 3) [10]. The surface temperature of CFRP tendons detected by six thermocouples is presented in Figure 4. It shows that the temperature distribution at the specimen surface was not uniform. The temperature difference within the 50 mm range of the extensometer was small, and the maximum temperature value appeared near the no. 3 point. Thus, the temperature at no. 3 point was chosen as the failure temperature. Finally, when the temperature was stable, the longitudinal deformations of CFRP tendons were recorded by the extensometer.

Testing Procedure
The tests of the material properties of CFRP tendons at elevated temperatures were divided into three series, namely thermal expansion test, steady state test, and transient state test. A total of 42 specimens were performed in this study, as shown in Table 2. (1) Thermal expansion tests In the thermal expansion test, one end of the specimen was gripped by the upper grip, and the other end was relaxed. In order to make the outer and inner temperatures of the tendons to be consistent, the furnace temperature (target temperature) was held constant for about half an hour after it was raised to a target temperature ( Figure 3) [10]. The surface temperature of CFRP tendons detected by six thermocouples is presented in Figure 4. It shows that the temperature distribution at the specimen surface was not uniform. The temperature difference within the 50 mm range of the extensometer was small, and the maximum temperature value appeared near the no. 3 point. Thus, the temperature at no. 3 point was chosen as the failure temperature. Finally, when the temperature was stable, the longitudinal deformations of CFRP tendons were recorded by the extensometer.

Testing Procedure
The tests of the material properties of CFRP tendons at elevated temperatures were divided into three series, namely thermal expansion test, steady state test, and transient state test. A total of 42 specimens were performed in this study, as shown in Table 2. (1) Thermal expansion tests In the thermal expansion test, one end of the specimen was gripped by the upper grip, and the other end was relaxed. In order to make the outer and inner temperatures of the tendons to be consistent, the furnace temperature (target temperature) was held constant for about half an hour after it was raised to a target temperature ( Figure 3) [10]. The surface temperature of CFRP tendons detected by six thermocouples is presented in Figure 4. It shows that the temperature distribution at the specimen surface was not uniform. The temperature difference within the 50 mm range of the extensometer was small, and the maximum temperature value appeared near the no. 3 point. Thus, the temperature at no. 3 point was chosen as the failure temperature. Finally, when the temperature was stable, the longitudinal deformations of CFRP tendons were recorded by the extensometer.   (2) Steady state tests In order to avoid the influence of free expansion, the installation process at the beginning of the steady state test was similar to that of expansion test. The specimens were exposed to the target temperature for additional 30 min as soak time. The temperature distribution on the surface of the tendons was similar to that shown in Figure 4. Then, the lower grip was tightened. According to the ACI 440.3R [21], the specimens were loaded with displacement control at a rate of 3 mm/min (585 MPa/min) until failure.

(3) Transient state tests
In the transient test, the specimens were first applied to a target load at a rate of 3 mm/min (585 MPa/min). After load stabilization for 1 min, the temperature was raised at a heating rate of 20 °C/min while maintaining the pre-set load until failure occurred. According to GB 50608-2010 [22], the allowable service stress of CFRP bars is limited to not more than 71% of the ultimate tensile strength. Therefore, 29-67% of the ultimate tensile strength at room temperature was chosen as the stress levels for the loaded specimens in this study.

Thermal Expansion Tests
According to the ACI 440.1R [23], the coefficient of thermal expansion (CTE) of CFRP tendons is between −9.0 × 10 −6 /°C and 0.0 × 10 −6 /°C in the longitudinal direction. However, the variation of CTE with temperatures is not clear. The experimental results show that the longitudinal deformation of CFRP tendons decreased with the increase of temperature ( Figure 5), which verifies the fact that the FRP composites shortened along the fiber direction at elevated temperatures [24]. This is mainly attributed to fact that carbon fibers shrink at elevated temperatures in the longitudinal direction [25]. With the increase of temperature, the shrinkage of carbon fibers dominated the longitudinal deformation of CFRP tendons due to the softening of the resin. When the temperature was low, the CTE decreased slowly. As the temperature rose, the resin of FRP composites began to soften, and the CTE of CFRP tendons decreased to a larger negative value. Especially, when the temperature exceeded 200 °C, the CTE decreased rapidly. Finally, the longitudinal deformation of CFRP tendons was unstable after 300 °C due to the decomposition of the resin, resulting in the failure of obtaining its thermal expansion properties.
Based on the experimental data, the longitudinal CTE (10 −6 /°C) of CFRP tendons is proposed in the form of polynomial function (Equation (1)). The obtained curve is in good agreement with the experimental results as shown in Figure 5.  (2) Steady state tests In order to avoid the influence of free expansion, the installation process at the beginning of the steady state test was similar to that of expansion test. The specimens were exposed to the target temperature for additional 30 min as soak time. The temperature distribution on the surface of the tendons was similar to that shown in Figure 4. Then, the lower grip was tightened. According to the ACI 440.3R [21], the specimens were loaded with displacement control at a rate of 3 mm/min (585 MPa/min) until failure.

(3) Transient state tests
In the transient test, the specimens were first applied to a target load at a rate of 3 mm/min (585 MPa/min). After load stabilization for 1 min, the temperature was raised at a heating rate of 20 • C/min while maintaining the pre-set load until failure occurred. According to GB 50608-2010 [22], the allowable service stress of CFRP bars is limited to not more than 71% of the ultimate tensile strength. Therefore, 29-67% of the ultimate tensile strength at room temperature was chosen as the stress levels for the loaded specimens in this study.

Thermal Expansion Tests
According to the ACI 440.1R [23], the coefficient of thermal expansion (CTE) of CFRP tendons is between −9.0 × 10 −6 / • C and 0.0 × 10 −6 / • C in the longitudinal direction. However, the variation of CTE with temperatures is not clear. The experimental results show that the longitudinal deformation of CFRP tendons decreased with the increase of temperature ( Figure 5), which verifies the fact that the FRP composites shortened along the fiber direction at elevated temperatures [24]. This is mainly attributed to fact that carbon fibers shrink at elevated temperatures in the longitudinal direction [25]. With the increase of temperature, the shrinkage of carbon fibers dominated the longitudinal deformation of CFRP tendons due to the softening of the resin. When the temperature was low, the CTE decreased slowly. As the temperature rose, the resin of FRP composites began to soften, and the CTE of CFRP tendons decreased to a larger negative value. Especially, when the temperature exceeded 200 • C, the CTE decreased rapidly. Finally, the longitudinal deformation of CFRP tendons was unstable after 300 • C due to the decomposition of the resin, resulting in the failure of obtaining its thermal expansion properties.
Based on the experimental data, the longitudinal CTE (10 −6 / • C) of CFRP tendons is proposed in the form of polynomial function (Equation (1)). The obtained curve is in good agreement with the experimental results as shown in Figure 5.

Steady State Tests
In the steady state tests, the experimental results of CFRP tendons at various temperatures are listed in Table 3. It can be seen that maximum temperature on the surface (no. 3 point) of CFRP tendons was less than the furnace temperature (target temperature). When the temperature in the furnace exceeded 500 °C, the surface temperature of CFRP tendons was higher than the furnace temperature because of the combustion phenomenon of CFRP composites. The ultimate tensile strength of CFRP tendons at room temperature was 2070 MPa, but the strength decreased with the increase of temperature, as shown in Figure 6. When the temperature reached and exceeds Tg and Td, the tensile strength of CFRP tendons decreased dramatically. The main reason is that the softening and decomposition of the resin not only reduced the strength of the resin itself, but also weakened the bonding effect of fibers, resulting in the rapid reduction of the strength of FRP composites. This phenomenon further verifies the effect of temperatures on the mechanical properties of FRP composites due to different state of the resin at elevated temperatures. In order to determine the fire resistance of concrete members strengthened with FRP composites, the temperature at which the composites lose 50% of its tensile strength was used as the critical temperature of FRP composites by Wang [26]. Therefore, the Tc of CFRP tendons was determined to be 324 °C by the linear interpolation method according to Table 3.

Steady State Tests
In the steady state tests, the experimental results of CFRP tendons at various temperatures are listed in Table 3. It can be seen that maximum temperature on the surface (no. 3 point) of CFRP tendons was less than the furnace temperature (target temperature). When the temperature in the furnace exceeded 500 • C, the surface temperature of CFRP tendons was higher than the furnace temperature because of the combustion phenomenon of CFRP composites. The ultimate tensile strength of CFRP tendons at room temperature was 2070 MPa, but the strength decreased with the increase of temperature, as shown in Figure 6. When the temperature reached and exceeds T g and T d , the tensile strength of CFRP tendons decreased dramatically. The main reason is that the softening and decomposition of the resin not only reduced the strength of the resin itself, but also weakened the bonding effect of fibers, resulting in the rapid reduction of the strength of FRP composites. This phenomenon further verifies the effect of temperatures on the mechanical properties of FRP composites due to different state of the resin at elevated temperatures. In order to determine the fire resistance of concrete members strengthened with FRP composites, the temperature at which the composites lose 50% of its tensile strength was used as the critical temperature of FRP composites by Wang [26]. Therefore, the T c of CFRP tendons was determined to be 324 • C by the linear interpolation method according to Table 3.   Figure 7a shows the failure mode of CFRP tendons in the steady state test. Due to the limitation of the location of the furnace, the heating area was not in the middle of the specimen. However, all specimens failed within the length of the specimen, and the clip anchorage provided effective anchoring effect. In Figure 7a, it is obvious that the temperature distribution of CFRP tendons was  Figure 7a shows the failure mode of CFRP tendons in the steady state test. Due to the limitation of the location of the furnace, the heating area was not in the middle of the specimen. However, all specimens failed within the length of the specimen, and the clip anchorage provided effective anchoring effect. In Figure 7a, it is obvious that the temperature distribution of CFRP tendons was not uniform, and the deterioration of resin was more serious near no. 3 point. As the temperature increased, the bonding effect of the resin decreased gradually. This resulted in a gradual separation of the carbon fibers and the resin. At 200-300 • C, there were obvious longitudinal cracks on the surface of the tendons, and the CFRP tendons were divided into several bunches of fibers. When the temperature exceeded 500 • C, only carbon fibers in the CFRP tendons were retained due to decomposition of the resin. At this time, the majority of the mechanical properties of CFRP tendons was lost.  Figure 7a shows the failure mode of CFRP tendons in the steady state test. Due to the limitation of the location of the furnace, the heating area was not in the middle of the specimen. However, all specimens failed within the length of the specimen, and the clip anchorage provided effective anchoring effect. In Figure 7a, it is obvious that the temperature distribution of CFRP tendons was not uniform, and the deterioration of resin was more serious near no. 3 point. As the temperature increased, the bonding effect of the resin decreased gradually. This resulted in a gradual separation of the carbon fibers and the resin. At 200-300 °C, there were obvious longitudinal cracks on the surface of the tendons, and the CFRP tendons were divided into several bunches of fibers. When the temperature exceeded 500 °C, only carbon fibers in the CFRP tendons were retained due to decomposition of the resin. At this time, the majority of the mechanical properties of CFRP tendons was lost.   Table 4 presents failure temperature of the loaded specimens in transient state tests. It should be noted that the failure temperature gradually decreased with the increase of the load level. Corresponding to the steady state test results, the critical temperature of CFRP tendons was 341 °C in the transient test. In theory, the presence of initial stress may intensify and accelerate degradation of CFRP tendons at elevated temperature. However, as shown in Figure 8, most of the data values in transient state tests were slightly higher than those in steady state tests. There are two main reasons: (1) In the steady state test, the degradation of tensile strength may have been further accelerated by the additional 30 min soaking time [27]; (2) In the transient test, the temperature at the surface part of CFRP tendons was higher than that in the internal part.   Table 4 presents failure temperature of the loaded specimens in transient state tests. It should be noted that the failure temperature gradually decreased with the increase of the load level. Corresponding to the steady state test results, the critical temperature of CFRP tendons was 341 • C in the transient test. In theory, the presence of initial stress may intensify and accelerate degradation of CFRP tendons at elevated temperature. However, as shown in Figure 8, most of the data values in transient state tests were slightly higher than those in steady state tests. There are two main reasons: (1) In the steady state test, the degradation of tensile strength may have been further accelerated by the additional 30 min soaking time [27]; (2) In the transient test, the temperature at the surface part of CFRP tendons was higher than that in the internal part.  In the transient tests, the failure mode of CFRP tendons is shown in Figure 7b. For the specimens under loads of 50-70 kN, the time from the beginning of the test to the failure of CFRP tendons was shorter. The tendons were still characterized by brittle fracture at failure. This was mainly attributed to the fact that most of the undecomposed resin still had an effective bonding effect, while for the specimens under loads of 30-40 kN, the resin was obviously decomposed, and the duration time of the test was longer before the failure of CFRP tendons.

Description of Models
In the present section, the experimental results of CFRP tendons at elevated temperatures are used to evaluate the applicability and accuracy of the models proposed in different literatures. In order to fully reveal the influence of thermo-physical responses on thermo-mechanical properties, a new model is proposed with the form of exponential function.
According to the characteristics of FRP composites at elevated temperatures, the temperature (Tref) at which the tensile strength and Young's modulus vanish and the room temperature (T0) are applied in the expression by Gu [28]. The equation is provided as follows: where P(T) and P0 represents the mechanical property at T (°C) and room temperature, respectively; The value of the power law index m can be chosen between 0 and 1. Subsequently, in order to fit the curve simply and accurately, a cubic polynomial function was proposed by Liu [29]. The glass temperature was successfully applied in the expression as follows: More recently, in the study of mechanical properties of GFRP composites at elevated temperatures, Correia [12] proposed an exponential function model which was derived from Gompertz statistical distribution theory. In the transient tests, the failure mode of CFRP tendons is shown in Figure 7b. For the specimens under loads of 50-70 kN, the time from the beginning of the test to the failure of CFRP tendons was shorter. The tendons were still characterized by brittle fracture at failure. This was mainly attributed to the fact that most of the undecomposed resin still had an effective bonding effect, while for the specimens under loads of 30-40 kN, the resin was obviously decomposed, and the duration time of the test was longer before the failure of CFRP tendons.

Description of Models
In the present section, the experimental results of CFRP tendons at elevated temperatures are used to evaluate the applicability and accuracy of the models proposed in different literatures. In order to fully reveal the influence of thermo-physical responses on thermo-mechanical properties, a new model is proposed with the form of exponential function.
According to the characteristics of FRP composites at elevated temperatures, the temperature (T ref ) at which the tensile strength and Young's modulus vanish and the room temperature (T 0 ) are applied in the expression by Gu [28]. The equation is provided as follows: where P(T) and P 0 represents the mechanical property at T ( • C) and room temperature, respectively; The value of the power law index m can be chosen between 0 and 1. Subsequently, in order to fit the curve simply and accurately, a cubic polynomial function was proposed by Liu [29]. The glass temperature was successfully applied in the expression as follows: More recently, in the study of mechanical properties of GFRP composites at elevated temperatures, Correia [12] proposed an exponential function model which was derived from Gompertz statistical distribution theory. P(T) = (1 − e Be C×T ) × (P 0 − P r ) + P r (4) in which coefficients B and C are determined by experimental data; P 0 and P r are the mechanical properties of the composites at ambient temperature and in the decomposed state, respectively. As mentioned in the previous section, the critical temperature can be used as a criterion for evaluating the fire resistance of strengthened concrete members. On the basis of Gibson [14] model, Yu [6] further proposed the following hyperbolic tangent function model. where P R is the relaxed (high temperature) value of the mechanical properties; k is a constant describing the extent of relaxation; T c is critical temperature of CFRP tendons. In order to reveal the thermophysical properties of CFRP composites at elevated temperatures, a modified model is suggested by the authors, which is based on the exponential function model. Equation (6) contains the comprehensive effects of fiber content, resin softening, and decomposition on the mechanical properties of CFRP composites at elevated temperatures.
where k 1 and k 2 are constants determined by fitting the experimental data; V f is the volume fractions of fibers.  (4), Correia defined P r as the mechanical property after glass transition (but before decomposition), corresponding to the strength in the leathery state. However, the experimental results in this paper show that it was more reasonable to refer to P r as the strength in the decomposed state. Thus, P r had the same meaning as P R in Yu's model. As discussed above, the resin of CFRP tendons was completely decomposed in steady state and transient tests at 518 • C and 468 • C, respectively. Therefore, the strength at 518 • C and 468 • C was used as P r or P R .

Results and Discussion
In statistics, the coefficient of determination (R-Square) can represent the accuracy of a fitting model. The normal range of its value is between 0 and 1. The closer R-Square approaches 1, the better the model fits with the data. The comparison of different models shows that Liu model and the model proposed in this paper have better fitting accuracy. At the same time, the proposed model not only has fewer parameters, but also reflects the thermophysical properties of CFRP composites. The strength degradation curves derived from different empirical relations are plotted in Figure 9. All the models were in good agreement with the experimental data in the range of temperatures analyzed. It is a fact that the mechanical properties of CFRP composites after the decomposition of the resin approached zero. However, when the temperature exceeded 500 • C, the results fitted by the Gu model and the Liu model became negative, which did not satisfy with the physical truth. Again, it confirms the applicability and accuracy of the proposed model in this paper for calculating the mechanical properties of CFRP composites at elevated temperatures. the resin approached zero. However, when the temperature exceeded 500 °C, the results fitted by the Gu model and the Liu model became negative, which did not satisfy with the physical truth. Again, it confirms the applicability and accuracy of the proposed model in this paper for calculating the mechanical properties of CFRP composites at elevated temperatures.

Conclusions
In this paper, detailed experimental studies are presented on the material properties of CFRP tendons at elevated temperatures. Based on the experimental data obtained, the fitting accuracy of different strength prediction methods is evaluated. The main conclusions can be drawn as follows: (

Conclusions
In this paper, detailed experimental studies are presented on the material properties of CFRP tendons at elevated temperatures. Based on the experimental data obtained, the fitting accuracy of different strength prediction methods is evaluated. The main conclusions can be drawn as follows: (1) In the thermal expansion tests, experimental results reported herein confirmed that the longitudinal deformation of CFRP tendons is shrinkage at elevated temperatures. The coefficient of thermal expansion decreases with the increase of temperature. The CTE of CFRP tendons predicted by polynomial function is in good agreement with the experimental data before 300 • C. (2) In the steady state tests, the tensile strength of CFRP tendons shows a characteristic of degradation with temperature. Especially, in the temperature range of glass transformation and decomposition of the resin, it decreases rapidly. When heated 324 • C, the retained strength of CFRP tendons is 50%. Hence the critical temperature of CFRP tendons in the steady state tests is 324 • C.