Bond Behavior of CFRP–Concrete Bonded Joints with Additional GFRP Layer: Effect of Bonding Sequence

: Existing studies have shown that bonding a ± 45 ◦ biaxial GFRP under CFRP laminate can significantly improve the load-carrying capacity and ultimate deformation of CFRP–concrete bonded joints. In such a bonding configuration, the GFRP interlayer is wider than the CFRP laminate so that the interfacial stress can be redistributed to achieve a higher fracture work; however, the effect of the bonding sequence,—specifically, the position of the GFRP layer—on the bond behavior is not yet clear. In this study, considering the same CFRP and GFRP usage, three types of CFRP–concrete bonded joints with CFRP bonded under, above, and between GFRP layers were prepared and tested under single-shear loading. Digital image correlation (DIC) was used to measure the deformation of the bonded joints during the test. Afterward, the failure mode, load–displacement behavior, and principal strain distribution were analyzed. The experimental results show that the dominant failure mode is the combined cohesion failure mode within the concrete and GFRP delamination, which is not affected by the bonding sequence. Compared to conventional CFRP–concrete bonded joints, bonding the CFRP laminate above, under, and between the GFRP layers achieved a 157.6%, 175.0%, and 177.2% increase in load-carrying capacity, respectively. Accordingly, the ultimate deformation also recorded an 83.0%, 103.6%, and 86.3% increase. However, the bonding sequence showed a slight influence on the initial stiffness of the load–displacement curve with a maximum difference of 16.1%, taking the minimum as a reference, which could be attributed to the differences in the strength and stiffness between the CFRP–concrete and CFRP–GFRP–concrete interfaces.


Introduction
Externally bonded (EB) carbon-fiber-reinforced polymers (CFRP) have been increasingly used in strengthening reinforced concrete (RC) structures due to their advantages of being lightweight, high in strength, and with excellent durability over conventional strengthening materials such as steel [1][2][3][4][5].For EB-CFRP-strengthened RC members subjected to flexural loading, the stress is transferred between the CFRP and concrete through the adhesively bonded interface.As such, bond behavior plays an important role in assessing the strengthening performance.Existing studies have shown that two major failure modes, i.e., plate-end debonding and intermediate crack (IC)-induced debonding, often determine the load-carrying capacity.For plate-end debonding, additional anchors should be used to prevent premature debonding caused by the high interfacial peeling stress, whereas IC-induced debonding is caused by the high interfacial shear stress [6].Therefore, it is of great significance to understand the bond behavior of CFRP-concrete bonded joints.
While investigating the bond behavior of CFRP-concrete bonded joints, among the existing test methods-such as the single-shear pull-off test [7], double-shear pull-off test [8], and flexural beam test [9]-the single-shear pull-off test is often used [10][11][12] due to its simplicity and acceptable accuracy, wherein a CFRP laminate is bonded to a concrete block before applying shear loading.Results of such experiments have clearly shown that the dominant failure mode is the cohesion failure within the concrete.Therefore, for a given CFRP laminate, the fracture toughness of the concrete mainly determines the load-carrying capacity of the bonded joints [13,14].It is also found that when the bonding length is greater than the effective bonding length, the load-carrying capacity of the bonded joints cannot be further increased.As a result, the stress of the CFRP is merely 10-50% of the tensile strength of the CFRP laminate when the bonded joints reach the load-carrying capacity, resulting in a low material utilization rate of the CFRP [15,16].This low utilization rate significantly hinders the application of CFRP in the strengthening of RC structures.
To improve the material efficiency of CFRP, numerous anchoring devices including steel-FRP plate anchors [17][18][19], shooting nail anchors [20], FRP anchors [21,22], and U-shaped anchors [23] have been proposed to prevent premature debonding.Taking advantage of the interlocking effect, dowel action, and the enhanced friction achieved by the anchorage devices, existing results have shown that the above approach can significantly improve the bonding performance.However, due to the stress concentration at the anchoring position, installing anchors through drilling the CFRP laminate may result in a brittle fracture of the CFRP [3], while U-jacketing requires access to the side of the RC members.In addition, the installation of anchors usually involves drilling and alignment procedures, during which the internal reinforcement may be damaged.In addition to anchoring the CFRP, changing the geometric or mechanical properties of the bonding surface on the concrete substrate is also proposed.These techniques include, but are not limited to, plastering high-performance concrete [24], grooving or drilling holes [25][26][27][28], or using flexible adhesives [29]; however, these measures are either labor-intensive or show limited improvement on the bonding performance and so their widespread application remains challenging.
Most recently, the Author's group proposed using a GFRP interlayer to enhance the bonding performance [30,31].In this approach, a ±45 • biaxial GFRP sheet was bonded between the CFRP and concrete substrate.Compared to the existing approach of using the CFRP patch beneath the CFRP laminate as an anchor [32], the GFRP interlayer has a lower stiffness that can effectively transfer the interfacial stress and enhance the deformation capability.In addition, GFRP sheets are much cheaper than CFRP, and thus, a more economical solution can be achieved.Experimental results showed that the width of the GFRP interlayer and the width ratio between the GFRP interlayer and CFRP laminate can significantly affect the bond behavior in terms of the load-displacement curve, the loadcarrying capacity, and the ultimate deformation.Specifically, the force plateau diminished as the width ratio increased and the load-carrying capacity achieved, at most, a 203.0%increase compared to the conventional CFRP-concrete bonded joints.A simplified model was proposed to explain the bond behavior.It was shown that sequential activation of the GFRP-concrete interface in both the longitudinal and transverse directions determines the bond behavior.In addition, the existence of effective bonding length in such bonded joints depends on the width of the GFRP interlayer and vice versa.Detailed discussion can be found in [31].
Considering the high load-carrying capacity and ultimate deformation achieved by the CFRP-concrete bonded joints with a GFRP interlayer, bonding the GFRP fabric beneath the CFRP laminate in the intermediate region of the RC beam or slab is a promising application to improve the interfacial performance; for the plate-end debonding, it might be preferable to bond the GFRP on top of the CFRP laminate or sandwich the CFRP between two layers of GFRP sheet such that the peeling performance can be improved at the same time.However, compared to bonding the CFRP to the GFRP interlayer and then the concrete substrate, bonding the CFRP directly to the concrete substrate could affect the stress transfer path since part of the interfacial stress will be directly resisted by the concrete substrate before being redistributed by the GFRP interlayer.As a result, the fracture process and the activation of the GFRP interlayer in resisting external force could be affected.
Against this background, this paper conducts a series of single-shear pull-off tests on CFRP-to-concrete bonded joints with GFRP layers.Three bonding sequences are considered, i.e., bonding GFRP beneath the CFRP laminate, bonding GFRP on top of CFRP laminate, and bonding CFRP laminate between two GFRP sheets.The failure mode, load-carrying capacity, ultimate deformation, and principal strain development are presented and discussed to investigate the effects of bonding sequence on the stress transfer of CFRP-GFRP-concrete bonding joints.

Test Program
To investigate the effect of the bonding sequence, a total of 12 single-shear pull-off test specimens were prepared and tested.Among these specimens, three CFRP-to-concrete bonded joints without a GFRP layer were tested as reference specimens, and the other nine CFRP-GFRP-concrete bonded joints were prepared with three bonding sequences.Detailed descriptions of the specimen design, test set-up, and instrumentation are presented in the subsequent sections.

Materials
Table 1 summarizes the mechanical properties of the adhesive, CFRP, and GFRP used in this study.The structural epoxy adhesive supplied by Carbon was used to bond the CFRP and GFRP to the concrete substrate.The tensile strength was 60.9 MPa and the elastic modulus was 3.7 GPa, respectively, according to the manufacturer's data sheet.Unidirectional CFRP at 300 g/mm 2 , used in all specimens, was supplied by Shanghai Niugu Construction Technology Company(Shanghai, China).The nominal thickness of the CFRP laminate after impregnation was 0.56 mm and the tensile strength and elastic modulus were 617.1 MPa and 75.6 GPa, respectively.Two types of ±45 • biaxial GFRP, i.e., DB400-1270 (400 g/m 2 ) and DB800-1270 (800 g/m 2 ), were used in preparing the specimens.Both types of GFRP were provided by Changzhou Hualike New Material Company(Changzhou, China).The nominal thickness of the GFRP laminate after impregnation was 1.68 mm and 3.36 mm, respectively.In addition, the tensile strength of the GFRP along the fiber direction was 341.7 MPa and the elastic modulus was 22.9 GPa.The concrete blocks were cast with commercially available concrete.The cubic concrete compressive strength was tested to be 39.84 MPa according to GB/T 50081-2019 [26].

Specimen Design
Figure 1 presents the schematics of single-shear pull-off test specimens investigated in this study.The dimensions of the concrete blocks are 400 mm × 200 mm × 150 mm.For CFRP-concrete bonded joints with a GFRP layer (Figure 1a), the CFRP sheet (50 mm × 700 mm) was bonded to the concrete block directly through the wet lay-up process (4 layers), whereas for CFRP-GFRP-concrete bonded joints, three bonding sequences were employed (Figure 1b-d).In the first bonding sequence, named the G/C group-in which 'C' and 'G', respectively, represent the CFRP and GFRP sheets, and the order of appearance of 'G' and 'C' represents the sequence in which CFRP and GFRP are bonded to the concrete block surface-a GFRP interlayer (300 mm × 150 mm) consisting of 1 layer of DB800-1270 sheet was bonded to the concrete prior to the bonding of CFRP sheet.For the second bonding sequence (named the C/G group), the CFRP sheet was bonded to the concrete prior to the bonding of the DB800-1270 GFRP sheet.Finally, for the third bonding sequence (named the G/C/G group), the CFRP sheet was bonded between two layers of DB400-1270 GFRP sheets.For all three bonding sequences, the planar dimension of the GFRP layer was identical and the fibers were positioned at ±45 • to the longitudinal direction of the bonded joints.In addition, the free end and loaded end of the bonded joints were 50 mm away from the edges of the concrete block to avoid wedge failure.Table 2 summarizes the specimens prepared and tested in this study.Three repeated tests were conducted for each specimen group to reach a reliable test result.
and 'G', respectively, represent the CFRP and GFRP sheets, and the order of appearance of 'G' and 'C' represents the sequence in which CFRP and GFRP are bonded to the concrete block surface-a GFRP interlayer (300 mm × 150 mm) consisting of 1 layer of DB800-1270 sheet was bonded to the concrete prior to the bonding of CFRP sheet.For the second bonding sequence (named the C/G group), the CFRP sheet was bonded to the concrete prior to the bonding of the DB800-1270 GFRP sheet.Finally, for the third bonding sequence (named the G/C/G group), the CFRP sheet was bonded between two layers of DB400-1270 GFRP sheets.For all three bonding sequences, the planar dimension of the GFRP layer was identical and the fibers were positioned at ±45° to the longitudinal direction of the bonded joints.In addition, the free end and loaded end of the bonded joints were 50 mm away from the edges of the concrete block to avoid wedge failure.Table 2 summarizes the specimens prepared and tested in this study.Three repeated tests were conducted for each specimen group to reach a reliable test result.

Specimen Preparation and Instrumentation
Before bonding the CFRP and GFRP, the bonding area of the concrete surface was abraded with an electric hammer with a 32-millimeter-working-diameter point groove gouge (Figure 2a).Then, the dust and concrete debris were removed through a portable blower.To achieve clean bonding between the FRPs and the concrete substrate, the bonding areas were marked with masking tape (Figure 2b) to prevent excessive epoxy adhesive application outside these areas.After that, the GFRP and CFRP were successively bonded to the specified area according to the designated sequence (Figure 2c).During the wet lay-up process, a defoaming roller was used to squeeze out the trapped air bubbles.It should be noted that the screw roller should be moved in the fiber direction to avoid dispersing the fibers (Figure 2d).Once all the FRP layers were applied, the excessive epoxy adhesive around the bonding area was carefully removed before curing under ambient temperature (Figure 2e).All prepared specimens were cured for more than 4 weeks before testing.
abraded with an electric hammer with a 32-millimeter-working-diameter point groove gouge (Figure 2a).Then, the dust and concrete debris were removed through a portable blower.To achieve clean bonding between the FRPs and the concrete substrate, the bonding areas were marked with masking tape (Figure 2b) to prevent excessive epoxy adhesive application outside these areas.After that, the GFRP and CFRP were successively bonded to the specified area according to the designated sequence (Figure 2c).During the wet layup process, a defoaming roller was used to squeeze out the trapped air bubbles.It should be noted that the screw roller should be moved in the fiber direction to avoid dispersing the fibers (Figure 2d).Once all the FRP layers were applied, the excessive epoxy adhesive around the bonding area was carefully removed before curing under ambient temperature (Figure 2e).All prepared specimens were cured for more than 4 weeks before testing.Figure 3 illustrates the set-up and configuration of the single-shear pull-off test.During the test, the specimen was fixed in a steel frame (Figure 3b), which was mounted to a 300 kN universal testing machine (Lishi MDC-350, shown in Figure 3a).During the test, the external load was applied at a loading rate of 0.05mm/min.To measure the deformation of the bonded interface during the test, a 3D digital image correlation (3D-DIC) system from Correlated Solutions (VIC-3D) was used (Figure 3a).More details about the concept and technical knowledge of DIC for calculating the deformation can be found in [33].Figure 3 illustrates the set-up and configuration of the single-shear pull-off test.During the test, the specimen was fixed in a steel frame (Figure 3b), which was mounted to a 300 kN universal testing machine (Lishi MDC-350, shown in Figure 3a).During the test, the external load was applied at a loading rate of 0.05mm/min.To measure the deformation of the bonded interface during the test, a 3D digital image correlation (3D-DIC) system from Correlated Solutions (VIC-3D) was used (Figure 3a).More details about the concept and technical knowledge of DIC for calculating the deformation can be found in [33].

Failure Mode
The photographs of the failed specimens are presented in Figure 4.It can be se the typical failure mode of the CFRP-concrete bonded joints is cohesion failure with concrete (Figure 4a), which is consistent with existing research [7, 10,34].For CFR crete bonded joints with a GFRP layer, regardless of the bonding sequence, the ma ure mode is the combined cohesion failure within the concrete and the GFRP dela tion.Specifically, the former failure mode mainly occurred in regions close to the laminate, while the latter was observed in areas close to the edge of the bonding are ure 4b,c).Such behavior could be caused by the debonding propagation toward

Failure Mode
The photographs of the failed specimens are presented in Figure 4.It can be seen that the typical failure mode of the CFRP-concrete bonded joints is cohesion failure within the concrete (Figure 4a), which is consistent with existing research [7, 10,34].For CFRP-concrete bonded joints with a GFRP layer, regardless of the bonding sequence, the main failure mode is the combined cohesion failure within the concrete and the GFRP delamination.Specifically, the former failure mode mainly occurred in regions close to the CFRP laminate, while the latter was observed in areas close to the edge of the bonding area (Figure 4b,c).Such behavior could be caused by the debonding propagation toward the far end that the interfacial stress near the free end was no longer purely shear stress.This complex interfacial stress state led to GFRP rupture and delamination failures.Such a failure mode has also been observed in CFRP-concrete-and CFRP-steel-bonded joint tests [23,24].A detailed inspection of the failure surfaces for specimens C-3, G/G-1, C/G-1, and G/C/G-2 is presented in Figure 5.In the conventional CFRP-concrete bonded joints, there were several deep cracks perpendicular to the loading direction (Figure 5a).Regardless of the bonding sequence, a similar crack pattern was observed in the overlapping area of A detailed inspection of the failure surfaces for specimens C-3, G/G-1, C/G-1, and G/C/G-2 is presented in Figure 5.In the conventional CFRP-concrete bonded joints, there were several deep cracks perpendicular to the loading direction (Figure 5a).Regardless of the bonding sequence, a similar crack pattern was observed in the overlapping area of CFRP and GFRP in the CFRP-GFRP-concrete specimens (zone III in Figure 5b-d).It is worth noting that deep cracks were also observed in the zones on either side of zone III where only the GFRP was present (zone I and II in Figure 5b).These cracks were at an angle of approximately 45 • to the loading direction.However, in contrast to CFRP-concrete bonded joint specimens, a greater concrete chunk near the free end within the bonding area was pulled out in the CFRP-GFRP-concrete bonded joint specimens.angle of approximately 45° to the loading direction.However, in contrast to CFRP-concrete bonded joint specimens, a greater concrete chunk near the free end within the bonding area was pulled out in the CFRP-GFRP-concrete bonded joint specimens.

Load-Displacement Behavior
The load-displacement curves at the loaded end of all specimens are shown in Figure 6.The displacement at the loaded end was measured by the VIC-3D system [33] and the corresponding force values were exported from the universal testing machine.As can be observed in Figure 6, for CFRP-concrete bonded joints, the force initially increases with the displacement at the loaded end before reaching point D1.After that, a general force plateau occurred.Such behavior has been commonly observed in the existing literature and the force plateau is caused by the debonding initiation and propagation of the bonded joints [11,14].For CFRP-concrete bonded joints with GFRP layers, regardless of the bonding sequence, the force-displacement curves are significantly different to the CFRP-concrete bonded joints without GFRP layers.As indicated by the force and displacement values of A2, A3, and A4 in Figure 6b-d, the initial slope of the load-displacement curves at the loaded end of the bonded joints with the GFRP layer is generally greater than that of the CFRP-concrete bonded joints (calculated by C1 in Figure 6a).In addition, compared to the evident force plateau observed in the CFRP-concrete bonded joints, the force increases with the displacement before reaching the ultimate load-carrying capacity (F2, E3, and F4).

Load-Displacement Behavior
The load-displacement curves at the loaded end of all specimens are shown in Figure 6.The displacement at the loaded end was measured by the VIC-3D system [33] and the corresponding force values were exported from the universal testing machine.As can be observed in Figure 6, for CFRP-concrete bonded joints, the force initially increases with the displacement at the loaded end before reaching point D1.After that, a general force plateau occurred.Such behavior has been commonly observed in the existing literature and the force plateau is caused by the debonding initiation and propagation of the bonded joints [11,14].For CFRP-concrete bonded joints with GFRP layers, regardless of the bonding sequence, the force-displacement curves are significantly different to the CFRP-concrete bonded joints without GFRP layers.As indicated by the force and displacement values of A2, A3, and A4 in Figure 6b-d, the initial slope of the load-displacement curves at the loaded end of the bonded joints with the GFRP layer is generally greater than that of the CFRP-concrete bonded joints (calculated by C1 in Figure 6a).In addition, compared to the evident force plateau observed in the CFRP-concrete bonded joints, the force increases with the displacement before reaching the ultimate load-carrying capacity (F2, E3, and F4).After that, softening behavior in the load-displacement curve can be observed in specimens G/C-1, C/G-1 and 3, and G/C/G-2 and 3 as well as brittle failure in specimens G/C-2 and 3, C/G-2, and G/C/G-1.Moreover, compared to specimens in the G/C and G/C/G groups, the load-displacement curves of the C/G specimens show a more scattered behavior after point C.Such difference might be attributed to the non-homogeneity of the concrete substrate that may affect the fracture path and toughness when debonding propagates.More advanced non-destructive testing methods, such as 3D-CT [35], should be employed in the future to investigate the fracture process of the concrete during the test.Except for this difference, the load-displacement curves of the bonded joints with GFRP layers are similar to each other, regardless of the bonding sequence.
Buildings 2024, 14, x FOR PEER REVIEW 10 of 1 Except for this difference, the load-displacement curves of the bonded joints with GFR layers are similar to each other, regardless of the bonding sequence.Table 3 summarizes the load-carrying capacity and ultimate displacement at th loaded end of each specimen.Compared to the CFRP-concrete bonded joints, the bonde joints using GFRP layers exhibited a significantly higher load-carrying capacity and ult mate displacement.Specifically, for the load-carrying capacity, G/C specimens show 157.6% increase, C/G specimens record a 175.0%increase, and G/C/G specimens show 177.2% increase on average.Regarding the ultimate displacement at the loaded end, th C/G specimens exhibited the maximum increase at 103.6%, while the G/C and G/C/G spec imens recorded a similar increase of 83.0% and 86.3%, respectively.From the above ex perimental results, it can be briefly concluded that among the three bonding sequence considered in this study, bonding the CFRP between or under the GFRP layer provided higher load-carrying capacity and ultimate displacement compared to bonding CFRP o top of the GFRP layer.Nevertheless, all three bonding sequences can effectively improv the bonding performance.Table 3 summarizes the load-carrying capacity and ultimate displacement at the loaded end of each specimen.Compared to the CFRP-concrete bonded joints, the bonded joints using GFRP layers exhibited a significantly higher load-carrying capacity and ultimate displacement.Specifically, for the load-carrying capacity, G/C specimens show a 157.6% increase, C/G specimens record a 175.0%increase, and G/C/G specimens show a 177.2% increase on average.Regarding the ultimate displacement at the loaded end, the C/G specimens exhibited the maximum increase at 103.6%, while the G/C and G/C/G specimens recorded a similar increase of 83.0% and 86.3%, respectively.From the above experimental results, it can be briefly concluded that among the three bonding sequences considered in this study, bonding the CFRP between or under the GFRP layer provided a higher load-carrying capacity and ultimate displacement compared to bonding CFRP on top of the GFRP layer.Nevertheless, all three bonding sequences can effectively improve the bonding performance.

Strain Distributions on CFRP and GFRP Laminates
To further investigate the bond behavior of CFRP-concrete bonded joints with GFRP layers, the principal strain distribution on the GFRP and CFRP of the specimens at different displacement levels (corresponding to the marked points on the load-displacement curve in Figure 6) are presented in Figure 7.As the principal distribution of the different specimens from the same group shows a very similar pattern, only the principal strain distribution of one representative of the specimens from each group is presented.In the figures, the black arrows represent the direction of the principal strains.However, as the regions with purple colors in the principal strain distribution contour were not activated while loading, the strain measurement was interpreted as measurement noise and is not discussed in the subsequent sections.
From the principal strain distribution contour of specimen C-3, it can be observed that when the displacement at the loaded end is less than 0.107 mm (i.e., point D1 in Figure 6a), only a small region (with greenish color) close to the loaded end was activated to resist the external force.In addition, the principal strain direction in most areas on the CFRP was consistent with the loading direction, only a few areas at the edge of the CFRP exhibited an angle with the external load direction.When the displacement at the loaded end was 0.332 mm (point E1 in Figure 6a), more bonding regions were activated to resist the external force.In the meantime, the regions where the direction of the principal strain of CFRP was diagonal to the loading direction expanded (marked with ellipses in Figure 7a).Such behavior might be attributed to the non-uniform debonding within CFRP-concrete in the width direction.As the displacement at the loaded end increases (i.e., at 0.741 mm-point G1 in Figure 6a), more bonding areas have debonded as a general uniform strain can be observed in the rectangular region.The above experimental observation is consistent with the existing literature, thus validating the test set-up and data acquisition.
The principal strain development of specimens G/C-1, C/G-3, and G/C/G-3 are presented in Figure 7b-d.For all three types of specimens, when the slip at the loaded end is 0.045 mm, only a small area of the CFRP and GFRP laminate showed visible strain distribution, which indicated the activation of these bonding areas in resisting the external force.Such behavior also explains the slightly higher initial slope of the load-displacement curves at the loaded end compared to that of the C-3.However, G/C-1 exhibited the highest force among the three types of specimens.As all specimens were in the elastic stage (Figure 6), G/C-1 can provide the greatest initial stiffness.As the slip at the loaded end increased to 0.503 mm, the activated bonding area expanded significantly in both longitudinal and transverse directions.However, it is evident that the G/C-1 specimen expanded wider in the longitudinal direction and C/G-3 showed further expansion in the transverse direction, while specimen C/C/G-3 exhibited a more balanced activation in both directions.Meanwhile, specimen G/C/G-3 recorded a much higher external force at this point than the other two specimens.Such activation pattern of the bonding area continued when the bond slip increased to 0.945 mm when the principal strain distribution of specimens C/G-3 and G/C/G-3 reached 132 mm and 130 mm in the transverse direction, and 176 mm and 205 mm in the longitudinal direction, respectively.For the above two specimens, the activated bonding area reached the edge when the displacement at the loaded end reached 1.163 mm (indicated by the alignment of the principal strain and loading direction).As such, no more bonding area in the width direction can be activated to provide additional interfacial resistance to the external force.Such behavior explained the softening in the load-displacement curves in specimens C/G-1 and 3 and G/C/G-2 and 3.As for the brittle failure of the other specimens in both specimen groups, this might be attributed to the non-homogeneity and dynamic effect when the complete debonding is about to happen; however, more experimental and numerical research should be conducted in the future.As a comparison, the principal strain distribution of specimen G/C-1 reached 138 mm and 217 mm in the transverse and longitudinal directions, respectively, and it is evident that the activated bonding area reached the edge when the displacement at the loaded end reached 1.418 mm.
It should be noted that regardless of the bonding sequence, the direction of principal strain on the GFRP layer is approximately 45 • to the loading direction, which aligns with the fiber direction of the GFRP layer.Such behavior can be attributed to the fact that the GFRP layer is subjected to in-plane shear loading.As a result, the principal stress within the GFRP layer is approximately 45 • to the loading direction.Such principal stress will further cause cracks within the concrete substrate in zones I and II, as observed in the failed specimens (Figure 5).Additionally, the activated bonding area is significantly greater in the bonded joints with GFRP layers compared to their CFRP-concrete bonded joints counterparts.Such phenomena can explain the remarkable increase in the load-carrying capacity in the bonded joints with GFRP layers.
To further investigate the effect of the GFRP layers on the bond behavior of the CFRPconcrete bonded joints, the strain distribution curves on CFRP and GFRP at different loading levels are presented in Figure 8.For specimen C-3, the strain distribution on the CFRP sheet is presented, while for specimen G/C-1, the principal strain distribution along the CFRP sheet and five lines along the fiber direction of the GFRP layer are displayed.As the CFRP sheet is covered by the GFRP layer in specimens C/G-3 and G/C/G-3, only the strain in the loading direction along the midline of the bonded joints (i.e., axis-x) as well as the principal strain along the fiber direction of GFRP on one side of the CFRP sheet are presented.However, considering the deformation continuity of the bonded CFRP and GFRP, the strain on the centerline of the GFRP can be taken as that of the CFRP laminate.
The longitudinal strain distribution curve of the C-3 specimen is shown in Figure 8a-1.A clear strain plateau can be observed when the displacement at the loaded end reaches 0.332 mm.This indicates that the CFRP-concrete bonded joints started debonding.As the external force increases, the increase in the strain plateau indicates the debonding propagation.In comparison, the evident strain gradient of the CFRP in the loading direction of specimen G/C-1 throughout the test can be observed.The maximum strain of CFRP reached as high as 15,000 µε.However, for specimens C/G-3 and G/C/G-3, a similar strain plateau with specimen C-3 can be observed when the displacement at the loaded end reached 0.945 mm and 0.503 mm, respectively.Such strain plateaus meant the interfacial stress transfer was minimal in this region.Additionally, the maximum strain that both specimens can reach are approximately 12,000 µε and 10,000 µε, which were less than that in G/C-1.Such differences indicate a distinct interfacial stress transfer in the G/C-1 specimen.
Regarding the principal strain distribution along the fiber direction of the GFRP layer, a similar strain distribution along lines L1 to L5 can be observed in the bonded joints with GFRP layers.For the principal strain along a specific line (i.e., L1 to L5), a clear increase in the strain values close to the loaded side can be observed.In addition, the activated bonding area with the significant strain propagated toward the free side as the external load increased.When the displacement at the loaded end increased to 0.945 mm, evident strain plateaus occurred along L1 of specimens G/C-1 and G/C/C-3.Such plateaus can be also observed along L3 of specimen G/C-1 and L4 of all three discussed specimens as the displacement further increased to 1.418 mm.From the above observations, it can be seen that the GFRP layer is sequentially activated in both transverse and longitudinal directions to resist the external load.From the principal strain distribution contour of specimen C-3, it can be observed that when the displacement at the loaded end is less than 0.107 mm (i.e., point D1 in Figure 6a), only a small region (with greenish color) close to the loaded end was activated to resist the external force.In addition, the principal strain direction in most areas on the CFRP

Discussion
The above results clearly indicate that inclusion of a GFRP layer can significantly affect the bond behavior of CFRP-concrete bonded joints regarding the load-carrying capacity, ultimate deformation, and strain distribution.As discussed in Zhou et al. [31], when the GFRP layer is positioned under the CFRP laminate as an interlayer (i.e., the G/C specimen group), the axial stress within the CFRP is firstly transferred to the GFRP inter-

Discussion
The above results clearly indicate that inclusion of a GFRP layer can significantly affect the bond behavior of CFRP-concrete bonded joints regarding the load-carrying capacity, ultimate deformation, and strain distribution.As discussed in Zhou et al. [31], when the GFRP layer is positioned under the CFRP laminate as an interlayer (i.e., the G/C specimen group), the axial stress within the CFRP is firstly transferred to the GFRP interlayer and then distributed to the concrete substrate.The sequential activation of the bonding area in the transverse and longitudinal directions can provide a greater active bonding area when subjected to external force.As the load-carrying capacity of the bonded joints is mainly determined by the fracture work, a greater bonding area leads to a higher total fracture work, thus, a significant increase in the load-carrying capacity can be observed.Detailed discussions can be found in Zhou et al. [31].
Regardless of the bonding sequence, the failure modes of the CFRP-concrete bonded joints with a GFRP layer include the combined cohesion failure within the concrete and the GFRP delamination at the edge of the bonding area.Except for the approximately 45 • cracking to the load direction, the above failure mode is consistent with the conventional CFRP-concrete bonded joints.Such consistency indicates that with proper concrete surface treatment, the inclusion of a GFRP layer would not introduce a weak interface to the conventional CFRP-concrete bonded joints.Additionally, as the interfacial stress is redistributed by the GFRP layer in the principal direction (i.e., approximately 45 • to the loading direction), a similar crack pattern in the concrete substrate can be expected.
When bonding the CFRP laminate under or between the GFRP laminate, though the bond behavior of the GFRP-concrete interface on both sides of the CFRP laminate is identical as they possess the same thickness and fiber volume ratio, the slight change in the stress transfer path will affect the initial elastic stiffness of the load-displacement curve (maximum difference of 16.1% taking the minimum as a reference).Specifically for the C/G specimens, the axial stress within the CFRP laminate is transferred to the GFRP laminate and concrete substrate simultaneously, then, the GFRP laminate on both sides of the CFRP will distribute the stress to the concrete substrate.However, for the G/C specimen, the axial stress within the CFRP laminate is transferred to the GFRP interlayer first and then redistributed to the concrete substrate.Therefore, compared to the C/G specimens, more concrete is activated to resist the external force in the G/C specimen, which leads to a higher initial slope in the load-displacement curve.Meanwhile, for the G/C/G specimens, the axial stress is transferred to the single layer under and above the CFRP laminate; however, since only one layer of GFRP laminate was bonded between the CFRP and concrete, the overall stiffness will be greater than that of the C/G specimen but less than the G/C specimen.
Except for affecting the initial stiffness, the bonding sequence also exhibited influence on the principal strain distribution and propagation.Compared to the G/C specimens, the CFRP laminate was bonded to the concrete directly in the C/G specimens.As the CFRP-concrete interface tended to debond earlier than the CFRP-GFRP (two layers)concrete interface due to the deformation and stress distribution of the GFRP interlayer, the displacement of the CRRP laminate will be greater for a given load.Considering the displacement continuity between the CFRP and bonded GFRP, more of the GFRP-concrete interface will be activated to resist the force.As such, the principal strain would propagate more in the transverse direction and failure in the width direction can be first observed.In comparison to the G/C specimens, the principal strain tends to develop in the longitudinal direction, while for the G/C/G specimens, the stiffness of the CFRP-GFRP (one layer)concrete interface lies between the C/G and G/C specimens in that the principal strain distribution shows development in both transverse and longitudinal directions when subjected to shear loading.
Nevertheless, the bonding sequence showed less of an effect on the overall activated bonding area when the maximum load was attained, thus, the load-carrying capacity was similar for different specimens.As a result, the bonding scheme can be customized to achieve the optimal strengthening performance in practice.For example, when the GFRP layer is employed for strengthening a reinforced concrete aqueduct structure with CFRP laminate, the GFRP can be bonded on top of the CFRP at the inner side of the aqueduct so that a smooth surface can be achieved to reduce the water blocking.The water permeability can be also improved in such as bonding configuration.For the RC beam or slab strengthening, the GFRP layer can also be bonded on top of the CFRP laminate at the plate end to improve the shear and peeling performance simultaneously.However, if a large bonding area is expected, e.g., bonding the GFRP layer in the intermediate region of the RC beam or slab, a bonding GFRP layer beneath the CFRP sheet is preferred to avoid the possible gaps at the edge of the CFRP laminate, especially when a thick CFRP laminate is used.Additionally, the C-G-C bonding configuration could be used in producing a commercial hybrid carbon and glass fiber fabric so that only one-time bonding is required in practice.However, more research should be conducted to investigate the water-resistance performance, Mode-I bond behavior, and production process for the above-mentioned applications of different bonding schemes.

Conclusions
This paper investigates the influence of the bonding sequence on the behavior of CFRP-concrete bonded joints with a GFRP layer.Three types of specimens, i.e., G/C, C/G, and G/C/G, were prepared and tested through a series of single-shear pull-off tests.The failure modes, load-carrying capacity, ultimate displacement at the loaded end, and strain distribution of the specimens are presented and discussed.The following conclusions can be drawn: 1.
The bonding sequence of the CFRP and GFRP showed negligible influence on the failure mode of tested specimens.The dominant failure mode was combined cohesion failure within the concrete and GFRP delamination.As such, the proper bonding position of the GFRP layer can be selected in different scenarios so that the benefits provided by the improvements in the Mode-I bond behavior and water permeability by the GFRP layer can be fully utilized; 2.
Compared to the CFRP-concrete bonded joints, the inclusion of a GFRP layer can significantly improve the bonding performance.A minimum of 157.6% increase in the load-carrying capacity and 83.0% in the ultimate deformation can be expected in G/C specimens.However, the effect of the bonding sequence on the bonding performance is insignificant.Bonding the CFRP laminate between two GFRP layers recorded the highest increase in load-carrying capacity (177.2%); the results for the other two types of specimens were 157.6% and 175.0%, respectively.The C/G specimen exhibited the maximum increase in ultimate displacement (103.6%),although 83.0% and 86.3% increases were achieved by the G/C and G/C/G specimens; 3.
The bonding sequence of the CFRP and GFRP sheets showed an influence on the initial stiffness of the load-displacement curves and the principal strain distribution propagation direction.Such influences could be attributed to the bonding stiffness and strength of the CFRP-concrete and CFRP-GFRP-concrete interfaces.However, the overall activated bonding area is insignificantly influenced by the bonding sequence.

Figure 1 .
Figure 1.Schematics of (a) conventional CFRP-concrete bonded joints; (b) CFRP-GFRP bonded joints where a biaxial GFRP was placed in the interlayer between CFRP and concrete; (c) CFRP-GFRP bonded joints where CFRP was placed between GFRP and concrete; (d) CFRP-GFRP bonded joints where CFRP was placed between GFRP sheets.

Figure 1 .
Figure 1.Schematics of (a) conventional CFRP-concrete bonded joints; (b) CFRP-GFRP bonded joints where a biaxial GFRP was placed in the interlayer between CFRP and concrete; (c) CFRP-GFRP bonded joints where CFRP was placed between GFRP and concrete; (d) CFRP-GFRP bonded joints where CFRP was placed between GFRP sheets.

Figure 2 .
Figure 2. Procedures of specimens' preparation: (a) abrading the bonding area of concrete; (b) pasting masking tape around the bonding area; (c) applying the biaxial glass fabric saturated with epoxy adhesive; (d) squeezing out air bubbles; (e) curing specimen under ambient temperature; (f) a speckled specimen.

Figure 2 .
Figure 2. Procedures of specimens' preparation: (a) abrading the bonding area of concrete; (b) pasting masking tape around the bonding area; (c) applying the biaxial glass fabric saturated with epoxy adhesive; (d) squeezing out air bubbles; (e) curing specimen under ambient temperature; (f) a speckled specimen.
Buildings 2024, 14, x FOR PEER REVIEW

Figure 3 .
Figure 3. Test instrumentation: (a) specimens were fixed within the steel frame that was inst the 300 kN universal testing machine and the 3D-DIC cameras were placed in front of the spe to measure the deformation; (b) the specific steel frame fixture used to fix specimens.

Figure 3 .
Figure 3. Test instrumentation: (a) specimens were fixed within the steel frame that was installed in the 300 kN universal testing machine and the 3D-DIC cameras were placed in front of the specimens to measure the deformation; (b) the specific steel frame fixture used to fix specimens.
After that, softening behavior in the load-displacement curve can be observed in specimens G/C-1, C/G-1 and 3, and G/C/G-2 and 3 as well as brittle failure in specimens G/C-2 and 3, C/G-2, and G/C/G-1.Moreover, compared to specimens in the G/C and G/C/G groups, the load-displacement curves of the C/G specimens show a more scattered behavior after point C.Such difference might be attributed to the non-homogeneity of the concrete substrate that may affect the fracture path and toughness when debonding prop-

Table 2 .
Parameters of specimens.

Table 2 .
Parameters of specimens.

Table 3 .
Load-carrying capacity and ultimate displacement of tested bonded joints.

Table 3 .
Load-carrying capacity and ultimate displacement of tested bonded joints.