Experimental Study on Fatigue Behaviour of BFRP-Concrete Bond Interfaces under Bending Load

Basalt fiber reinforced polymer (BFRP) composites are increasingly being used to retrofit concrete structures by external bonding. For such strengthened members, the BFRP-concrete interface plays the crucial role of transferring stresses. This study aims to investigate the fatigue behaviour of the interface under bending load. A series of tests were conducted on BFRP-concrete bonded joint, including static, fatigue, and postfatigue loading. The fatigue failure modes, the development of deflection, the evolution of BFRP strains, and the propagation of interfacial cracks were analysed. In addition, the debonding-induced fatigue life of BFRPconcrete bonded joints was studied. Finally, a new model of fatigue life was proposed by defining the effective fatigue bond stress. The results showed that the fatigue experience has a significant effect on the BFRP strength especially near the root of concrete transverse crack and on the bond performance of the adhesive near the interface crack tip. There are two main fatigue failure modes: BFRP rupture and BFRP debonding. The fatigue damage development of the bond interface has three stages: rapid, stable, and unstable growth. The proposed model for the debonding-induced fatigue life is more conservative for the BFRP-concrete bonded joints under pure shear load than for those under bending load.


Introduction
The strengthening technique of applying fiber reinforced polymer (FRP) composites in civil structures has many advantages, such as high-strength/weight ratio, good corrosion resistant, ease of handling and application, and the elimination of the need for heavy equipment [1,2].In particular, carbon fiber reinforced polymer (CFRP) has been applied widely in the strengthening concrete structure because of its superior tensile performance [3][4][5].Compared with CFRP, basalt fiber reinforced polymer (BFRP) has better resistance to high temperatures and is also cheaper and greener [6].Consequently, strengthening concrete structures with BFRP is receiving increasing attentions [7].
For strengthening with externally bonded FRP, the FRPconcrete bond interface is not only the crucial element of transferring stresses, but also vulnerable for debonding which could result in the failure of the whole structure.This has led to numerous studies on the mechanical properties of FRP-concrete bond interface.However, the majority of these studies are focused on the interface under monotonic loading, with a limited number under dynamic loading [8]; very limited information is available for the fatigue performance of the FRP-concrete bond interface which is of dominate importance in cases such as highway and railroad bridges [9].A number of studies have been conducted on the fatigue behaviour of FRP-concrete interface using both the single shear pull-off test [9][10][11][12][13][14][15] and double shear pull-off tests [11,[16][17][18][19]. Table 1 presents the experimental research progress of FRP-concrete interface submitted to repeated or cyclic loading (so-called fatigue load).The test methods, the specimen number, the type of fiber, the fatigue load range, and other important parameters are summarized in Table 1.As shown in Table 1, the test can be divided into two types: one of which conducted fatigue test up to failure of the specimens and another conducted postfatigue monotonic test up to failure.The main aim of the former is to investigate the fatigue failure mechanism and fatigue life of FRP-concrete interface, while that of the latter is to study the effect of fatigue experience on the bond properties of FRP-concrete interface.The reported results indicated that either static debonding or fatigue debonding initiated at the root of the Carloni et al. [9] Carloni and Subramaniam [14] Diab et al. [19] Iwashita et al. [17] Ferrier et al.
[11] Ko and Sato [16] Bizindavyi et al. [10] Mazzotti and Savoia [12] Yun et al. [18] Dai et al. [15] Number of specimens Fatigue up to failure   diagonal crack near the loaded end under pull-off test [10,17].In addition, the fatigue experience under pull-off test showed no significant effect on the CFRP but significantly reduced the bond performance of FRP-concrete interface especially near the loaded end [10,11,14].However, it should be pointed out that the majority of the reported experimental studies used the single shear pull-off test or double shear pull-off test to investigate the fatigue behaviour of FRPconcrete interface, which cannot fully reflect the mechanical properties of concrete beams reinforced with FRP sheets [20].This is because the specimens were just subjected to pure shear loading in the single shear pull-off or double shear pulloff tests, but the actual concrete structures commonly are subjected to bending loads.Thus, the bending test can well simulate the effects of moment variation and shear force in concrete, which is closer to the actual engineering situation than the pure shear tests.
To investigate the fatigue life of FRP-concrete interface, Bizindavyi et al. [10] and Ferrier et al. [11] fitted the relationship of the bond stress and cyclic number to calculate the fatigue life of CFRP-concrete interface based on a series of pull-off tests.Similarly, the fatigue life prediction models (- curves) in Bizindavyi et al. [10] and Ferrier et al. [11] simply took the mean bond stress (the fatigue load/the whole bond area) as  but did not consider the effects of the remaining bond strength on the fatigue life.Actually, the fatigue life of the bond interfaces is influenced by the bond length of FRP.Diab et al. [19] pointed out that the stable propagation of debonding impossibly goes along the whole bond length; once the fatigue load overcomes the remaining bond strength of the interface, the life of the structure is due.In addition, Iwashita et al. [17] fitted the - curves to predict the fatigue life by taking the ratio between the fatigue upper load and the static debonding strength as .However, this kind of model did not consider the effects of FRP bonding area, resulting in the fact that predicted fatigue life would obviously be lower than that of the interface bonded with a longer FRP.Both Diab et al. [19] and Carloni and Subramaniam [14] introduced iteration methods to calculate the fatigue life by establishing the interface crack propagation rate with modified Paris models.However, due to the lack of the test database of interface crack propagation rate, the widespread use of these methods is hindered now [21].
In addition, the majority of the reported studies were focused on CFRP-concrete bond interfaces; hardly any work has been carried out on the fatigue behaviour of BFRPconcrete interfaces.Thus, this paper presents an experimental study of the fatigue behaviour of BFRP-concrete bond interface under bending load.A series of tests are conducted on BFRP-concrete bonded joint, including static, fatigue, and postfatigue loading.The fatigue failure modes, fatigue life, and propagation of interface cracks as well as the evolution of midspan deflection and BFRP strains with cycle numbers are analysed.After that, the debonding-induced fatigue life of BFRP-concrete bonded joints was investigated based on the test results and the reported test results under pure shear loading.Finally, a new simple model of fatigue life is proposed for BFRP-concrete bond interfaces.

Experimental Program
2.1.Specimen Design and Materials.In this study, the bending test proposed by Chen et al. [20] was adopted for ease of test.The specimen consisted of two 150 mm × 150 mm × 300 mm plain concrete prisms which separated by a 10 mm gap and linked by two 10 mm steel bars of 620 mm in length near the top face.The bottom faces of the prisms were ground to clear the soft mortar layer before one layer of 460 mm × 120 mm × 0.167 mm BFRP sheet was bonded.One side of the BFRP was designed as the test side, and the other side was enhanced with a BFRP U-strip to prevent debonding failure to occur in this nontest side (Figure 1).Six 3 mm foil strain gauges were attached on the BFRP sheet at the test side for each specimen (Figure 2).
The BFRP used in this study had a nominal thickness of 0.167 mm.Tensile flat coupon tests showed that its tensile strength, Young's modulus, and elongation were, respectively, 1397 MPa, 74 GPa, and 2.15%.For the epoxy resin used in this study, its manufacturer guaranteed tensile strength and Young's modulus were, respectively, 41.7 MPa and 2.52 GPa.The concrete mix included ordinary Portland cement, continuously graded gravel, medium grained sand, and water.The 28-day average compressive strength of three-cylinder specimens was 28.3 MPa.

Testing
Procedure.Three specimens were tested under monotonic loading to determine the reference static capacity.The rest of the 8 specimens were tested under a sinusoidal waveform fatigue load with a frequency of 10 Hz by using an electrohydraulic servo fatigue testing machine PWS-50 (Figure 3).The upper limit load varied from 55% to 65% of the static capacity.The lower limit load to upper limit load ratio (stress ratio) was 0.2.The strain readings were recorded by a high-speed dynamic strain logger TMR-211 at a frequency of 100 Hz.The BFRP-concrete interface crack length was measured by a microscope DJCK-2 with a measurement precision of 0.01 mm.The test conditions as well as the key test results are shown in Table 2.Note that each specimen is assigned a designation which consists of a letter (S or F for static or fatigue test), followed by a number representing the upper limit load (55, 60, or 65 percent of the static loading capacity), followed by a final number representing the specimen number (1, 2, or 3) within the same group.

Results and Discussions
The main experimental results are listed in Table 2, including the fatigue life, the static loading capacity, and failure modes.
It can be seen that the postfatigue static capacity of specimen F-55-1 is not lower than the capacity of all three specimens under static test, indicating that the static capacity is not necessarily affected by previous fatigue loading, if the interface has not completely debonded during fatigue tests and the length of the intact bond is greater than the effective bond length such as that determined by Chen and Teng's model [22].A similar observation is also reported for shear pull-off fatigue test (Bizindavyi et al. [10], Yun et al. [18], Diab et al. [19], Carloni et al. [9], and Carloni and Subramaniam [14]).
3.1.Failure Modes.For the three specimens under monotonic static loading, debonding failure occurred in the concrete layer.The failure mode was similar with that of BFRP/CFRPconcrete in single or double shear tests [8,23].The failure process included three phases: an inclined crack of approximate 45 ∘ appeared from the bottom near the midspan in concrete; with the increase of loading, interface crack between BFRP and concrete initiated at the root of the diagonal crack; the interfacial crack then propagated under complete debonding of the BFRP.The duration from the initiation of the interface crack to complete debonding was very short, resulting in an obvious brittle failure.The debonding failure appeared to take place within the concrete, with a thin layer of concrete attached to the debonded BFRP sheet as shown in Figure 4. Specimens under fatigue loading failed in two difference failure modes: debonding at BFRP/adhesive interface and BFRP rupture with the former being more common.For the fatigue debonding, the interfacial crack appeared to occur at the interface between the BFRP sheet and adhesive rather than in concrete.The debonded surface of BFRP was usually very smooth, as shown in Figure 5(a).It may be noted  that debonding failure within the adhesive layer has been reported for p-phenylene benzobisoxazole fiber and CFRP sheets in single or double shear fatigue tests [9,11,14,17].
In addition, some researchers [13,17] also often observed the occurrence of CFRP debonding in concrete in single or double shear fatigue tests, which is same as in static test.
Thus, it appears that fatigue damage has a significant effect on the bond performance of the adhesive near the crack tip, resulting in the variation of the debonding location: within the concrete or adhesive or at the adhesive-FRP interface.
It should be mentioned that the majority of FRP fatigue debonding tests has a common feature that the occurrence of FRP debonding initiated in concrete.The variation of the interface crack propagation location during fatigue loading might be greatly influenced by the physical and mechanical properties of the adhesive, such as its strength and glass transition temperature [11,14].The role of the adhesive in the fatigue debonding mechanism and its quantitative effect are complex and require further investigation.
For the specimens failed in BFRP fatigue rupture, they had similar fatigue damage process with those failed in debonding in the early stages.With the increase of cyclic numbers, an inclined crack appeared in the concrete near the midspan from the bottom, forming a concrete wedge.After that, an interface crack in the epoxy started to propagate from the root of the concrete crack.When the fatigue life of the specimen was reached, BFRP rupture took place at the location of the root of the inclined concrete crack (Figure 5(b)) where high stress concentration is present.This is an obvious brittle failure.Bizindavyi et al. [10] reported a similar fatigue failure mode for CFRP-concrete interface in the single shear test.FRP rupture usually occurs when the maximum applied cyclic stress reaches the strength of the FRP [10] at the loaded end where local bending is evident [22].
Specimen F-55-1 did not fail after 2,352,361 loading cycles.It was then tested under monotonic loading and failed due to BFRP rupture (Figure 6) which indicates that fatigue has a significant negative effect on the strength of BFRP.It has been reported in the literature that fatigue has a much deteriorative effect on BFRP than CFRP [24][25][26].

Evolution of Deflection.
Figure 7 shows the midspan deflection of three typical specimens under the upper limit load during fatigue test.The midspan deflection shows a three-stage evolution: rapid, stable, and unstable growth.The deflection of the specimens increased rapidly within the early cycles due to the formation of an inclined crack in the concrete.The stiffness of the specimens reduced rapidly during this phase which represents about 4% of the fatigue life.After this first phase, the evolution of the deflection entered into a stable stage in which the degradation of stiffness is much slower.This phase represents about 95% of the fatigue life.In the final stage, the deflection increased very rapidly due to the occurrence of either FRP rupture or debonding.This last stage consists of only a few load cycles, representing less than 1% of the fatigue life.Figure 8 shows typical load versus midspan deflection cycles for specimen F-65-1.Clearly, the hysteresis curves experience decreased slopes as the cycle numbers increases, indicating that the fatigue load caused the damage accumulation of the BFRPconcrete interface.

Evolution of BFRP Strain.
Figure 9 shows the BFRP strain distribution along the fiber direction for specimen S-2 (monotonic loading test) at different loading levels.The FRP strain may be used to monitor the initiation and growth of debonding [10,19,27].If there is no bending effect, the strain within the dedonded zone shall remain constant and equal the strain at the midspan.This phenomenon may be used to determine the crack (debonding) initiation strain from the BFRP strain distribution evolution.Figure 9 shows that the BFRP strain of specimen S-2 was about 0.9% (at approximately 20 kN) at the initiation of debonding.
Figure 10 shows typical strain distributions along the BFRP fiber direction under the upper limit load of 14.2 kN for specimen F-60-2 which had a fatigue life of   = 843351 cycles.Based on the above criterion, the initiation of debonding occurred at about 16 loading cycles with a BFRP strain of about 0.7%, indicating that the fatigue loading can accumulate damage around the crack tip leading to debonding at a lower FRP strain compared with monotonic static test.The cyclic debonding strain can be related to many factors, including the amplitudes of the fatigue loading, the number of loading cycles, and the material behaviour which determines how damage is accumulated and thus when debonding may occur [16,21].

Propagation of Interface Crack.
It is difficult to observe the whole debonding propagation process in fatigue test by using strain gauges because the bonded gauges on the FRP surface are easy to debond due to fatigue, as evident in Figure 10.In this study the test was paused regularly to directly measure the debonding crack length at the BFRPconcrete interface.To facilitate this, the concrete on both sides of the FRP was sanded and cleaned with acetone before painted with a thin layer of water-soluble white paint.
Figure 11 shows the evolution of the interface crack length with the cycle number for specimens F-65-2 and F-60-2.As can be seen from Figure 11(a), the propagation of the interface crack also has three phases: rapid, stable, and unstable growth, similar to that of the deflection shown in Figure 7.A similar observation is reported in Diab et al. [19].

Debonding-Induced Fatigue Life of FRP-Concrete Interface
As mentioned above, two typical fatigue failure modes occur in a FRP/epoxy-concrete bond system: FRP fracture and FRP debonding.In terms of FRP fatigue rupture, it usually occurs when the maximum applied cyclic stress reaches the fatigue strength of the FRP.For pure shear tests, FRP rupture usually occurs at the loaded end where local bending is evident [22].
For the bending test, FRP rupture usually occurs near the transverse crack where high stress concentration is present, as noted above.Many studies have been conducted to investigate the fatigue life of the FRP composites subjected to fatigue load [24,26,28,29], but only the fatigue life of FRP debonding from the FRP-concrete bond interfaces is discussed in this section.FRP debonding is the most common failure mode for the FRP-concrete interface under fatigue load [9-14, 16, 18, 19].It should be noted that when the bond length is larger than the effective bond length, the static bond capacity of FRP-concrete interface is constant [22], but the fatigue life increases with the FRP bond length because a longer bond length takes more cycles for the fatigue crack to propagate [10], which can be explained by the debonding failure process mentioned above.Consequently, the fatigue life of a FRPconcrete bond interface is significantly influenced by the concrete strength, the cross-sectional area of FRP, and the FRP bond length.The fatigue life increases with an increase of these parameters.In terms of the fatigue life prediction of FRP-concrete interface, the shortcomings of the reported models [10,11,14,17,19] as mentioned above necessitate the development of a new model for practical design, which should be simple to use, rationally based, and capable of capturing the fundamental features of the bond behaviour and predicting the debonding-induced fatigue life with good accuracy.This section aims at proposing a new model of - curve for calculating the fatigue life of FRP-concrete interfaces based on an extensive literature survey.The fatigue test database of FRP-concrete interface under pure shear tests from the existing literature has been collected, as shown in Table 3.It should be mentioned that all the collected specimens in Table 3 failed due to debonding, but the tests that were not sufficiently well documented for analysis and interpretation have been excluded here.In addition, the effect of frequency on the fatigue life has not been considered  because the frequency of the collected tests is between 1∼5 Hz.Many researchers who conducted fatigue test for concrete members with the low frequencies (1∼5 Hz) did not find significant effect on the fatigue life.In general, civil engineering structure is usually subject to a frequency of loading of approximately 5 Hz or less.Therefore, the negative effect due to the frequency of actual load condition on the FRP-concrete interfaces is slight.
As mentioned in the reported studies by Bizindavyi et al. [10], Ferrier et al. [11], and Iwashita et al. [17], the maximum applied load (the fatigue upper load) and the bond area are main factors for the fatigue life of the FRP-concrete interface.In addition, the minimum resistant bond area of FRP also significantly affects the fatigue life, which correspondingly resists the monotonic debonding failure under the fatigue upper load.Thus, an effective fatigue bond stress (  ) coupling fatigue upper load, bond area, and minimum resistant bond area is defined by where   and  are the width and length of bonded FRP;  max is the tensile force of the FRP at the loaded end for the pure shear test or that at the midspan for the bending test under the fatigue upper load, as shown in Figure 12;   is the critical bond length for resisting the monotonic debonding failure under  max , which denotes that when the remaining bond length after the fatigue crack propagation of the interface is less than or equal to l f , the fatigue failure will occur.Referring to the relationship of  and   reported in Chen and Teng [22],   can be solved by

Concrete block
At the loaded end for the pure shear test or At the midspan for the bending test where   denotes the static debonding strength of the interface, which is the tensile force of the FRP at the moment of failure, as shown in Figure 12;   is the effective bond length, which can be calculated by using the equation proposed in Chen and Teng [22].Note that   is in megapascals,  max and   are in kilonewtons, and   , ,   , and   are in millimeters.
As shown in ( 1) and ( 2),   not only describes the fatigue parameters, but also the crucial ones of the monotonic behaviour of the interface.Moreover, each parameter used has a sound physical meaning, condition that facilitates their experimental evaluation.Figure 13 shows the test points (  and fatigue life   ) for the specimens listed in Table 3.Based on the pure shear test database, a new - curve is fitted out as (3) described by   and semilogarithmic (  ), as shown in Figure 13.
= 2.462 − 0.135 ln (  ) . ( It can be seen from Figure 13   is suitable for predicting the debonding-induced fatigue life of FRP-concrete bond interfaces under pure shear loading, but not for those under bending loading.As observed in Figure 13, (3) overestimates the debonding-induced fatigue life of FRP-concrete bond interfaces under bending loading.This may be explained by that stress concentration is higher near the transverse crack under bending loading than at the loaded end in pull-off test, resulting in a shorter fatigue life in bending test.By taking into account the above considerations, a simple debonding-induced fatigue life model may be proposed based on (3) and the regression of test data in Table 3.
To compare this new model with the experimental data in Table 3, Figure 13 shows that it is more conservative for the FRP-concrete bonded joints under pure shear loading and agrees well with the test data for the specimens under high-cycle bending load.Traditionally, the stress of - model is defined as the fatigue strength of the specimen which corresponds to the fatigue life of 2,000,000 load cycles.Thus, it is available from (4) that the fatigue strength   of FRP-concrete bonded joints is 0.274 MPa, which can be regarded as the design value for resisting the debondinginduced fatigue failure under bending load.

Conclusions
An experimental study has been conducted to investigate the fatigue performance of the BFRP-concrete bond interface in a bending test, including monotonic, fatigue, and postfatigue loading.In addition, the debonding-induced fatigue life of FRP-concrete bonded joints has been analysed based on the present experimental results and the reported tests under pure shear loading.Finally, a new simple model of fatigue life has been developed.The following conclusions can be drawn from the experimental results: (1) There are two main fatigue failure modes: BFRP rupture and BFRP debonding.Local bending of BFRP has a significant negative effect on its tensile strength near the root of the inclined concrete crack which leads to BFRP rupture failure after high-cycle fatigue loading.
(2) Irrespective of the failure mode, the development of the fatigue interface crack has three stages under fatigue loading: rapid, stable, and unstable growth.
The fatigue debonding failure process of FRPconcrete joint interface can be described as follows: the local debonding of the interface near the loaded end firstly takes place because of fatigue damage, and then debonding propagates with the increase of fatigue time leading to the deterioration of large part of the bonded area, and finally the applied load overcomes the remaining bond strength of the interface resulting in a very brittle and instantaneous failure.
(3) As a result, the critical bond area significantly influences the fatigue life of the FRP-concrete interface, which correspondingly resists the static debonding failure under the fatigue upper load, as well as the actual bond area and the fatigue load.
(4) By defining the effective fatigue bond stress as the main parameter of fatigue strength, which couples the fatigue upper load, the actual, and critical bond area, a new model of - curve has been proposed based on the existing database for predicting the fatigue life of FRP-concrete bond interfaces.The results indicate that this model is more conservative for the FRPconcrete bonded joints under pure shear loading.
(5) The fatigue strength   of FRP-concrete bonded joints under bending load is 0.274 MPa, which corresponds to the debonding-induced fatigue life of 2,000,000 load cycles.

Figure 1 :
Figure 1: Details of the test specimens (mm).

Figure 12 :
Figure 12: Schematic diagram of the FRP force of bond interface.

Table 1 :
Research progress in the fatigue behaviour of FRP/epoxy-concrete interface.

Table 2 :
Test parameters and main results.

Table 3 :
FRP-concrete interface fatigue test data collected from literature.