An Analysis of the Transfer Lengths of Different Types of Prestressed Fiber-Reinforced Polymer Reinforcement

The main aim of this paper is to provide a broader analysis of the transfer lengths of different types of fiber-reinforced polymers (FRPs) and to provide corrections to the existing theoretical models. Therefore, this paper presents a description of the main factors that influence the transfer lengths of different types of FRPs based on experimental results found in the literature. A database of more than 300 specimens was compiled with the results of the transfer lengths of different FRPs and the main influencing parameters. The analysis of the database results showed that the transfer length of the carbon fiber composite cable (CFCC) strands depends on the type of prestressed reinforcement release. Therefore, in this article, the new coefficient αt = 2.4 is proposed for the transfer length of suddenly released CFCC strands. Additionally, the transfer length of the aramid fiber reinforced polymer (AFRP) depends on its surface conditions. Therefore, new coefficients αt = 1.5 and αt = 4.0 are also proposed for the transfer lengths of smooth braided and sanded and rough AFRP bars, respectively. Furthermore, the proposed coefficients αt = 2.6, αt = 1.9, and αt = 4.8 found in the literature were validated with the analysis of a larger database of the transfer lengths of glass fiber-reinforced polymer (GFRP) bars, carbon fiber-reinforced polymer (CFRP) bars, and gradually released CFCC strands, respectively. Moreover, the main existing theoretical models are presented, and the comparison of theoretical and experimental transfer length results is discussed. However, the low number of specimens prestressed with basalt fiber-reinforced polymer (BFRP) bars prevented the deeper analysis of the results. the analysis of the transfer length and the proposed new values of the coefficient αt provides possibilities for adapting it to design codes for engineering applications and performing additional research that fills the missing gaps in the field.


Introduction
There has been increasing concern about the durability of prestressed concrete structures due to the corrosion of prestressed steel strands, especially in corrosive environments such as parking garages, bridges, marine structures, and railway sleepers [1][2][3]. Steel corrosion is accelerated in these structures as a result of the influence of a chloride-rich environment (deicing salts, etc.). Therefore, there is increasing interest in the use of FRP materials, namely, CFRP, AFRP, GFRP, and the relatively new BFRP [4,5] as replacements for steel reinforcements. BFRP was developed for applications in civil infrastructure as a structural reinforcement material. It has similar tensile strength, modulus of elasticity, and cost to GFRP. BFRP is more chemically stable, has a wider range of working temperatures, and is much cheaper than CFRP. Additionally, it is durable, resistant to high temperatures, corrosion, acid, radiation, ultraviolet (UV), and vibration; it also has excellent electromagnetic characteristics. Furthermore, BFRP is resistant to alkalis and is distinct from GFRP. AR-glass fibers are suggested for application in alkaline environments. However, it has a much higher cost. AFRP has good resistance to impact, high strength, high modulus of elasticity, the lowest density, and sufficient stiffness. AFRP can be utilized for impact-resistant Table 1. Basic physical and mechanical properties of different FRP bars [10][11][12][13][14]. FRPs have additional important properties that make them particularly attractive for prestressed concrete applications: high strength, which is similar to or greater than that of steel and low modulus of elasticity, which results in lower concrete prestress losses due to concrete creep and shrinkage as well as the relaxation of the prestressing element.
The major difficulty in using FRP reinforcement for prestressing is that anchorage systems require greater attention than those for steel strands. Therefore, three types of anchorage systems are developed for FRP reinforcement: mechanical, bonded, and composite [9].
The mechanical anchorage of reinforcement consisting of a steel barrel and wedges is ensured by friction between the reinforcement surface and the inner conical anchorage surface. Therefore, compressive stresses appear perpendicular to the reinforcement. However, FRP is an anisotropic material that has weak material properties perpendicular to the fibers. Therefore, practitioners suggest enhancing the mechanical anchorage of the FRP reinforcement using a small slope angle between the inner surface of the barrel and the outer surface of the wedges, the rounded edges of the wedges, and a soft metal sleeve (aluminum or copper) around the FRP. Other mechanical anchorages were described in [9,15] such as the curved wedge anchorage system [16], the nonmetallic wedge anchorage system [17], the integrated sleeve-wedge anchorage system [18], the spike anchorage system, and the clamping anchorage system. The mechanical anchors are usually of small size. However, this type of anchorage lacks stability and reliability.
The bonded-type anchorage consists of a steel barrel with a bonding material (cementitious or resin) surrounding the FRP bar. The inner part of the barrel can have three different shapes: straight; conical with decreased length but larger diameter; or straight and Polymers 2022, 14, 3931 3 of 37 conical with increased length and similar diameter with the purpose of relieving the stress concentration. The bonded-type anchorage depends on the bonding force between the CFRP cable and the bonding material to balance the tension of the cable. The performance of the bonded-type anchorage is mainly influenced by the dimensions and configuration of the steel barrel, the material type and surface condition of the FRP, and the properties of the bonding material. The main drawbacks are a large size, long curing time, creep of the bonding material, sensitivity to moisture, and high temperature.
The composite anchorage takes advantage of mechanical and bonded-type anchorage. In [19], the straight steel pipe filled with resin was directly clamped with a split-wedge anchor that improved the anchorage of the FRP. Additionally, [20] proposed a compoundtype anchorage system that adapted load transfer through wedge extrusion and epoxy bonding. The composite anchorage is smaller than the bonded anchorage but larger than the mechanical anchorage and presents good anchoring performance. At the same time, it has complicated construction and installation processes. Mechanical and bond anchorage systems are simpler than the composite anchorage system. Therefore, the mechanical and bond anchorage systems are still under investigation.
The use of FRP reinforcement in prestressed concrete structures is highly dependent on the reliability of the reinforcement anchorage zone. That is, the behavior of concrete members is mainly dependent on the bond between reinforcement and concrete [21][22][23]. In pretensioned concrete structures, the prestressing force is transferred to the concrete by the bond between reinforcement and concrete, which depends on three mechanisms: adhesion, Hoyer effect (wedge action), and mechanical interlocking. The bond of prestressed FRP is mainly influenced by the large Poisson ratio related to the Hoyer effect and the high axial strain capacity of the FRP materials [24]. Furthermore, if the effect of Poisson's ratio is dominant, the surface conditions of the FRP will have little influence on the bond [25]. In the case of FRP materials, the chemical bond (adhesion) is particularly weak [24]. Lastly, mechanical interlock and friction are quite dependent on the surface characteristics of the reinforcement.
In pretensioned members, the length from the end of the member where the tendon stress is zero to the point along the tendon where the prestress is fully effective is called the transfer length [12]. Ref. [11] defined the transfer length as the length required to transfer the prestressing force to the concrete after the release of a prestressing reinforcement. In other words, it is the distance along the member in which the effective prestressing force is developed.
One of the most important factors that affect the transfer length of prestressed reinforcement is the Hoyer effect, which is caused by the swelling of the strand in the transfer zone after release as a result of the Poisson ratio. During the transfer, the induced confining stresses normal to the tendon enhance the bond strength at the interface, as the lateral deformation is resisted by the surrounding concrete. Enhancement of the bond caused by the Hoyer effect is directly related to the strength of the concrete and the coefficient of friction between the two materials. Additionally, according to [26], the distribution of the bond stress over the transfer length is not linear due to the changing values of transverse strain (Hoyer effect) and slip over the transmission length.
Knowledge of the transfer length is essential to maintaining the integrity of the structure and to preventing the bond slip failure of the member. Additionally, an accurate estimate of the transfer length is important for checking the stress limit at the serviceability limit state, determining the actual bond stress in the anchorage zone of the pretensioned reinforcement, and designing the shear of the prestressed members. The underestimation of the transfer length leads to unconservative shear calculations, while overestimation leads to unconservative stress calculations at the serviceability limit state.
The simulation of the anchorage zone of FRP reinforcement introducing the finite element method (FEM) allows for performing a deeper analysis of the transfer length results that are difficult to measure in real experimental research. The simulation of the transfer lengths of BFRP [27], CFCC, and AFRP [28] showed that the optimization of the transfer length prediction of the FRP can be performed, eliminating unnecessary and expensive future experimental research.

Parameters That Influence the Transfer Length
The transfer length is an important parameter for the anchorage zone of prestressed reinforcement. In addition, it depends on various mechanical and geometric properties of the material. Therefore, this section presents an overview of the influence of different parameters on the transfer length of different prestressed FRP reinforcements (GFRP, CFRP, AFRP, CFCC and BFRP). Unfortunately, only a few studies on the transfer length of BFRP bars were found in the literature [27,55,56]. The lack of results and similar experimental parameters do not allow for qualitatively determining the influence of different parameters on the transfer length of BFRP bars.

Effect of Concrete Compressive Strength
Many authors investigated the influence of concrete compressive strength at the time of reinforcement release on the transfer length of concrete specimens pretensioned with different types of FRP reinforcement. Table 2 provides a summary of the average transfer length results, showing the influence of concrete compressive strength. The lower (f ci.min ) and higher (f ci.max ) concrete strength correspond to higher (L t.max ) and lower (L t.min ) transfer length, respectively. Zou [33] concluded that compared with the effect of the level of prestressing, the concrete strength at transfer is a more significant factor that affects the transfer length of the spirally indented CFRP Leadline bar. Additionally, [33] stated that all tested beams prestressed with AFRP bars had similar transfer lengths but different concrete strains with respect to concrete strength. Therefore, it was concluded that the transfer length of the AFRP bars is not significantly affected by the concrete strength.
The authors of [36,37] concluded that the transfer length of the prestressed GFRP bars decreased by about 32% when the concrete strength at the time of reinforcement release increased by 2.3 times. Additionally, Crossett et al. [55] showed that the transfer length of the BFRP bar decreased by 10% with an increase in concrete strength by 43%.
The FEM model of Motwani and Laskar [28] showed that peak elastic strain of concrete at the level of prestressed CFCC and AFRP Arapree reinforcement decreases with an increase in concrete strength.
Based on [33,38,39], it was stated in [36,37] that the concrete strength had a minimum effect on the transfer length of the AFRP reinforcement due to its small diameter (normally ≤9.5 mm) and shorter transfer length. Additionally, the authors of [36,37] stated that the larger diameter of the GFRP bars (Ø12-Ø16 mm), the longer the transfer length, allowing the effect of concrete strength on the transfer length to be apparent. However, in [38,39], a higher diameter of AFRP bars was also used (Ø12-Ø16 mm), and it resulted in transfer lengths almost twice as long as in [36,37] for GFRP bars with similar prestress levels and concrete strengths. Although the modulus of elasticity of AFRP and GFRP bars is similar, the surface of AFRP bars is braided (almost plain), and it is ribbed for GFRP bars. Therefore, the different surface conditions of AFRP and GFRP reinforcement may be the deciding factor for the difference in transfer length.
According to Table 2, it can be stated that with the increase i concrete compressive strength, the FRP transfer length decreases. However, the rate of decrease of the transfer length is lower for AFRP and GFRP reinforcement compared with that for CFCC and CFRP reinforcement. Therefore, the concrete strength is more significant for CFCC and CFRP reinforcement.

Effect of Prestress Level
The FRP reinforcements for prestressed concrete members can have different tensile strengths and initial stresses. Therefore, it is better to compare the ratios between initial stresses and tensile strength (f pi /f pu ) of different types of FRPs with the transfer length instead of the initial stresses (f pi ). Table 3 provides a summary of average transfer length results showing the influence of prestress level on the FRP reinforcement. The lower (f pi /f pu (min)) and higher (f pi /f pu (max)) prestress levels correspond to the lower (L t.min ) and higher (L t.max ) transfer lengths, respectively.
For the Ø12.5 mm CFCC strand and the concrete strength of 28-30 MPa, the average transfer length increased by 26% with an increase in the prestress level of 25%. For the Ø15.2 mm CFCC strand and the concrete strength of 22-24 MPa, the transfer length increased by 25% with an increase in prestress level of 19%. Therefore, the authors [31,32] stated that the prestress level can have a greater influence on the transfer length of the CFCC strand than the concrete compressive strength due to similar concrete strengths and a greater reduction of the transfer length at a higher prestress level. However, the results from [29,30,39] showed that the transfer length of CFCC strands is not significantly influenced by the prestress level. Ehsani et al. [47] calculated that the transfer length of the AFRP Arapree bars (Ø9.9 mm) with a smooth surface increased by 81% with an increase in the prestress level of 43% to 59% with concrete strength of 29-31 MPa.
The authors of [36,37] tested concrete beams prestressed with GFRP bars with ribbed surfaces and calculated that for Ø16 mm GFRP bar and concrete strength of 29-31 MPa, the increase in prestress level from 26% to 41% induced a 32% increase in transfer length.
It is evident that in most cases for CFRP Leadline bars, CFCC strands, AFRP, and GFRP bars, the transfer length is proportional to the prestress level (Table 3).

Effect of Reinforcement Diameter
Several authors have investigated the influence of reinforcement diameter on the transfer length of pretensioned FRP reinforcement. Table 4 provides a summary of average transfer length results showing the influence of reinforcement diameter. The lower (Ø min ) and higher (Ø max ) reinforcement diameters correspond to the lower (L t.min ) and higher (L t.max ) transfer lengths, respectively. Domenico and colleagues [29,30] and Mahmoud and colleagues [31,32] calculated that the transfer length of CFCC strands increased by 9-34% and 32%, respectively, with an increase in reinforcement diameter from 12.5 mm to 15.2 mm. Stark and Hegger [26] tested UHPFRC (ultra-high-performance fiber-reinforced concrete) beams prestressed with CFCC 7-wire helical indented strands (Ø7.5 mm) and CFRP spirally indented bars (Ø5 mm). The test results showed that the transfer length of the CFCC strand was 73% higher than the CFRP bar. As both reinforcements had indented surfaces, the helical form of the CFCC 7-wire strand may have an additional positive influence on bond conditions. Nanni and colleagues [24,38] calculated that at a low prestress level (24%), the transfer length increased by 38% with an increase in the diameter of the AFRP bar (with a smooth braided surface) from 8 mm to 12 mm. For a higher prestress level (about 50%), the transfer length increased to 42% with an increasing diameter of the AFRP bar from 8 mm to 16 mm. Furthermore, Taerwe and Pallemans [35,48] showed that the transfer length increased by 53% with an increasing diameter of the AFRP Arapree bar (with sanded surface) from 5.3 mm to 7.5 mm (the cross-sectional area of reinforcement increased twice).
In [36,37], it was calculated that with the increase in GFRP bar diameter from 12 mm to 16 mm, the average transfer length increased by 19%.
The FEM model of Motwani and Laskar [28] showed that the peak elastic strain of concrete at the level of prestressed CFCC and AFRP Arapree reinforcement increases with an increase in reinforcement diameter.
The results for the influence of reinforcement diameter on the transfer length of pretensioned CFRP, CFCC, AFRP, and GFRP reinforcement showed that it is generally consistent with the literature and that the transfer length is directly proportional to the diameter of the prestressed FRP reinforcement. However, expressing the transfer length in terms of the bar diameter only can result in inaccurate results, as it also depends on the prestressing level and the concrete compressive strength.

Effect of Concrete Cover
The effect of concrete cover is important for the durability of prestressed concrete structures. However, the FRP reinforcement covers this aspect. Therefore, a sufficient concrete cover is more relevant to controlling concrete splitting during the release of pretensioned reinforcement. Additionally, it is more relevant to analyze the influence of concrete cover on the cracking of the anchorage zone and the transfer length of the FRP reinforcement through the ratio between the concrete cover and the reinforcement diameter (c/Ø).
In [29,30], it was concluded that concrete cover c/Ø of 4 and 5 had no significant influence on the transfer length results for the CFCC strand sizes of Ø12.5 mm and Ø15.2 mm, respectively, for concrete strength of 30-56 MPa and prestress level of 50-75% for a given range of concrete cover (from 50 mm to 75 mm). According to [31,32], a concrete cover of 4·Ø is sufficient to prevent concrete splitting of the anchorage zone of the CFCC strand and the CFRP Leadline bar during reinforcement release for concrete strength between 22 MPa and 42 MPa, reinforcement diameter of 8-15.2 mm, and prestressing level of 60-80%. Therefore, with sufficient concrete confinement, all prestress force can be transferred to the concrete. However, all tested beams had a debonded length of 50 or 100 mm at the end of the beam. Therefore, the radial pressure moved deeper into the beam and could increase the splitting resistance of the transfer zone. Additionally, Stark and Hegger [26,49] calculated that a concrete cover c/Ø of at least 3.0 (absolute concrete cover 22.5 mm) is required for CFCC strand (Ø7.5 mm) and a CFRP bar (Ø5 mm) c/Ø of at least 4.0 (absolute concrete cover 20 mm) is required to ensure the crack-free introduction of the prestressing force for UHPFRC concrete beams.
Taerwe and Pallemans [35] calculated that for sand-coated AFRP bars (Ø5.3 mm and Ø7.5 mm), the critical concrete cover to be used to avoid concrete splitting is approximately equal to 2.8 times the bar diameter (2.8·Ø) for concrete strength of 54 MPa and prestress level of 50%. Additionally, the test results showed that the transfer length decreased with an increase in c/Ø up to approximately 3.5. The further increase in concrete cover did not have a significant influence on the transfer length of the AFRP bars.
Khin et al. [50] performed tested the minimum concrete cover (5 mm, 10 mm, and 15 mm) of concrete prisms (100 × 100 × 600 mm) pretensioned with CFRP, AFRP (prestress level of 60%), and GFRP (prestress level of 50%) bars (Ø8 mm). Based on the experimental test results, the recommended minimum concrete cover was 15 mm (c/Ø = 1.9) for the CFRP bar, but for the GFRP and AFRP bars, the concrete cover could be greater than 15 mm (c/Ø > 1.9). This could be related to the lower modulus of elasticity of the GFRP and AFRP reinforcement, which induces higher splitting stresses in the anchorage zone due to the Hoyer effect.
Motwani and Laskar [28] performed simulation tests for the anchorage zone of the prestressed CFCC strand and AFRP Arapree bar. They established that significant amounts of plastic strain appeared near the prestressing reinforcement, indicating concrete cracking at transfer for Ø12.7 mm, c = 30 mm, and c/Ø = 2.4. Additionally, they determined that the peak elastic strain of concrete at the level of prestressed CFCC and AFRP Arapree reinforcement decreases with an increase in concrete cover.
The small concrete cover could lead to the splitting of the concrete at transfer due to the insufficient confinement of the concrete. Therefore, the additional confinement of the concrete can reduce the risk of splitting the concrete during reinforcement release.

Effect of Shear Reinforcement
As stated in the previous section, additional concrete confinement can reduce the risk of concrete splitting during the release of pretensioned FRP reinforcement and therefore have a positive effect on the transfer length. Several authors have investigated the different types of concrete confinement at the anchorage zone of pretensioned FRP reinforcement. Table 5 provides a summary of the average transfer length results showing the influence of shear reinforcement. c-concrete cover, Ø-reinforcement diameter, E p -modulus of elasticity of reinforcement, f pi -initial prestress of reinforcement, f pu -tensile strength of reinforcement, f ci -concrete compressive strength at release, L t (with stirrups)-transfer length of specimen with shear reinforcement, L t (without stirrups)-transfer length of specimen without shear reinforcement.
Saeed [45] tested concrete beams prestressed with Ø12.7 mm CFRP sand-coated and helically wrapped bars. The distance between shear reinforcement of 75 mm and 50 mm for the level of prestressing of 50% and 55% for beams B2-4-55 and B3-4-60, respectively, was analyzed. The shorter distance between the shear reinforcements increased the confinement and decreased the transfer length (495 mm) in beam B3-4-60 compared with the transfer length (533 mm) in beam B2-4-55 despite the higher prestress level in beam B3-4-60. Finally, it was concluded that confinement, represented by the number of stirrups, played a significant role in the bond behavior between the concrete and CFRP bars [45]. Mahmoud and colleagues [31,32] calculated that for the Ø12.5 mm CFCC strands, the addition of shear reinforcement reduced the average transfer length by 21%. For the Ø8 mm CFRP Leadline bar, the addition of shear reinforcement reduced the transfer length by 6%.
Nanni et al. [57] investigated the influence of different levels of concrete confinement on the splitting of concrete during the release of smooth braided AFRP bars. It was calculated that cracking of the anchorage zone can be avoided by using a small-pitch (25 mm) CFRP coil (length 225 mm) that confines the concrete immediately surrounding the AFRP bar or partially blanketing the AFRP bar, intermittently applying the tape for lengths of one and two tendon diameters (16 and 32 mm) at a spacing of one tendon diameter over total lengths of 240 and 272 mm, respectively. However, shear reinforcement (Ø6 mm) (distance between stirrups Polymers 2022, 14, 3931 9 of 37 20 mm and 25 mm) does not prevent cracking. The transfer lengths of the specimens with shear reinforcement, CFRP coil, and partial blanketing were 625, 600, and 700 mm, respectively. Partial blanketing was determined to be very effective due to the reduction in tensile stress field despite an increase in transfer length of up to 17%.
Despite the low difference between prestress level and concrete compressive strength, it can be stated that specimens with shear reinforcement can induce a reduction in the transfer length of CFCC strands and the CFRP Leadline bar [31,32] (Table 5). Additionally, it was concluded in [31,32] that the absence of shear reinforcement in specimens pretensioned with CFRP Leadline and CFCC strand increases the transfer length with an averages of 10% and 17%, respectively, more than the values predicted by the proposed theoretical models.

Effect of Reinforcement Surface Conditions
The bond properties between the reinforcement and concrete are influenced by different surface roughnesses of the reinforcement. Furthermore, reinforcement surface conditions can enhance the anchorage zone behavior of pretensioned reinforcement. Table 6 provides a summary of average transfer length results that shows the influence of reinforcement surface conditions. Ehsani et al. [47] investigated three different types of AFRP bars. The lowest transfer length (327 mm) was calculated for the AFRP Technora bar with a rough (ribbed) surface. For the AFRP Fibra bar with a smooth braided surface, the transfer length was longer (339 mm), and the transfer length of the AFRP Arapree bar with a smooth (plain) surface was the longest (385 mm). Despite the differences in reinforcement diameter (Ø7.4-10.4 mm), it can be concluded that with increasing surface roughness of AFRP reinforcement, the transfer length decreases.
Additionally, in [35,48], three different types of surface finishing (sanded, "expancel" and expancel sanded) for a Ø5.3 mm AFRP Arapree bar were investigated. The expancel coating was applied as a thin compressible layer around the bar to absorb part of the radial expansion of the bar due to thermal expansion and the Hoyer effect [35,48]. The test results revealed that the transfer length of the AFRP bars with the expancel coating (smooth surface) was greater by about 60-80% compared with the transfer length of the AFRP bars with the sanded and expancel-sanded surfaces. Additionally, the reduction of the consequences of the Hoyer effect by applying an expancel coating to the AFRP bar reduces the c/Ø up to approximately 2.33·Ø [35,48]. This indicates that the sand coating of the AFRP bars can significantly decrease the transfer length. However, it can increase the risk of concrete splitting during reinforcement release; additional concrete confinement can reduce the risk of splitting concrete.
Nanni and colleagues [24,38] calculated that the transfer length (225 mm) of sandcoated AFRP bars (Ø12 mm) was more than half the transfer length (450 mm) of smooth braided AFRP bars (Ø12 mm) when the prestress level was 48%. Although the concrete strength of sand-coated AFRP bars (39 MPa) was higher compared with the braided AFRP bars (34 MPa), the silica sand on the AFRP bars with respect to that of smooth bar increased the bond characteristics and significantly reduced the transfer length of the AFRP bar.
The test results revealed that the ribbed surface of the GFRP bar [36,37] and the sandcoated surface of the AFRP bar [24,38] give similar transfer length results (221 mm and 225 mm, respectively) despite different material properties (modulus of elasticity of 60 GPa and 68 GPa, concrete strength of 31 MPa and 39 MPa, and prestress level of 36% and 48%, respectively).
The test results of Khin et al. [50] showed that CFRP reinforcement with increased surface roughness (spirally indented, sanded, and cross-wounded) had a shorter transfer length (300 mm) than CFRP reinforcement with a helical plain surface (400 mm). Additionally, AFRP and GFRP spirally indented bars had similar moduli of elasticity (54 GPa and 47 GPa respectively) and the same transfer length (500 mm).

Effect of Modulus of Elasticity of Reinforcement
The difference in the bond characteristics in the anchorage zone of reinforcement can be attributed to the different moduli of elasticity of different FRP reinforcements. A lower modulus of elasticity causes more longitudinal deformation during prestressing and consequently more transverse deformation due to the Poisson ratio during the release of reinforcement for the same prestress level. Higher transverse deformations of FRP reinforcement with a lower modulus of elasticity improve bond strength at the transfer zone due to the lateral expansion of the bar, creating a wedge action known as the Hoyer effect. Table 7 provides a summary of the average transfer length results showing the influence of modulus of elasticity of reinforcement. According to [51,52], the bond characteristic of the GFRP reinforcement is superior to that of steel. This can be attributed to the lower modulus of elasticity of the GFRP reinforcement. It could also be due to the better adhesion and interlock between the GFRP reinforcement and the concrete. Thus, for the same force, the transfer length for the GFRP reinforcement is smaller. Additionally, Zawam and Soudki [36] stated that the lower modulus of elasticity of the GFRP bar will induce more radial expansion compared with CFRP, resulting in an increase in the confining stresses normal to the tendon during the prestress transfer process. This characteristic improved the bond strength at the interface between the bar and the surrounding concrete upon release, resulting in a shorter transfer length compared with CFRP bars [36].
Zou [33] calculated that the transfer length of CFRP Leadline bars (Ø8 mm) was 2-3.8 times higher compared with the transfer length of the AFRP Arapree sanded bar (Ø7.8 mm). The large difference in transfer length can be attributed to the significantly lower modulus of elasticity of the AFRP bar (54 GPa) compared with the CFRP Leadline bar (172 GPa). Additionally, the sanded surface of the AFRP bar can enhance bonding conditions, in contrast with the spirally indented CFRP Leadline bar.
Ehsani et al. [47] calculated that the transfer length of the CFCC (helical plain surface) and CFRP Leadline (spirally indented surface) reinforcement was similar (432 mm). However, the transfer length of the AFRP Technora bar (rough surface) was shorter (327 mm) and may be related to the higher surface roughness and lower modulus of elasticity (69 GPa) of the AFRP bar compared to the CFCC strand and CFRP bar.
Lu et al. [53,54] calculated that the transfer length of the AFRP Technora bar (rough surface) was 13% shorter compared with the CFRP Leadline bar (spirally indented surface). It can be related to a higher surface roughness and a lower modulus of elasticity (45 GPa) of the AFRP bar compared to the CFRP Leadline bar (171 GPa).
Since the modulus of elasticity of FRPs is generally less than that of steel, the transfer length of most FRPs is less than that of steel [11].

Dead End versus Live End for Transfer Length Results
Different transfer lengths may appear at two ends of the pretensioned concrete member. This can be attributed to more than one specimen prestressed with the same reinforcement. In this case, one end of the reinforcement is prestressed (Live end), while the other at the same time is fixed to the rigid support (Dead end). During the release of prestressed reinforcement, differences can appear in the transfer lengths of different FRP reinforcements.
In [58], the measured transfer lengths at the live and dead ends of the box beams were in close agreement with the sudden release of the CFCC strands and CFRP Leadline bars. The same trend was identified for the gradual release of CFRP Leadline bars [45], AFRP bars [24], and GFRP ribbed bars [36,37].
The experimental results provided in [59][60][61] showed that the transfer lengths of the CFCC strands at the live ends are higher than the transfer lengths at the dead ends for the sudden release. As explained by [62], this could be influenced by factors such as concrete casting location, cutting location, and the use of multiple batches of concrete. However, the average dead end to live end ratios of transfer lengths ranged from 0.74 for pile 3 to 0.86 for piles 4 and 5. This shows that the casting location had minimal to no effect on the transfer length. Additionally, Krem [43] calculated that the transfer length at the live end was slightly greater than at the dead end for the same beam. The increase in transfer length at the live end could be a result of the dynamic impact of the sudden release of the CFRP bar.
The results of Crossett et al. [55] showed that the transfer length of the BFRP bars is 16-48% higher at the live end than at the dead end. Additionally, it was explained that the release of stress is more gradual at the dead (anchoraged) end, and therefore, shorter transfer lengths were calculated [53].

Long-Term Effect
The transfer of the prestressing force to the concrete during the release of reinforcement is a short-term effect. However, the prestressed concrete member is intended for long-term operation. After the release of the prestressed reinforcement, the concrete members are affected by many long-term effects (shrinkage and creep of the concrete, relaxation of the reinforcement, etc.) that can influence the anchorage zone and transfer length of the prestressed concrete member.
Zou [33] stated that the transfer length of AFRP Arapree sanded bars is not affected by time (up to 238 days), although the average concrete strain beyond the transfer length increases significantly. Nanni et al. [24] have also drawn the same conclusion for smooth braided AFRP bars for up to three weeks.
According to Zou [33], there is no apparent deterioration of the bond between the CFRP Leadline bar and the concrete with time, and the measured transfer length did not change with time for normal and high strength concrete beams. Furthermore, the application of two-point loads in the midspan did not affect the transfer length up to 390 days.
Grace [39] showed that the increase in the transfer length of CFRP Leadline bar after 300 days and the CFCC strands after 391 days is only about 7.8% and 7%, respectively, and there is no significant effect of time on the transfer length. Approximately the same result (6%) is for steel strands [63].
Soudki et al. [46] calculated that 200 days after CFRP Leadline bar (prestress level of 50%) release for the T-beam specimen, no significant change in the measured transfer length was observed.
Lu et al. [54] also determined that there are little changes in transfer length with time for CFRP Leadline and CFRP (both spirally indented) and bars after 28 days and for AFRP Teachnora (rough) bars after 90 days.
Mahmoud et al. [31,32] determined that the helical shape of the CFCC and steel strands enhanced the mechanical component of the bond and did not result in an increase in the transfer length over time. The 22% increase in the transfer length of CFRP Leadline bar over time could be due to its relatively smooth surface compared to CFCC and steel strands.
In [51,52], it was calculated that the transfer length increased by 40% and 20% with time (after 600 days) for the GFRP and steel strands, respectively. However, the rate of increase was almost double for GFRP reinforcement compared with steel.
According to [55], the transfer length of the BFRP bars increased after 5 days of reinforcement release by 4-39%.
It is suggested that at the ends of the member, the concrete strain increased with an increase in time, mainly due to the shrinkage of the concrete [46]. Away from the ends, the increase in strain includes the combined effects of creep, shrinkage, and the relaxation of the prestressing reinforcement [24,33,46]. Table 8 presents a summary of the recommended expressions for the transfer length of the FRP reinforcement (Equation (1)-(3)) and steel strands (Equations (4)- (6)). The nomenclature in this table is presented.

Theoretical Models
Equation (5) provided in [64] is evaluating the fewest different parameters (f pi and Ø) influencing the transfer length. Additionally, with an empirical coefficient (20.7) which is based on the large database of transfer length results, Equation (5) was developed for steel strands. Mitchell et al. [63] suggested supplementing the ACI-318 [64] equation with the compressive strength at transfer. As concrete strength enhances the bond of reinforcement, it becomes a good additional parameter for increasing the accuracy of the transfer length prediction. Additionally, Equation (6) with the empirical coefficient (20.7) is proposed for the steel strands. Equation (3) proposed by Domenico [30] replaces the reinforcement diameter (Ø) with a cross-sectional area (A p ) of the reinforcement and proposes an empirical coefficient C T = 80 for the CFCC strand. Additionally, it calculates f 1/2 ci as also in Equations (2) and (6). However, Zou [33] stated that the influence of prestress on the transfer length of CFRP reinforcement is significantly lower than the concrete strength at transfer. Therefore, it can be neglected, and the transfer length can be calculated according to Equation (2) with an empirical coefficient κ = 480 (in N·mm units) for the prestressed CFRP Leadline bar. According to [58,65], the theoretical results of the transfer length of the CFRP bar predicted by Equations (1) and (5) were up to 10% and 6% higher compared with the experimental ones, respectively. For CFCC strands [58,66], the theoretical values predicted by Equations (1) and (5) were up to 15% higher and up to 12% lower compared with the experimental results, respectively. Furthermore, Equation (2) provided the least inaccurate prediction and the theoretical values of the transfer length of the CFRP and CFCC reinforcement were up to 2.8 and 2.9 times higher than the experimental ones, respectively. Therefore, these results contradict the concluding remarks of Zou [33], who states that for all practical purposes, the effect of prestress on the transfer length of the tendons can be neglected. Based on the measured data in [31,32], the transfer length of CFRP reinforcement is proportional to the reinforcement diameter, the initial prestress level, and the concrete compressive strength at transfer, and Equation (1) was proposed for the transfer length of CFRP reinforcement. Equation (1) was adopted in design codes [10,12]. The main difference from other theoretical models is that Equation (1) proposes an empirical coefficient α t dependent on the type of FRP reinforcement. Therefore, the coefficient α t can be calibrated for different types of FRP reinforcement (GFRP, CFCC, CFRP, AFRP, BFRP) with different surface conditions. Additionally, it presents the concrete strength as f 2/3 ci instead of f 1/2 ci (Equations (2), (3) and (6)). The presentation of concrete strength as f 2/3 ci can be explained by the correlation of concrete compressive strength with concrete tensile strength f ctm = 0.3·f 2/3 ci provided in [67 -69]. Additionally, Marti-Vargas et al. [70] proposed Equation (4) for prestressed seven-wire steel strands, which is similar to Equation (1). The only differences are in the empirical coefficient (2.5) and that the reinforcement diameter (Ø) is replaced with an area of cross-section of reinforcement (A p ). Table 8. Summary of the theoretical models for the transfer length of FRP reinforcement.

Reference Equation Equation
No. Notes [10,12,31,32] L t -is the transfer length, f pi -is the initial prestress level, Ø-is the reinforcement diameter, f ci -is the concrete compressive strength at the time of transfer, α t -is a material dependent coefficient.
κ-is a factor equal to 480 in N·mm units (for the CFRP Leadline bar with a spirally indented surface), Ø-is the reinforcement diameter, f ci -is the concrete compressive strength at the time of transfer. [30] L t = A p -cross-sectional area of prestressed reinforcement, f pe -effective prestressing stress in the CFCC strand, C T -constant is equal to 80 for CFCC strands, f ci -is the concrete compressive strength at the time of transfer.
A p -cross-sectional area of prestressed reinforcement, f pi -is the initial prestress level, f ci -is the concrete compressive strength at the time of transfer. [64] L t = f pi ·Ø 20.7 (5) f pi -is the initial prestress level, Ø-is the reinforcement diameter. [63] f pi -is the initial prestress level, Ø-is the reinforcement diameter, f ci -is the concrete compressive strength at the time of transfer. Table 8 presents a summary of the recommended expressions for the transfer length of the FRP reinforcement (Equation (1)-(3)) and steel strands (Equations (4)-(6)). Up until now, Equation (1) is still the main equation to calculate the transfer length of CFRP and CFCC reinforcements. There are different types of FRP reinforcement (CFRP, CFCC, GFRP, AFRP, BFRP) with different surface conditions and material properties. Therefore, other researchers proposed values of the empirical coefficient α t for different types of FRP reinforcement ( Table 9). The values proposed by [31,32] and adopted in [10,12] are α t = 1.9 and α t = 4.8 for CFRP Leadline bars and CFCC strands respectively. The authors of [39] proposed α t = 1.95 and α t = 2.12 for CFRP Leadline bars and CFCC strands, respectively. In [36,37], α t = 2.6 was proposed for GFRP ribbed bars. Additionally, [43,44] investigated specimens made of self-compacting concrete (SCC) and prestressed with CFRP bars and proposed α t = 2.84 − f pi /800.

Database of Transfer Length Results
A literature review of the experimental results of the transfer lengths of different types of pretensioned FRP reinforcement was performed. In total, 318 specimens were found. Table 10 (Table 1), CFCC (Table 2), CFRP (Table 3), AFRP (Table 4), and BFRP (Table 5) reinforcement were found with the transfer length results, respectively. Tables 1-5 present the original specimen number from the experimental research, type of FRP reinforcement, surface conditions of FRP reinforcement, specimen type and dimensions (b × h × l-width and height of the cross-section and length of the specimen), release type of prestressed reinforcement, information about the existence of shear reinforcement in the cross-section of the specimen, protective concrete cover (c), reinforcement diameter (Ø), cross-sectional area of one prestressed bar (A p ), modulus of elasticity (E p ), tensile strength (f pu ), initial stresses (f pi ) of reinforcement, the ratio between initial stresses and tensile strength of reinforcement (f pi /f pu ), concrete compressive strength at transfer (f ci ), average transfer length (L t ) of pretensioned reinforcement.

Analysis of Experimental Results
The experimental results for the transfer length sof different types of FRP reinforcement (GFRP, CFCC, CFRP, AFRP, and BFRP) found in the literature are analyzed in this chapter. Figure 1 shows the distribution of the experimental transfer lengths (L t ) of different types of pretensioned FRP reinforcement under the influence of different experimental parameters: concrete compressive strength at transfer (f ci ), the ratio between initial stress in pretensioned reinforcement, and tensile strength (f pi /f pu ), reinforcement diameter (Ø), concrete protective cover (c), the ratio between the concrete cover and reinforcement diameter (c/Ø) and modulus of elasticity of FRP reinforcement (E p ). The ranges of influential parameters (f ci , f pi /f pu , Ø, c, c/Ø and E p ) presented in Figure 1 are presented in Table 10.
The results show that the transfer lengths of different types of pretensioned FRP reinforcement decrease with increasing concrete compressive strength at the time of pretensioned reinforcement release (Figure 1a). In general, the increased compressive strength of concrete enhances the bond between the reinforcement and the concrete. Therefore, it provides better anchorage properties for the pretensioned reinforcement. Research by other authors confirms the trend that the transfer length decreases with the increase of concrete strength. However, according to the results of Zou [25] and Zawam with colleagues [26,27], the rate of decrease in transfer length is lower for the AFRP and GFRP bars. Therefore, the concrete strength is more significant for CFCC and CFRP reinforcement. The results show that the transfer lengths of different types of pretensioned FRP reinforcement decrease with increasing concrete compressive strength at the time of pretensioned reinforcement release (Figure 1a). In general, the increased compressive strength of concrete enhances the bond between the reinforcement and the concrete. Therefore, it provides better anchorage properties for the pretensioned reinforcement. Research by other authors confirms the trend that the transfer length decreases with the increase of concrete strength. However, according to the results of Zou [25] and Zawam with colleagues [26,27], the rate of decrease in transfer length is lower for the AFRP and GFRP bars. Therefore, the concrete strength is more significant for CFCC and CFRP reinforcement.
The transfer length of different types of pretensioned FRP reinforcement tends to increase with an increase in the fpi/fpu ratio (Figure 1b). The higher prestress level transfers a higher prestress force to the concrete, which induces greater damage to the bond between the reinforcement and the concrete. Therefore, the transfer length is usually higher for a higher level of prestress. Some results in the literature can be found that show little influence of prestress level on the transfer length. It can be attributed to CFCC [29,30,39]  The transfer length of different types of pretensioned FRP reinforcement tends to increase with an increase in the f pi /f pu ratio (Figure 1b). The higher prestress level transfers a higher prestress force to the concrete, which induces greater damage to the bond between the reinforcement and the concrete. Therefore, the transfer length is usually higher for a higher level of prestress. Some results in the literature can be found that show little influence of prestress level on the transfer length. It can be attributed to CFCC [29,30,39] and CFRP [31,32,39,45,46] reinforcement (Table 3). On the other hand, in [31,32] it is stated that the prestress level may have a greater effect on the transfer length of CFCC strands compared with the concrete strength. Despite some discrepancies between the results of the transfer length of CFCC and the CFRP reinforcement of different authors, the main trend of increasing the transfer length with an increase in prestress level remains for different types of FRP reinforcement (GFRP, CFCC, CFRP, AFRP) (Figure 1b).
The transfer lengths of different types of pretensioned FRP reinforcement (CFCC, CFRP, and AFRP) increase with increasing diameter of reinforcement (Figure 1c). For the larger reinforcement diameter, the prestress force will be higher than for the lower reinforcement diameter for the same prestress level. Therefore, higher prestress force transferred to the concrete will cause more damage to the surface between the reinforcement and the concrete. Additionally, this damage will be distributed in a larger contact area with a larger reinforcement diameter. Therefore, the damage to the bond of a larger reinforcement diameter will be larger and will cause a longer transfer length. However, the results of GFRP reinforcement are the opposite (Figure 1c). Therefore, it is evident that reinforcement diameter alone cannot be assessed as the main factor influencing the transfer length.
The transfer lengths of different types of pretensioned FRP reinforcement (CFCC, CFRP, and AFRP) increase with an increase in concrete protective cover (Figure 1d). The release of pretensioned reinforcement induces splitting tensile stresses in concrete due to the Hoyer (wedge) effect, which is the result of the swelling of the reinforcement at transfer. Therefore, the concrete protective cover plays an important role in restricting the splitting of concrete and maintaining sufficient reinforcement bond conditions at transfer. Additional confinement may be introduced in the form of shear stirrups or spiral reinforcement to take over splitting tensile stresses. As in the case of reinforcement diameter (Figure 1c), the transfer length of the GFRP reinforcement decreases with increasing concrete cover (Figure 1d). Therefore, it shows that the transfer length should be compared by evaluating the combination of several influencing factors. Additionally, the concrete cover is more important in the case of durability and splitting of the concrete anchorage zone at transfer.
As in the case of reinforcement diameter (Figure 1c) and concrete cover (Figure 1d), the transfer lengths of different types of pretensioned FRP reinforcement increase with increasing ratio c/Ø (Figure 1e). It should be noted that the increase in the ratio c/Ø with an increase in transfer length is significantly reduced compared to the results presented in Figure 1c,d. Therefore, the evaluation of these two parameters (c and Ø) provides a more uniform distribution of the relationship between the transfer length and c/Ø. Additionally, Taerwe and Pallemans [35] determined that for sand-coated AFRP bars the transfer length decreased with an increase of c/Ø up to approximately 3.5. The further increase in concrete cover did not have a significant influence on the transfer length of the AFRP bars. Therefore, it shows the importance of choosing the appropriate value of the concrete cover (c) for a certain reinforcement diameter (Ø). Additionally, it is again proven that the transfer length should be evaluated in a combination of several influencing factors. The ratio c/Ø is important to ensure the crack-free introduction of the prestressing force into the concrete, while the absolute concrete protective cover is important for the durability of the member.
A comparison of the transfer length and modulus of elasticity of different types of FRP reinforcement shows that in most cases, the transfer length of FRP reinforcement (GFRP, CFCC, and CFRP) increases with increasing modulus of elasticity (Figure 1f). In [36,51,52], it was stated that a lower modulus of elasticity will cause a greater swelling of the reinforcement, resulting in an increase in the confining stresses at transfer. It improves the bond strength of the reinforcement, resulting in a shorter transfer length [36]. The results of [33,47,53,54] showed that the transfer length of AFRP bars was lower than that of the CFRP Leadline bar due to the lower modulus of elasticity of the AFRP reinforcement. However, Figure 1f shows that the transfer length of AFRP bars is slightly decreasing with increasing modulus of elasticity. This may be related to the higher range of modulus of elasticity of AFRP reinforcement (Table 10) and the additional variation of other parameters (nonidentical).
The transfer length of the BFRP bar also decreases with an increase in concrete strength (Figure 1a). However, the trends of other parameters (f pi /f pu , Ø, c, c/Ø, and E p ) with respect to the transfer length are opposite compared to GFRP, CFCC, CFRP, and AFRP reinforcements. The results of only six specimens pretensioned with BFRP reinforcement are presented in the database of transfer length ( Table 5). The number of specimens and the variation of the parameters are too small. Therefore, it is difficult to distinguish a clear influence of different parameters on the transfer length of BFRP bars. Therefore, broader experimental research needs to be performed for a better understanding the behavior of anchorage zone of prestressed BFRP bar.
The prestress force can be transferred to the concrete in two ways: gradually or suddenly. Therefore, the influence of the type of release of CFCC reinforcement (sudden or gradual) on the transfer length and parameters (f ci , f pi /f pu , Ø, c, c/Ø and E p ) influencing the transfer length is presented in Figure 2. It can be seen that there is a clear difference between the results, and in most cases, a sudden release gives higher transfer length results compared to a gradual release of CFCC strands. The same tendency is observed with pretensioned steel strands. The influence of reinforcement release type on the transfer length of CFCC strand can be affected by the the similar shape of the the CFCC and steel strand (seven helically wounded wires). elasticity of AFRP reinforcement (Table 10) and the additional variation of other parameters (nonidentical).
The transfer length of the BFRP bar also decreases with an increase in concrete strength (Figure 1a). However, the trends of other parameters (fpi/fpu, Ø, c, c/Ø, and Ep) with respect to the transfer length are opposite compared to GFRP, CFCC, CFRP, and AFRP reinforcements. The results of only six specimens pretensioned with BFRP reinforcement are presented in the database of transfer length (Table A5). The number of specimens and the variation of the parameters are too small. Therefore, it is difficult to distinguish a clear influence of different parameters on the transfer length of BFRP bars. Therefore, broader experimental research needs to be performed for a better understanding the behavior of anchorage zone of prestressed BFRP bar.
The prestress force can be transferred to the concrete in two ways: gradually or suddenly. Therefore, the influence of the type of release of CFCC reinforcement (sudden or gradual) on the transfer length and parameters (fci, fpi/fpu, Ø, c, c/Ø and Ep) influencing the transfer length is presented in Figure 2. It can be seen that there is a clear difference between the results, and in most cases, a sudden release gives higher transfer length results compared to a gradual release of CFCC strands. The same tendency is observed with pretensioned steel strands. The influence of reinforcement release type on the transfer length of CFCC strand can be affected by the the similar shape of the the CFCC and steel strand (seven helically wounded wires).

Derivation of the Coefficient α t
Previous analysis of the results (Figures 1 and 2) showed that the influence of separate factors on the transfer length of prestressed FRP reinforcement does not always provide a qualitative comparison. Equation (1) provided in Table 8 is proposed for prestressed FRP reinforcement and takes into account initial prestress (f pi ), reinforcement diameter (Ø), and concrete compressive strength at transfer (f ci ). Therefore, for a better analysis of the results based on Equation (1), a graphical comparison of the transfer length of different FRP reinforcements (GFRP, CFCC, CFRP, and AFRP) and f pi ·Ø/f 2/3 ci is presented in Figures 3-6. Additionally, these graphs represent a distribution of the results with the proposed average values for the coefficient α t .
concrete cover (c), (e) ratio between the concrete cover and reinforcement diameter (c/Ø), (f) modulus of elasticity (Ep) of different types of CFCC pretensioned reinforcement under different methods of prestress transfer.

Derivation of the Coefficient αt
Previous analysis of the results (Figures 1 and 2) showed that the influence of separate factors on the transfer length of prestressed FRP reinforcement does not always provide a qualitative comparison. Equation (1) provided in Table 8 is proposed for prestressed FRP reinforcement and takes into account initial prestress (fpi), reinforcement diameter (Ø), and concrete compressive strength at transfer (fci). Therefore, for a better analysis of the results based on Equation (1), a graphical comparison of the transfer length of different FRP reinforcements (GFRP, CFCC, CFRP, and AFRP) and Figures 3-6. Additionally, these graphs represent a distribution of the results with the proposed average values for the coefficient αt.   concrete cover (c), (e) ratio between the concrete cover and reinforcement diameter (c/Ø), (f) modulus of elasticity (Ep) of different types of CFCC pretensioned reinforcement under different methods of prestress transfer.

Derivation of the Coefficient αt
Previous analysis of the results (Figures 1 and 2) showed that the influence of separate factors on the transfer length of prestressed FRP reinforcement does not always provide a qualitative comparison. Equation (1) provided in Table 8 is proposed for prestressed FRP reinforcement and takes into account initial prestress (fpi), reinforcement diameter (Ø), and concrete compressive strength at transfer (fci). Therefore, for a better analysis of the results based on Equation (1), a graphical comparison of the transfer length of different FRP reinforcements (GFRP, CFCC, CFRP, and AFRP) and Figures 3-6. Additionally, these graphs represent a distribution of the results with the proposed average values for the coefficient αt.   It was calculated that for the database of GFRP reinforcement, the average coefficient α t is 2.58 with the standard deviation (STD) of 0.33 and the coefficient of variation (COV) of 12.8% (Table 8) (for the concrete strength 29-71 MPa, the level of prestressing 26-47% and the diameter of the reinforcement 9.5-16 mm). The graphical representation is presented in Figure 3a with a very strong correlation coefficient R 2 = 0.98. Almost all GFRP bars had ribbed surfaces, and the influence of the reinforcement surface conditions could not be determined. Additionally, Figure 3b shows that the influence of shear reinforcement and the type of prestressed reinforcement release (sudden or gradual) has an almost negligible effect on the transfer length of the GFRP reinforcement. However, it should be mentioned that the database of transfer length results of GFRP reinforcement is quite small (Table 1) compared with other FRP types (Tables 2-4): only five specimens reinforced with shear reinforcement and prestressed with GFRP reinforcement affected by a sudden transfer of prestressing were found in the literature. Therefore, more research is needed to investigate the effect of shear reinforcement and the type of release of prestressed GFRP reinforcement on the transfer length.  It was calculated that for the database of GFRP reinforcement, the average coefficient αt is 2.58 with the standard deviation (STD) of 0.33 and the coefficient of variation (COV) of 12.8% (Table 8) (for the concrete strength 29-71 MPa, the level of prestressing 26-47% and the diameter of the reinforcement 9.5-16 mm). The graphical representation is presented in Figure 3a with a very strong correlation coefficient R 2 = 0.98. Almost all GFRP  It was calculated that for the database of GFRP reinforcement, the average coefficient αt is 2.58 with the standard deviation (STD) of 0.33 and the coefficient of variation (COV) of 12.8% (Table 8) (for the concrete strength 29-71 MPa, the level of prestressing 26-47% and the diameter of the reinforcement 9.5-16 mm). The graphical representation is presented in Figure 3a with a very strong correlation coefficient R 2 = 0.98. Almost all GFRP The type of CFCC strand release has already been shown to affect the transfer length even when the influence of separate factors was analyzed ( Figure 2). Furthermore, the effect is more obvious when comparing the transfer length with f pi ·Ø/f 2/3 ci (Figure 4a).  (Table 11) (for concrete strength of 37-48 MPa, level of prestressing 30-65% and diameter of reinforcement 12.5-15.2 mm). Additionally, the higher variation of the results (27.7% and 12.5% for sudden and gradual release, respectively) of the sudden type of reinforcement release shows that this type of release is more dangerous and can cause more damage to the anchorage zone of prestressed CFCC strands and, therefore, increase the transfer length. According to Figure 4b, there is no clear influence of shear reinforcement on the transfer length of the CFCC strand. All CFCC seven-wire strands in the collected database have helical plain or almost plain surfaces, and therefore, their effect on the transfer length could not be determined.  Figure 5a presents the transfer length distribution of CFRP bars with respect to f pi ·Ø/f 2/3 ci with an average α t = 1.92 with STD = 0.48, COV = 24.8%, and R 2 = 0.94 (Table 11) (for concrete strength of 26-101 Mpa, prestress level 26-86%, and reinforcement diameter 5.3-12.7 mm). According to Figure 5b,c, there is no influence of the type of transfer of prestressing and shear reinforcement on the transfer length of CFRP bars. Most CFRP bars were Leadline bars with a spirally indented surface (Table 3), and a few studies were performed on spirally indented, sanded, and spirally indented and sanded CFRP bars. Khin et al. [50] determined that CFRP reinforcement with spirally indented or sanded surface had up to 25% lower transfer length compared with CFRP reinforcement with a helical plain surface. The trend lines show that the higher the roughness of the CFRP bar surface, the lower the transfer length ( Figure 5d). However, the scatter of the bigger database results is too large to qualitatively confirm this trend. Figure 6a presents the transfer length distribution of AFRP bars with respect to f pi ·Ø/f 2/3 ci with an average α t = 2.9 with STD = 1.8, COV = 62.1%, and R 2 = 0.84 (Table 11) (for concrete strength of 27-81 MPa, prestressing level 23-82% and reinforcement diameter 5.3-16 mm). The variation in the results is very high (62%). Therefore, the additional influence of other variables was investigated on the transfer length of the AFRP bars. According to Figure 6b, shear reinforcement may have some influence on the transfer length of AFRP bars. However, since the transfer length is shown to be higher for specimens with shear reinforcement, it contradicts the results for GFRP (Figure 3b) and CFCC (Figure 4b) reinforcement. Therefore, it was decided that the influence of shear reinforcement will be neglected in determining the coefficient α t . Additionally, Figure 6c shows that two reinforcement surface conditions can be distinguished for the AFRP bars. Therefore, the transfer length is higher for smooth braided AFRP bars compared to sanded and rough FRP bars. The same trend of decrease in transfer length with an increase in reinforcement surface roughness was observed by [24,35,38,47,48]. Additionally, the ribbed surface of the GFRP bar [36,37] and the sand-coated surface of the AFRP bar [24,38] gave similar transfer length results; the increased roughness of reinforcement increases the bond between reinforcement and concrete and therefore the transfer length. Furthermore, reinforcement classification according to surface roughness gives different results of α t = 1.53 with STD = 0.55, COV = 35.8%, and R 2 = 0.93 for smooth braided AFRP bars (for concrete strength of 29-39 MPa, prestressing level 23-58%, and reinforcement diameter 8-16 mm) and α t = 3.99 with STD = 1.70, COV = 42.7%, and R 2 = 0.86 for sanded and rough AFRP bars (Table 11) (for concrete strength of 27-81 MPa, prestressing level 37-82%, and reinforcement diameter 5.3-13.5 mm). These results show that an additional classification of AFRP bars according to surface conditions reduces the variation of the results from 62% to 36%-43%. However, this level of variation in the results is still high compared with GFRP, CFCC, and CFRP reinforcement. Therefore, the surface condition of AFRP bars should be evaluated in additional comprehensive research. Figure 7a presents the transfer length distribution of the BFRP bars with respect to f pi ·Ø/f 2/3 ci with an average α t = 2.1 with STD = 1.7, COV = 79.2%, and R 2 = 0.68 (Table 11) (for concrete strength of 27-52.4 MPa, prestressing level 34-48%, and reinforcement diameter 8-12 mm). The low number of results gives high variation in α t . Therefore, it cannot be applied to calculate the transfer length of BFRP bars. Additionally, Figure 7b shows a preliminary comparison of the transfer length and f pi ·Ø/f 2/3 ci with the influence of release type of reinforcement. The results show that sudden transfer of prestress induces a longer transfer length of the BFRP bar compared to gradual transfer of prestress (Figure 7b). This confirms the results of the CFCC strands influenced by different types of reinforcement release (sudden or gradual) (Figure 4a). Therefore, the influence of the type of BFRP reinforcement release and other important factors (f ci , f pi /f pu , c) on the transfer length could be the topic of future research.
GFRP, CFCC, and CFRP reinforcement. Therefore, the surface condition of AFRP bars should be evaluated in additional comprehensive research.  Figure 7b shows a preliminary comparison of the transfer length and

Comparison of Experimental and Theoretical Results
In this section, theoretical models for the calculation of transfer length (Table 8, Section 3) are compared with experimental results from the literature (Tables 1-4). In particular, models are presented by [10,12,31,32] (Equation (1)), [63] Equation (6), [64] Equation (5), and [30] Equation (3). Equation (2) proposed by Zou [33] is not analyzed due to the lack of correlation with the experimental results found in the literature. Furthermore, Equation (4) is not taken into account because it is proposed for steel strands [70] and is very similar to Equation (1) proposed for FRP reinforcement. The results presented in Figures 8-10 show the relationships between the experimental and theoretical results of the transfer length of different FRP reinforcements. In addition, they show the differences between different theoretical models for calculating the transfer length. The theoretical model proposed by [31,32] is presented in Figures 8-10 with the α t coefficients proposed in this article (Table 8).
From Figure 8a, it is clear that the theoretical models presented by [30,64] are less accurate than the models presented by [31,32,63]. Furthermore, the [64] model overestimates experimental results on average by 47% (L t,teor /L t.exp = 1.47) with STD = 0.32 and COV = 21.7% and [30] underestimates experimental results of the transfer length of prestressed GFRP reinforcement on average by 38% (L t,teor /L t.exp = 0.62) with STD = 0.19 and COV = 31.2%. The models of Mitchell [63] and Mahmoud [31,32] (α t = 2.6) show similar results with L t,teor /L t.exp = 1.05, STD = 0.13, COV = 12.1% and with L t,teor /L t.exp = 1.0 STD = 0.13, COV = 12.6%, respectively. The models presented by [31,32,63] are in the best agreement with the experimental results (difference up to 5%) and the variation of the results is the lowest (COV = 12%). Therefore, it shows the best agreement between theoretical and experimental results. In the case of CFRP reinforcement (Figure 8b), Equation (3) gives the most inappropriate results with an underestimation of the experimental results by 70% (L t,teor /L t.exp = 0.30) with STD = 0.12, COV = 40.2%. Equation (6) also underestimates the experimental results with L t,teor /L t.exp = 0.82, STD = 0.28, COV = 34%, but Equation (5) overestimates the results with L t,teor /L t.exp = 1.17, STD = 0.55, COV = 47.2%. Despite the relatively high variation of the results the most accurate prediction of experimental results was received by Equation (1) (α t = 1.9) with L t,teor /L t.exp = 1.01, STD = 0.33, COV = 32.5%. Figure 9 distinguishes the comparison of experimental and theoretical results of the transfer length of CFCC strands for the gradual (Figure 9a) and sudden (Figure 9b (Figure 9b) types of release with L t,teor /L t.exp = 1.0, STD = 0.12, COV = 12.5% and L t,teor /L t.exp = 1.02, STD = 0.27, COV = 26.6%, respectively. The higher variation in the results for sudden prestress transfer shows that it has more damage to the bond of reinforcement and results in a longer transfer length. Part of the results of Equation (6) coincide with the results of Equation (1) for sudden transfer of prestress. It shows that the combination of empirical coefficient (20.7) and f 1/2 ci in Equation (6) is similar to the combination of coefficient α t = 2.4 and f 2/3 ci in Equation (1) in case of sudden transfer of prestress. However, Equation (1) provides more consistent results for both gradual and sudden transfer of prestress.   (Figure 10a). The prediction of the transfer length of AFRP reinforcement with sanded and rough surfaces (Figure 10b) was also significantly underestimated according to Equation (3) with L t,teor /L t.exp = 0.3, STD = 0.12, COV = 36.6%. However, in this case, Equations (5) and (6) significantly overestimate the transfer length with L t,teor /L t.exp = 2.9, STD = 1.4, COV = 48.9% and L t,teor /L t.exp = 1.8, STD = 0.77, COV = 42.3%, respectively. A similar trend can be observed for all results from the database (without grouping according to surface conditions) (Figure 10c) calculated according to Equations (3), (5) and (6) with L t,teor /L t.exp = 0.3, STD = 0.14, COV = 42.6%; L t,teor /L t.exp = 1.8, STD = 1.41, COV = 78.6%; and L t,teor /L t.exp = 1.2, STD = 0.79, COV = 66%, respectively. The most accurate prediction of the results was obtained according to Equation (1) for AFRP reinforcement with sanded and rough surfaces (α t = 4.0) (Figure 10b) and all types of AFRP reinforcement (α t = 2.9) (Figure 10c) with L t,teor /L t.exp = 1.0, STD = 0.43, COV = 42.7% and L t,teor /L t.exp = 1.0, STD = 0.62, COV = 62.1%, respectively. However, the variation in the results is still very high for AFRP reinforcement.
Different values of coefficient α t (used in Equation (1)) were proposed by other authors (Table 9) and calculated in this article (see Section 4.3 and Table 11). The results in this article confirm the α t = 2.6 proposed by [36,37] for calculating the transfer length of the GFRP bars. Additionally, the α t = 1.92 calculated in this article is in close agreement with the proposals of other authors [10,12,31,32] α t = 1.90 and [39] α t = 1.95 for CFRP bars. The proposed α t = 5.0 for gradually released CFRP bars is slightly higher than α t = 4.8 [10,12,31,32] and significantly higher than α t = 2.12 [36,37]. The suggested α t = 2.4 is new, as there is no proposal of this coefficient for the sudden release of pretensioned CFCC reinforcement in the literature. This value is between the α t = 4.8 [10,12,31,32] and α t = 2.12 [36,37] suggested in the literature for the gradual release of pretensioned CFCC reinforcement. Furthermore, the literature review does not provide any proposal of α t for the AFRP reinforcement. Therefore, three possible values of α t were proposed in this article: α t = 2.9 for all different types of AFRP reinforcement, α t = 1.5 for smooth braided AFRP bars, and α t = 4.0 for sanded and rough AFRP bars. However, the high variation in the transfer length results of AFRP reinforcement suggests additional research and analysis for better understanding its behavior.
The comparison of the ratio between theoretical and experimental results of the transfer length (L t.teor /L t.exp ) of different types of FRP reinforcement for different values of α t is presented in Figure 11. Table 12 presents an additional statistical analysis. M-mean L t.teor /L t.exp , STD-standard deviation, COV-coefficient of variation, N total -the total amount of specimens, UV-underestimated values, OV-overestimated values, M+STD-upper bound, M-STD-lower bound.
The results for the GFRP bars with α t = 2.6 (Table 12 and Figure 11a) show a low COV = 12.6% and a high inbound value (between the upper and lower bounds) of 80%. Therefore, the proposed coefficient is suitable for GFRP bars.
The α t = 2.12 proposed by [39] for the gradual release of CFCC strands is not suitable due to a very high overestimation of the experimental transfer length with a mean L t.teor /L t.exp = 2.35, and all 41 specimens are on the safe side (Table 12 and Figure 11b). Furthermore, more CFCC strands with α t = 4.8 are on the safe side compared with α t = 5.0 (Table 12), and considering that all other statistical parameters are almost identical, α t = 4.8 may be a more appropriate choice for gradual transfer. For sudden CFCC strand release, the proposed α t = 2.4 gives better agreement with the experimental results for transfer length with theoretical values higher by 1% compared with results higher by 14% with α t = 2.12 (Table 12). Additionally, the linear trend line presented in Figure 11c shows that the theoretical results with α t = 2.4 are in closer agreement with the experimental transfer length. length results of AFRP reinforcement suggests additional research and analysis for better understanding its behavior. The comparison of the ratio between theoretical and experimental results of the transfer length (Lt.teor/Lt.exp) of different types of FRP reinforcement for different values of αt is presented in Figure 11. Table 12 presents an additional statistical analysis.   The results for the relationship between the theoretical and experimental transfer lengths are almost identical for CFRP bars, with α t ranging from 1.90 to 1.95 (Table 12 and Figure 11d). For α t = 1.95, the theoretical results on average are 2% lower than the experimental ones. Therefore, it is better to use α t = 1.9 for determining the transfer length of the CFRP bars.
The proposed α t = 2.9 for the AFRP bars provides 65% inbound values (Table 12 and Figure 11e). However, the 62% variation in the results is too large. Additionally, a separate α t = 1.5 for FRP bars with the smooth braided surface was proposed and provided a lower variation of the results (35%) and a similar inbound 61%. The α t = 4.0 proposed for the FRP reinforcement with sanded and rough surface shows a higher variation in the theoretical results (42%) compared with the results of the smooth braided AFRP bars (35%), but less than the results for all AFRP reinforcement with α t = 2.9. As the results of the sanded and rough AFRP bars provide about 49% inbound values, the amount of overestimated (OV = 20) and underestimated (UV = 19) values is similar.

Conclusions
A large database of the transfer lengths of different FRP reinforcements was collected, and the analysis of experimental results, description of theoretical models, and comparison of experimental and theoretical results led to the following conclusions and proposals: 1.
The analysis of the literature showed that the concrete compressive strength has a greater influence on the transfer lengths of CFRP and CFCC reinforcement, and the prestress level has a greater effect on the transfer lengths of AFRP and GFRP reinforcement. Additionally, most transfer lengths of FRP reinforcements (GFRP, CFCC, CFRP, and AFRP) are directly proportional to prestress level, reinforcement diameter, and modulus of elasticity of reinforcement and inversely proportional to concrete strength.

2.
The minimum concrete cover to avoid concrete splitting in the anchorage zone of the FRP reinforcement is 1.9·Ø for the CFRP bar and 2.8·Ø for the AFRP and GFRP bars. Additional research is needed for the CFCC strand, but the sufficient concrete cover for the CFCC strand is 4·Ø. Therefore, the concrete cover of 3.5·Ø and above does not have a significant influence on the transfer length of FRP reinforcement.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.