Post-Fire Characteristics of Concrete Beams Reinforced with Hybrid FRP Bars

One of the main concerns of experimental and numerical investigations regarding the behavior of fiber-reinforced polymer reinforced concrete (FRP-RC) members is their fire resistance to elevated temperatures and structural performance at and after fire exposure. However, the data currently available on the behavior of fiber-reinforced polymer (FRP) reinforced members related to elevated temperatures are scarce, specifically relating to the strength capacity of beams after being subjected to elevated temperatures. This paper investigates the residual strength capacity of beams strengthened internally with various (FRP) reinforcement types after being subjected to high temperatures, reflecting the conditions of a fire. The testing was made for concrete beams reinforced with three different types of FRP bars: (i) basalt-FRP (BFRP), (ii) hybrid FRP with carbon and basalt fibers (HFRP) and (iii) nano-hybrid FRP (nHFRP), with modification of the epoxy matrix of the rebar. Tested beams were first loaded at 50% of their ultimate strength capacity, then unloaded before being heated in a furnace and allowed to cool, and finally reloaded flexurally until failure. The results show an atypical behavior observed for HFRP bars and nHFRP bars reinforced beams, where after a certain temperature threshold the deflection began to decrease. The authors suggest that this phenomenon is connected with the thermal expansion coefficient of the carbon fibers present in HFRP and nHFRP bars and therefore creep can appear in those fibers, which causes an effect of “prestressing” of the beams.


Introduction
For the last few decades, the use of fiber-reinforced polymers (FRP) composites for concrete structures has experienced explosive growth. A major reason of its implementation lies in the unique characteristics of these materials and their constituents, as well as the appropriate use of FRP composites for its considered design purpose. Numerous investigations have shown that FRP composites are effective materials for use in concrete members [1,2].
The application of FRP bars as internal reinforcement in concrete members has many advantages, as opposed to structural members reinforced with steel bars [3]. Durability, strength and stability are the main criteria when selecting a material in the design of concrete structures with internal FRP reinforcement [4][5][6][7].
The progress in manufacturing technology has introduced new types of FRP bars, and furthermore hybrid FRP (HFRP), which allow one to adjust FRP characteristics to a specific design situation. HFRP bars contain several types of reinforcing constituents and/or several types of adhesive constituents in a matrix. Moreover, some authors suggest that such hybridization of the constituents of FRP bars can prevent changes in their behavior, making it semi-ductile instead of linear [8,9]. with GFRP bars. Two unloaded slabs, after being heated, attained deflections of approximately 70 mm, due to their own weight and thermal strains, and their residual resistance, evaluated after the cooling phase, which was about 55% to 100% of the ultimate design bearing moment resistance. Moreover, it was stated that the typical values of concrete clear cover (i.e., between 30 and 50 mm) can be adopted.
On the other hand, the relatively non-corrosive behavior of the fibers prevents a significant reduction of concrete clear cover without compromising the material's durability [28][29][30]. This enables thinner structures to be built, whose construction requires less raw materials and is more sustainable, environmentally friendly, durable, and cost-efficient. However, thin-walled elements with thin concrete covers may subject the reinforcement to substantial effects when exposed to elevated temperatures [31].
As known from experiments on beams reinforced with steel reinforcement, the concrete clear cover, and the fire scenario have the most significant influence on the response of the concrete element subjected to fire [32]. Nevertheless, in the design of FRP-RC structures, the reinforcement type and obtaining of a specific high-temperature resistance for FRP bars is one of the main concerns. The use of FRP in structures vulnerable to fire needs further research since this is one of the main reasons that limit the widespread use of FRP in construction.
However, the behavior of such reinforcement during a fire is still unknown. In addition, the mentioned literature demonstrates a lack of insight on the behavior of basalt FRP (BFRP) and HFRP reinforcement at and after applying fire conditions. Therefore, the authors attempted to investigate the post-fire behavior of RC flexural members reinforced with these types of reinforcement. The FRP-RC members were not subjected to loading during heating phase to allow to concentrate on the performance of FRP bars after being subjected to elevated temperatures. Different reinforcement configurations were examined for the tension zone to check the influence of different types of the bars. This paper investigates the residual capacity of full-scale FRP-RC beams after being subjected to elevated temperatures. For residual testing, the beams were preliminary loaded with a force of 50% of their ultimate strength capacity (checked at room temperature), then the beams were unloaded. Afterwards the beams were placed into a furnace and heated for approximately one hour. After heating, the beams were allowed to cool, and finally reloaded flexurally until failure.
The results of the investigations were discussed, with particular emphasis on the structural performance of the concrete members and on the types of FRP bars used.

Novelty and Purpose of the Work
Tested specimens were subjected to specific fire actions, where the midsection, approximately 1/3 of the length of the beams, was exposed to heating from both the bottom and the sides. In conventional testing, the entire span length of concrete members is usually subjected to heating and loading is applied simultaneously. The presented investigation is focused on residual fire resistance testing, where the samples were heated and the strength capacity was checked afterwards.
As one of the main aims was to examine the influence of different reinforcement types, the testing was made for concrete beams reinforced with three different types of FRP bars, including newly developed hybrid FRP bars: (i) basalt-FRP (BFRP), (ii) hybrid FRP with carbon and basalt fibers (HFRP) and (iii) nano-hybrid FRP (nHFRP), with modification of the epoxy matrix of the rebar.

The Concept of Hybrid FRP Bars
The concept behind the invention of HFRP bars came as a desire to obtain a material with improved and adjustable properties, where the final product can exploit the most favorable qualities of every individual constituents.
In the context of this work, the term "hybridization" should be understood as a physical substitution of some of the fibers by another type. The proposed constituents for these hybrid bars: carbon and basalt fibers, were embedded in epoxy resin creating HC/BFRP (HFRP) bars.
Processing HFRP bars is analogous to BFRP bars, a certain amount of the basalt roving is physically replaced with carbon roving during the pultrusion process; subsequently both were embedded in epoxy resin. Different constituents and their location were estimated. Basalt and carbon fibers of low strength (LS) type were selected due to their similar strain parameters. Different volume fractions for carbon and basalt (C:B) fibers were considered as well, before production of the bars.
Analytical and numerical estimations were performed, prior to the production of HFRP and nHFRP bars. The properties of composite bars in the longitudinal direction can be calculated using the rule of mixtures (ROM) (Voigt model), as it comes from the literature [33][34][35]. The transverse properties can be obtained with Halpin-Tsai and other semi-empirical models [36,37]. However, these formulas do not consider bars configuration, i.e., the location of fibers, so, it was agreed to prepare an additional numerical simulation for this purpose.
The numerical simulation of tensile strength test for HFRP bars with different configurations of fibers was performed by finite element methods (FEM) in the software ANSYS ® Academic Research Mechanical, Release 16.2 (Ansys Inc., Canonsburg, PA, USA) [38]. Two different bar configurations were proposed, one where carbon fibers substituted the basalt fibers in the core region, the other one with carbon fibers located in the near-surface region.
The bars were modelled as cylindrical elements with a diameter of 8 mm and a length of 850 mm. A constant pressure of 500 MPa was applied on the side edges. One central point was fixed in the y and z directions (non-longitudinal directions). The structure of the HFRP bars consisted of a core and surface region, which were perfectly interconnected.
The obtained results from numerical modelling were compared with analytical considerations and are found to be in a good convergence with each other. The results for different fibers arrangement and different volume fractions of fibers are described in Table 1. The comparison between analytical considerations and numerical simulations of FRP bars was explained in more detail in [8]. ed in epoxy resin. Different constituents and their location were estimated. Basalt and carbon low strength (LS) type were selected due to their similar strain parameters. Different volume s for carbon and basalt (C:B) fibers were considered as well, before production of the bars. alytical and numerical estimations were performed, prior to the production of HFRP and bars. The properties of composite bars in the longitudinal direction can be calculated using of mixtures (ROM) (Voigt model), as it comes from the literature [33][34][35]. The transverse ies can be obtained with Halpin-Tsai and other semi-empirical models [36,37]. However, rmulas do not consider bars configuration, i.e. the location of fibers, so, it was agreed to an additional numerical simulation for this purpose. e numerical simulation of tensile strength test for HFRP bars with different configurations of as performed by finite element methods (FEM) in the software ANSYS® Academic Research ical, Release 16.2 (Ansys Inc., Canonsburg, PA, USA) [38]. Two different bar configurations oposed, one where carbon fibers substituted the basalt fibers in the core region, the other one bon fibers located in the near-surface region. e bars were modelled as cylindrical elements with a diameter of 8 mm and a length of 850 constant pressure of 500 MPa was applied on the side edges. One central point was fixed in d z directions (non-longitudinal directions). The structure of the HFRP bars consisted of a surface region, which were perfectly interconnected. e obtained results from numerical modelling were compared with analytical considerations found to be in a good convergence with each other. The results for different fibers ment and different volume fractions of fibers are described in Table 1. The comparison analytical considerations and numerical simulations of FRP bars was explained in more [8].  e results showed that bar configuration is less important than the volume fraction of fibers. erence between various bar configurations can be a maximum of 2%, meanwhile, the volume of all analyzed configurations can influence the final stiffness by 74.6%. eral technological issues were observed while placing the carbon fibers located in the nearregion, including increased heterogeneity at fiber locations and local scorching of bars caused erature increases. Consequently, the most suitable location for carbon fibers is in the core f the HFRP rebar. addition, nHFRP bars were prepared with an addition of nano-silica fillers to the resin ncy. The additives were used for improving the chemical cohesion between the agents of and to improve the overall properties of the bars, as suggested in some sources [39][40][41]. ure 1 represents an example of scanning electron microscope (SEM) images, showing the ents of nHFRP bars, where it is possible to distinguish different fibers and nano-additives (by mensions and color). Visible areas of nanosilica locations are marked as Q, bright spots basalt fibers, while carbon fibers appear as darker areas.
Materials 2020, 13, x FOR PEER REVIEW 4 of 15 embedded in epoxy resin. Different constituents and their location were estimated. Basalt and carbon fibers of low strength (LS) type were selected due to their similar strain parameters. Different volume fractions for carbon and basalt (C:B) fibers were considered as well, before production of the bars. Analytical and numerical estimations were performed, prior to the production of HFRP and nHFRP bars. The properties of composite bars in the longitudinal direction can be calculated using the rule of mixtures (ROM) (Voigt model), as it comes from the literature [33][34][35]. The transverse properties can be obtained with Halpin-Tsai and other semi-empirical models [36,37]. However, these formulas do not consider bars configuration, i.e. the location of fibers, so, it was agreed to prepare an additional numerical simulation for this purpose.
The numerical simulation of tensile strength test for HFRP bars with different configurations of fibers was performed by finite element methods (FEM) in the software ANSYS® Academic Research Mechanical, Release 16.2 (Ansys Inc., Canonsburg, PA, USA) [38]. Two different bar configurations were proposed, one where carbon fibers substituted the basalt fibers in the core region, the other one with carbon fibers located in the near-surface region.
The bars were modelled as cylindrical elements with a diameter of 8 mm and a length of 850 mm. A constant pressure of 500 MPa was applied on the side edges. One central point was fixed in the y and z directions (non-longitudinal directions). The structure of the HFRP bars consisted of a core and surface region, which were perfectly interconnected.
The obtained results from numerical modelling were compared with analytical considerations and are found to be in a good convergence with each other. The results for different fibers arrangement and different volume fractions of fibers are described in Table 1. The comparison between analytical considerations and numerical simulations of FRP bars was explained in more detail in [8]. The results showed that bar configuration is less important than the volume fraction of fibers. The difference between various bar configurations can be a maximum of 2%, meanwhile, the volume fraction of all analyzed configurations can influence the final stiffness by 74.6%.
Several technological issues were observed while placing the carbon fibers located in the nearsurface region, including increased heterogeneity at fiber locations and local scorching of bars caused by temperature increases. Consequently, the most suitable location for carbon fibers is in the core region of the HFRP rebar.
In addition, nHFRP bars were prepared with an addition of nano-silica fillers to the resin consistency. The additives were used for improving the chemical cohesion between the agents of nHFRP and to improve the overall properties of the bars, as suggested in some sources [39][40][41]. Figure 1 represents an example of scanning electron microscope (SEM) images, showing the constituents of nHFRP bars, where it is possible to distinguish different fibers and nano-additives (by their dimensions and color). Visible areas of nanosilica locations are marked as Q, bright spots indicate basalt fibers, while carbon fibers appear as darker areas. The results showed that bar configuration is less important than the volume fraction of fibers. The difference between various bar configurations can be a maximum of 2%, meanwhile, the volume fraction of all analyzed configurations can influence the final stiffness by 74.6%.
Several technological issues were observed while placing the carbon fibers located in the near-surface region, including increased heterogeneity at fiber locations and local scorching of bars caused by temperature increases. Consequently, the most suitable location for carbon fibers is in the core region of the HFRP rebar.
In addition, nHFRP bars were prepared with an addition of nano-silica fillers to the resin consistency. The additives were used for improving the chemical cohesion between the agents of nHFRP and to improve the overall properties of the bars, as suggested in some sources [39][40][41].

Experimental Program
The experimental program is involved in the design and fabrication of the six FRP reinforced concrete beams without any fire protection system. The investigation consisted of fire tests performed on three simply supported concrete beams strengthened internally with different reinforcement in the tensile zone (Set 1) and three corresponding reference beams that were used for comparison (Set 2).
The beams of Set 1 were preliminarily loaded by a force of 50% of their ultimate strength capacity to simulate realistic situations, i.e., the appearance of cracks. As for the next step, unloaded cracked beams were placed into a furnace in such a way that the mid-section of the beams was inside it. After the cooling phase, the beams' capacity was checked with a four-point flexural test. The beams of Set 2 were tested with four-point flexural tests without preliminary loading and were not exposed to elevated temperature. Table 2 shows the description of specimens used in the tests.

Experimental Program
The experimental program is involved in the design and fabrication of the six FRP reinforced concrete beams without any fire protection system. The investigation consisted of fire tests performed on three simply supported concrete beams strengthened internally with different reinforcement in the tensile zone (Set 1) and three corresponding reference beams that were used for comparison (Set 2).
The beams of Set 1 were preliminarily loaded by a force of 50% of their ultimate strength capacity to simulate realistic situations, i.e., the appearance of cracks. As for the next step, unloaded cracked beams were placed into a furnace in such a way that the mid-section of the beams was inside it. After the cooling phase, the beams' capacity was checked with a four-point flexural test. The beams of Set 2 were tested with four-point flexural tests without preliminary loading and were not exposed to elevated temperature. Table 2 shows the description of specimens used in the tests. The concrete mixture design was identical for all beams: a typical concrete mix C40/45 was used. Ordinary Portland cement CEM III/A, ash, and crushed stone (silica), with a nominal maximum size of 16 mm, were used in concrete mixes. The beams were cured for a complete 28-day period in the lab, before they were moved to the testing frame. The cubic specimens from the same concrete mixture were checked to confirm the concrete class.

Bars
A more detailed description for choosing the configuration of the bars and their characteristics is reported in the following companion papers [42][43][44]. For these tests, the bar types BFRP, HFRP, and nHFRP were chosen.
The selected volume fractions of carbon-to-basalt fibers for HFRP bars were assumed as 1:4 (i.e., 16% of carbon fibers, 64% of basalt fibers and 20% of epoxy resin). For the nHFRP bars, the matrix was modified by adding a four-component 1300 System ® (CIECH Sarzyna S.A., Nowa Sarzyna, Poland) to the epoxy resin. A sol with the nanosilica particles with a concentration of 25 to 30% by weight was used. Average diameter of particles was equal 24.7 nm, containing two fractions: fine (80%) and coarse (20%).
The mean values of maximum strength, F u , limit stress, f u , modulus of elasticity, E 11 , and the limit strain, ε u , for BFRP, HFRP and nHFRP bars were obtained from tensile tests and are shown in Table 3.

Specimen Dimensions and Reinforcement Configuration
All specimens had a rectangular cross-section with dimensions 140 mm wide and 260 mm high and were 3200 mm long. The clear cover, c nom , was equal to 30 mm from all sides for all tested specimens.
The top reinforcement (compression zone) and shear reinforcement (stirrups) were kept constant for all beams. Two BFRP bars with a diameter of 8 mm were used as longitudinal top reinforcement. The stirrups made of BFRP with a diameter of 6 mm were used as shear reinforcement. The stirrup spacing was assumed as 100 mm and the mid-part of the beam did not have stirrups to simulate clear bending. The bottom reinforcement (tension zone) was a variable parameter and described based on the reinforcement type, as mentioned in Figure 2. Figure 2 shows the scheme of tested specimens.

Test Setup
Prior to casting, each specimen was instrumented with thermocouples of Type K (till 1200 °C) inside the specimens at different locations, in order to monitor the temperature during fire exposure. Eight thermocouples were used for each beam that were embedded at different depths. Seven of the thermocouples were added into the concrete and one of the thermocouples, T3, was embedded on the upper surface of a bar.
Three dial gauges were applied to the top face in order to measure the deflections in the midsection of the beams. In addition, two dial gauges were added on the sides to measure possible beam extensions.
The specimens were exposed to heat in the mid-section from below and from the sides. Outputs were recorded every half second during the heating and cooling process.
Before the temperature was applied, the beams from Set 1 were loaded to 50% of their ultimate strength load (as described in Table 1). Set 2 was tested without temperature exposure and preliminary loading. Figure 3 shows the test setup of the beams.

Test Setup
Prior to casting, each specimen was instrumented with thermocouples of Type K (till 1200 • C) inside the specimens at different locations, in order to monitor the temperature during fire exposure. Eight thermocouples were used for each beam that were embedded at different depths. Seven of the thermocouples were added into the concrete and one of the thermocouples, T3, was embedded on the upper surface of a bar.
Three dial gauges were applied to the top face in order to measure the deflections in the mid-section of the beams. In addition, two dial gauges were added on the sides to measure possible beam extensions.
The specimens were exposed to heat in the mid-section from below and from the sides. Outputs were recorded every half second during the heating and cooling process.
Before the temperature was applied, the beams from Set 1 were loaded to 50% of their ultimate strength load (as described in Table 1). Set 2 was tested without temperature exposure and preliminary loading. Figure 3 shows the test setup of the beams. Beams were heated from the bottom, as well as parts of both sides, for a period of approximately one hour. The standards provide the means of heating beams from room temperature to temperatures representative of a fire, taking into account a time parameter. Structural fire testing was performed under standard conditions in accordance with norms [45] and a standard heating curve ISO-834 (1999) [46], represented by formula: Beams were heated from the bottom, as well as parts of both sides, for a period of approximately one hour. The standards provide the means of heating beams from room temperature to temperatures representative of a fire, taking into account a time parameter. Structural fire testing was performed under standard conditions in accordance with norms [45] and a standard heating curve ISO-834 (1999) [46], represented by formula: where: T ISO is the temperature ( • C), T 0 is the room temperature (assumed to be 20 • C) and t is the time (min). Figure 4 illustrates the test setup during the heating phase. Beams were heated from the bottom, as well as parts of both sides, for a period of approximately one hour. The standards provide the means of heating beams from room temperature to temperatures representative of a fire, taking into account a time parameter. Structural fire testing was performed under standard conditions in accordance with norms [45] and a standard heating curve ISO-834 (1999) [46], represented by formula: where: TISO is the temperature (°C), T0 is the room temperature (assumed to be 20 °C) and t is the time (min). Figure 4 illustrates the test setup during the heating phase.

Results and Discussion
The performed tests provide a means to estimate the influence of elevated temperatures on beams reinforced with FRP and therefore investigate their thermal behavior being subjected to those temperatures. All specimens showed a typical flexural-ductile failure, the beams of Set 1 were destroyed due to reinforcement failure in the tensile zone, however the reason of the beams destruction of Set 2 was the concrete crushing in the compression zone. Figure 5 demonstrates the beam sample B2Ø14 from Set 1 just after it was taken out of the furnace.
As shown on Figure 6, after the hybrid FRP bars were uncovered by removing the clear cover, the temperature caused a burning of the FRP bars. This resulted in the evaporation of the matrix in the middle of the bars. The temperature caused a de-bonding with the concrete, completely separating the reinforcement from the concrete (Figure 6a). However, a major portion of the fibers remained in the same place and continued to sustain the load (Figure 6b). The performed tests provide a means to estimate the influence of elevated temperatures on beams reinforced with FRP and therefore investigate their thermal behavior being subjected to those temperatures. All specimens showed a typical flexural-ductile failure, the beams of Set 1 were destroyed due to reinforcement failure in the tensile zone, however the reason of the beams destruction of Set 2 was the concrete crushing in the compression zone. Figure 5 demonstrates the beam sample B2Ø14 from Set 1 just after it was taken out of the furnace. As shown on Figure 6, after the hybrid FRP bars were uncovered by removing the clear cover, the temperature caused a burning of the FRP bars. This resulted in the evaporation of the matrix in the middle of the bars. The temperature caused a de-bonding with the concrete, completely separating the reinforcement from the concrete (Figure 6a). However, a major portion of the fibers remained in the same place and continued to sustain the load (Figure 6b).  destruction of Set 2 was the concrete crushing in the compression zone. Figure 5 demonstrates the beam sample B2Ø14 from Set 1 just after it was taken out of the furnace. As shown on Figure 6, after the hybrid FRP bars were uncovered by removing the clear cover, the temperature caused a burning of the FRP bars. This resulted in the evaporation of the matrix in the middle of the bars. The temperature caused a de-bonding with the concrete, completely separating the reinforcement from the concrete (Figure 6a). However, a major portion of the fibers remained in the same place and continued to sustain the load (Figure 6b).  Analyses of the temperatures attained at different depths were possible with the collection of measurements from eight thermocouples embedded in the concrete and on the bars. An example temperature distribution is shown in Figure 7a,b, which demonstrate the measurements obtained from thermocouples for sample B2Ø14. The thermocouples are numbered in accordance with the scheme (Figure 3) and represent temperatures at different locations.
From Figure 7, it can be seen that the temperatures obtained at thermocouples T4 and T7 are similar, indicating that the temperatures were distributed uniformly. At the same time, measurements obtained from thermocouples T1 and T5, indicate that the gaps between the beam interfaces and edges of the furnace were carefully insulated with ceramic and rock wool. In addition, it can also be seen that the maximum temperatures measured at the bottom edge were approximately 20% lower (at the last point) than the applied temperatures. scheme ( Figure 3) and represent temperatures at different locations.
From Figure 7, it can be seen that the temperatures obtained at thermocouples T4 and T7 are similar, indicating that the temperatures were distributed uniformly. At the same time, measurements obtained from thermocouples T1 and T5, indicate that the gaps between the beam interfaces and edges of the furnace were carefully insulated with ceramic and rock wool. In addition, it can also be seen that the maximum temperatures measured at the bottom edge were approximately 20% lower (at the last point) than the applied temperatures. The sample beam B2Ø14 from Set 1 showed a time-vertical deflection relationship more similar to a typical one for steel reinforced concrete beams [13,32]; the deflections were only increasing. Despite the larger initial deflections for beam samples B2Ø14, H2Ø14 and N2Ø14, the increase of deflections during fire exposure started to be twice lower than in the case of analogous beams with steel reinforcement. However, samples H2Ø14 and N2Ø14 demonstrated an atypical behavior for beams subjected to elevated temperatures, since at a certain point the deflection began to decrease. At that time, the registered temperature on the bar surface was in the range of 550-570 °C. Figure 8 describes the deflections obtained by three dial gauges during the heating process of the beams from Set 1. The sample beam B2Ø14 from Set 1 showed a time-vertical deflection relationship more similar to a typical one for steel reinforced concrete beams [13,32]; the deflections were only increasing. Despite the larger initial deflections for beam samples B2Ø14, H2Ø14 and N2Ø14, the increase of deflections during fire exposure started to be twice lower than in the case of analogous beams with steel reinforcement. However, samples H2Ø14 and N2Ø14 demonstrated an atypical behavior for beams subjected to elevated temperatures, since at a certain point the deflection began to decrease. At that time, the registered temperature on the bar surface was in the range of 550-570 • C. Figure 8 describes the deflections obtained by three dial gauges during the heating process of the beams from Set 1.  Figure 9 shows the deflections of beams from Set 1, measured by the middle dial gauge, U2, during the heating and cooling phases. It was noticed that just after the heating phase, the deflection of sample B2Ø14 at first started to decrease and then returned to its maximum value (shown in the Figure 9a), however for both samples H2Ø14 and N2Ø14 the deflections during the cooling phase started to decrease until approx. 30% of their maximum value was reached (Figure 9b,c). In the opinion of the authors, this difference is due to the carbon fiber presence in hybrid FRP bars, which works as a kind of "prestressing" during heating (at certain thresholds as shown in Figure 8), as well  Figure 9 shows the deflections of beams from Set 1, measured by the middle dial gauge, U2, during the heating and cooling phases. It was noticed that just after the heating phase, the deflection of sample B2Ø14 at first started to decrease and then returned to its maximum value (shown in the Figure 9a), however for both samples H2Ø14 and N2Ø14 the deflections during the cooling phase started to decrease until approx. 30% of their maximum value was reached (Figure 9b,c). In the opinion of the authors, this difference is due to the carbon fiber presence in hybrid FRP bars, which works as a kind of "prestressing" during heating (at certain thresholds as shown in Figure 8), as well as during the cooling process, since beam displacement becomes smaller and then remains constant.
(c)  Figure 9 shows the deflections of beams from Set 1, measured by the middle dial gauge, U2, during the heating and cooling phases. It was noticed that just after the heating phase, the deflection of sample B2Ø14 at first started to decrease and then returned to its maximum value (shown in the Figure 9a), however for both samples H2Ø14 and N2Ø14 the deflections during the cooling phase started to decrease until approx. 30% of their maximum value was reached (Figure 9b,c). In the opinion of the authors, this difference is due to the carbon fiber presence in hybrid FRP bars, which works as a kind of "prestressing" during heating (at certain thresholds as shown in Figure 8), as well as during the cooling process, since beam displacement becomes smaller and then remains constant. The strength capacity for beams of Set 1 was compared with beams of Set 2, which were tested without the influence of elevated temperatures ( Figure 10). The beams from Set 2 were only subjected to loading in the four-point flexural test, and subsequently had a greater strength capacity. The samples of Set 2 were loaded till 12.5 kN and then loading was reduced to 5 kN, after one more The strength capacity for beams of Set 1 was compared with beams of Set 2, which were tested without the influence of elevated temperatures ( Figure 10). The beams from Set 2 were only subjected to loading in the four-point flexural test, and subsequently had a greater strength capacity. The samples of Set 2 were loaded till 12.5 kN and then loading was reduced to 5 kN, after one more analogous cycle of loading the beams until failure. The cyclic loading was applied in order to reduce effects of plastic strains.
A comparison between results of tested beams is represented in Table 4. Set 2 (Fu, reference) showed a higher strength capacity by 43%, 40% and 43% respectively for beams B2Ø14; H2Ø14 and N2Ø14 from Set 1 (Fu, tested). A comparison between results for beams from Set 1 and Set 2 is shown in Figure 10. The highest strength capacity that was obtained by the tested sample H2Ø14 for both beams was equal to 51 and 85 kN for beams from Set 1 and Set 2, respectively.

Conclusions
This study investigates the residual strength capacity for beams subjected to elevated temperatures that were preliminarily loaded by half of their ultimate capacity. Solely FRP reinforcement was used; beams were having different reinforcement in the tensile zone (bottom part of the beams): BFRP, HFRP or nHFRP bars. Reinforcement used in the compression zone and stirrups were made from BFRP.
Obtained results show that deflections obtained from current investigation for beams reinforced with FRP bars were much smaller than the results received for similar beams reinforced with steel bars in other studies [13,32]. In addition, the behavior of FRP-RC beams was quite uncharacteristic when subjected to elevated temperature, as well as during the cooling phase. Set 2 (F u, reference ) showed a higher strength capacity by 43%, 40% and 43% respectively for beams B2Ø14; H2Ø14 and N2Ø14 from Set 1 (F u, tested ). A comparison between results for beams from Set 1 and Set 2 is shown in Figure 10. The highest strength capacity that was obtained by the tested sample H2Ø14 for both beams was equal to 51 and 85 kN for beams from Set 1 and Set 2, respectively.

Conclusions
This study investigates the residual strength capacity for beams subjected to elevated temperatures that were preliminarily loaded by half of their ultimate capacity. Solely FRP reinforcement was used; beams were having different reinforcement in the tensile zone (bottom part of the beams): BFRP, HFRP or nHFRP bars. Reinforcement used in the compression zone and stirrups were made from BFRP.
Obtained results show that deflections obtained from current investigation for beams reinforced with FRP bars were much smaller than the results received for similar beams reinforced with steel bars in other studies [13,32]. In addition, the behavior of FRP-RC beams was quite uncharacteristic when subjected to elevated temperature, as well as during the cooling phase.
Based on the experimental testing of beams reinforced with FRP, the following main conclusive remarks can be drawn:

1.
All tested samples subjected to elevated temperatures were destroyed due to the tensile zone reaching its ultimate strength capacity. The authors suggest that this issue can be related to the reduction of mechanical properties experienced by the bars after being subjected to fire actions.

2.
The overall strength capacity of the FRP reinforced beams after being subjected to fire exposure was reduced; by approx. 43% for beams with the tensile zone reinforced with BFRP bars, 40% and 43% for beams reinforced with HFRP and nHFRP bars, correspondingly.