In the study, 51 standard concrete cylinders with a diameter of 150 mm and a height of 300 mm were cast and tested under monotonic axial compression. The specimens were divided into three groups: one unheated case used as a control group, four cases of heat-damaged concrete without strengthening, and 12 cases of heat-damaged concrete with different layers of BFRP jackets. For each design case, three identical specimens were prepared. The variables of this study were the target temperatures (20°C, 200°C, 400°C, 600°C, and 800°C) and the number of layers of the BFRP jackets (0, 2, 3, and 4). Each specimen is named by a label consisting of two letters (T stands for the target temperature and L stands for the number of BFRP layers), followed by a number to distinguish nominally identical specimens. For instance, “T400-L2-1” denotes the first specimen of three duplicates that was heat-damaged after exposure to 400°C and then confined with two layers of BFRP jackets.

All the specimens were cast with the same concrete mix design. The concrete mix included sand, crushed granitic rocks, ordinary Portland cement (OPC), and water. River sand was used as the fine aggregate, and crushed granitic rocks with a maximum size of 10 mm were used as the coarse aggregate. The mixing proportion of 1 cubic meter of concrete consisted of 438 kg cement, 704 kg fine aggregate, 1,103 kg coarse aggregate, and 206 kg water. The compressive strength and elastic modulus of concrete were determined to be 45.1 MPa and 21.9 GPa, respectively. The basalt fibers used in the study were in the form of continuous unidirectional fabric sheets with a nominal thickness of 0.121 mm and a width of 300 mm. The tensile strength and elastic modulus of the BFRP jackets were obtained according to the flat coupon tests recommended by the ASTM D3039 standard (ASTM, 2017). Table 1 lists the test results and the specified mechanical properties of BFRP sheets provided by the manufacturer.

Four representative high temperatures (i.e., 200°C, 400°C, 600°C, and 800°C) were used in this study to induce the light (after exposure to 200°C), moderate (after exposure to 400°C or 600°C), and severe (after exposure to 800°C) damage of concrete. Three standard concrete cylinders (unheated) were tested at 20°C as a control group. Prior to heat exposure, the remaining 48 concrete cylinders were oven-dried at 105°C for 24 h to reduce the water content of the concrete. It is well recognized that the high water content in concrete may cause explosive spalling of concrete during heating (Gao et al., 2013; Yang et al., 2019), and RC columns that have been serviced for many years generally exhibit much lower water content than fresh concrete. Therefore, the purpose of pre-drying treatment is to reduce the moisture content of fresh concrete so that the concrete reaches a relatively actual moisture content like that during service. The concrete cylinders were exposed to a high temperature of 200°C, 400°C, 600°C, or 800°C in an electric furnace. A heating rate of 5°C/min was set for the furnace temperature, which was monitored by a thermocouple placed in the chamber. After the high temperature was obtained, it was maintained for 2.5 h to endure that the entire cross-section of the concrete cylinder achieved an almost uniform high temperature. Such a heating period was determined by numerical analysis of heat transfer according to the finite element model proposed by the authors (Gao et al., 2013; Gao et al., 2015; Gao et al., 2018). After finishing the heating process, the door of the furnace was opened, and the concrete cylinders were left in the chamber to be cooled down naturally. Post-heating inspection of all the cylinders was carefully carried out. It was obvious that as the high temperature increased, some slight micro-cracks appeared on the surfaces of the concrete cylinders, and no severe explosive spalling occurred. Only some slight concrete spalling was observed for the cylinders after exposure to 800°C. The reason may be due to the positive impact on the concrete cylinders during the early preheating process.

Before applying the BFRP jackets, an epoxy mortar (Sikadur®-31 CF Normal) was first applied to repair the heat-damaged concrete cylinders to fill the concrete cracks caused by local concrete spalling. After that, the repaired specimens were cured in a laboratory environment for 7 days to make the epoxy mortar reach a sufficient hardening state according to the recommendations provided by the design specification (GB, 2006). Then, the concrete surfaces of the cylinders were burnished with abrasive paper, and acetone was used to wipe the surfaces of the concrete cylinders. Any dust or pollutants were removed by compressed air. The epoxy adhesive used to bond the BFRP jackets was a two-part resin consisting of an epoxy (Part A) and a hardener (Part B), and the mixing ratio was 3:2 by volume. The tensile strength and elastic modulus of the epoxy resin provided by the material manufacturer are 40 MPa and 2.5 GPa, respectively. A thin layer of two-part epoxy resin was applied on the cleaned concrete surface as a primer, then the BFRP sheets were wrapped along the hoop direction, and another layer of epoxy resin was brushed on the surface of the BFRP sheets and served as a polymer matrix. The BFRP sheets were continuously wrapped around the heat-damaged cylinders with the primary fibers orienting in the hoop direction, and there were additional 150-mm-length BFRP sheets in the overlapping zone. Two additional 20 mm BFRP jackets were used at both ends of the column to prevent possible premature failure near the two ends.

The mechanical properties of BFRP sheets were measured by tensile coupon tests following the ACI testing code (ASTM, 2017). Table 1 lists the detailed values of the tensile strength and elastic modulus of the BFRP sheets obtained from the flat coupon tests, in which the mechanical properties supplied by the material manufacturer were provided as a reference. The BFRP-confined heat-damaged specimens were cured under laboratory conditions for 7 days to ensure that the epoxy adhesive was fully cured according to the provisions supplied by the ACI design code (ACI, 2008). The prepared specimens were further placed in the laboratory at least 2 weeks before the axial compression tests.

The specimens were tested under axial compression using a 2,000 kN loading system. A displacement-controlled method was adopted for the axial compression tests, in which the specimens were loaded at a constant displacement rate of 0.2 mm/min. Before testing, each column was capped with dental plaster at its top and bottom to achieve parallel surfaces and help the load be distributed uniformly. Two clip-on extensometers with a gauge length of 100 mm were placed at the middle height of each specimen to measure the axial deformations. Lateral strain responses of BFRP jackets were sufficiently measured by seven strain gauges with a 10-mm gauge length, which were distributed at the mid-height of the cylinder along the hoop direction, as shown in Figure 1. Two strain gauges were located in the 150-mm overlapping zone, and the remaining five were evenly distributed in the non-overlapping zone. All the test data (i.e., load force, displacement, and strain measurements) were monitored and recorded by an automatic data acquisition system.

It is well known that the lateral confinement provided by the FRP jackets is proportional to the FRP hoop rupture strain as follows: f l = 2 n E f r p t f r p ε h o o p D (1) where f l is the confinement pressure provided by the FRP jackets, n is the number of the FRP jacket layers,   E f r p is the elastic modulus of FRP jackets obtained from the flat coupon tests, t f r p is the nominal thickness of the FRP jackets, ε h o o p   is the average hoop strain in the overlapping zone, D is the diameter of the confined concrete core. Since the tensile force of the FRP jackets can be regarded as a constant, the difference in the recorded strain responses between the overlapping zone and the non-overlapping zone can be determined by the following theoretical strain ratio: ε o v e r l a p ε n o n − o v e r l a p = n n o n − o v e r l a p n o v e r l a p (2) where ε o v e r l a p is the average hoop strain of the overlapping zone, ε n o n − o v e r l a p is the average hoop strain of the non-overlapping zone (i.e., ε h , ave in Table 2), n o v e r l a p is the number of FRP layers in the overlapping zone, and n n o n − o v e r l a p is the number of FRP layers in the non-overlapping zone. To further examine the strain distribution of BFRP jackets in the BFRP-confined heat-damaged concrete, Figure 4 shows the details of the measurements recorded by the seven strain gauges for the tested specimens after exposure to 200°C and 800°C. It is seen that the minimum strain ε m i n is obtained in the overlapping zone, which occurs at the strain gauge SG1 near the end of the jackets. Comparatively speaking, the lateral strain measured by the strain gauge SG2 in the overlapping zone is much closer to that measured at the strain gauge SG3 in the non-overlapping zone. In other words, the strain values measured by the two strain gauges in the overlapping zone are different, and the difference between them increases with the distance from the edge of the BFRP jackets. This observation was also reported by Lam and Teng (Lam and Teng, 2004).

Table 4 shows the strain difference recorded by the two strain gauges in the overlapping zone. It can be seen from the table that when the same BFRP jacket layer is used, the ratio of the measured lateral strain of ε h , S G 1 to ε h , a v e is close to the theoretical ratio (i.e., n o v e r l a p / n n o n − o v e r l a p ) , while the ratio of the hoop strain of ε h , S G 2 to ε h , a v e is usually higher than the theoretical ratio. The results further indicate that in the overlapping zone, the confinement efficiency of the BFRP jackets increases with the increase of the distance from the end edge of the FRP jackets. This is because when the concrete core expands, the tension force starts from the inside of the FRP, transmits in the hoop direction, and arrives at the end of the overlapping. After reaching the transition area between the overlapping and the non-overlapping zone, it will cause a sudden change in stiffness. Therefore, in the overlapping zone, the FRP jackets near the end edge exhibit a smaller tensile stress, while the FRP jackets far from the end edge exhibit a higher tensile stress.

In order to further illustrate the influence of historical high temperatures (i.e., different heat-damage levels of concrete) on the strain responses of the BFRP jackets, Figures 5A,B compare the strain distributions in the non-overlapping zones for the specimens confined by two layers and four layers of BFRP jackets, respectively, which are initially exposed to various high temperatures (i.e., 200°C, 400°C, 600°C, and 800°C). It can be seen that the hoop strain values of the non-overlapping zone are significantly higher than those of the overlapping zone, which are consistent with the observations reported in the literature (Lam and Teng, 2004; Wu and Jiang, 2013; Chen et al., 2013; Pham et al., 2015; Take and Bisby, 2009). Figure 6 shows the strain distributions of four typical specimens under different loading levels. The lateral strains measured in the non-overlapping zone are not uniform, mainly attributed to the inhomogeneity of the concrete that causes the uneven lateral deformations (Lam and Teng, 2004; Take and Bisby, 2009). In addition, when the number of the BFRP jackets is increased from two to four layers, the lateral strain distributions are more uniform. This indicates that the increase of the BFRP layers helps to improve the nonuniformity of the lateral strain distribution. However, the increase in the number of BFRP layers cannot enhance the hoop rupture strain, which means that the confinement efficiency of a single layer of BFRP will not increase with the growth in the number of BFRP jackets.

The relationship between the efficiency factor and the historical temperatures is presented in Figure 7. As the historical temperature increases, the efficiency factor grows, and the increasing rates for the specimens with different levels of heat damage after exposure to different temperatures are different. It is apparent that the efficiency factor increases significantly between the exposed temperature from 400°C to 600°C due to the abrupt deterioration of the compressive strength of concrete (Bisby et al., 2011; Lenwari et al., 2016; Jia et al., 2021).

Figure 8 illustrates the relationship between the values of the BFRP efficiency factor versus the confinement strengths provided by the BFRP jackets. The values of the BFRP efficiency factor decrease slightly with the increase of the confinement strength under various historical high-temperature conditions. For example, for a confinement strength of about 4.06 MPa, the values of the BFRP efficiency factor for the BFRP-confined heat-damaged specimens after exposure to 400°C, 600°C, and 800°C are 0.742, 0.798, and 0.810, respectively. When the confinement strength is increased to almost 8.12 MPa, the efficiency factor values of the corresponding specimens after exposure to 400°C, 600°C, and 800°C decrease by 1.46, 1.23, and 2.34%, respectively. Moreover, as the confinement strength increases, the difference between the values of the efficiency factor for the specimens with different levels of initial heat damage decreases. In other words, under monotonically axial compression, the effect of the confinement strength on the FRP efficiency factor is much smaller than those of compressive strength and ultimate axial strain (Jian and Ozbakkaloglu, 2015). Therefore, the effect of BFRP confinement strength on the efficiency factor is ignored in deducing the design equation of the efficiency factor of BFRP for confined heat-damaged concrete cylinders.

Ozabakkaloglu et al. (2013) collected a large amount of existing test data and proposed a design method of FRP efficiency factor based on the FRP-confined concrete columns tested at ambient temperature. An attempt to use the Ozabakkaloglu et al.’s design method to predict the BFRP efficiency factor for the confined heat-damaged concrete is conducted, and the predicted values of the efficiency factor are shown in Figure 9. The performance of the design method in predicting the efficiency factor of BFRP for the confined heat-damaged concrete columns is evaluated using statistical indicators including the mean squared error (MSE) and the average absolute error (AAE), as described in the following equations: M S E = 1 N ∑ i = 1 N ( Pr e i − E x p i E x p i ) 2 (4) A A E = ∑ i = 1 N | Pr e i − E x p i E x p i | N (5) where N is the total number of the data points, and P r e i and E x p i are the prediction and the test result of the i th specimen, respectively.

It can be observed that the method by Ozabakkaloglu et al. (2013) is not capable of predicting the experimental results accurately. When the historical high temperature increases from 200°C to 400°C, the predicted values of the efficiency factor agree reasonably well with the test results, which further verify the applicability of the design method for the FRP-confined concrete columns under ambient temperature conditions. However, when the historical high temperature further increases to 600°C or 800°C, almost all the test data points are lower than the line of equality (solid line in Figure 9), indicating that the method by Ozabakkaloglu et al. (2013) underestimates the efficiency of the BFRP jackets for confining the heat-damaged concrete. Therefore, a new and more accurate design equation is needed to describe the efficiency of BFRP in practical design of BFRP-confined heat-damaged concrete columns, especially for these columns after experiencing moderate or severe heat- or fire-induced damage.

As stated in the above sections, the efficiency factor of BFRP jackets used for the confinement of heat-damaged concrete columns relies on the initial concrete damage level. Therefore, a regression curve between the detailed test values of the BFRP efficiency factor versus the historical high temperatures is proposed as follows to describe the effect of historical high temperature: k ε T = 0.0137 ( T 100 ) + 0.6962 (6)

The accuracy of the proposed design method is described in Figure 10. It is seen that the BFRP efficiency factor increases almost linearly with the historical high temperature, and the value of R 2 between the regression curve and the test results is 0.8815, indicating that there is a good agreement between the test results and the model predictions. It should be noted that the design equation (i.e., Eq. 6) is only described as a function of historical high temperature (i.e., the initial heat damage of concrete), so it has only considered the effect of initial heat damage on the efficiency factor of BFRP jackets for confining heat-damaged concrete. The detailed value of k ε can be determined by multiplying k ε T (calculated by Eq. 6) by the corresponding value determined by the exiting design method of the efficiency factor proposed at ambient temperature ( k ε 0 ) in the literature [such as those provided in the Refs (Lam and Teng, 2004; Chen et al., 2013; Wu and Jiang, 2013; Jian and Ozbakkaloglu, 2015; Pham et al., 2015)]. That is, the efficiency factor k ε = k ε T × k ε 0 .