Experimental and Analytical Investigations on Glass-FRP Shear Transfer Reinforcement for Composite Concrete Construction

Abstract

In accelerated bridge construction, precast concrete girders are connected to cast-in-place concrete slab using shear transfer reinforcement across the interface plane to ensure the composite action. The steel transverse reinforcement is prone to severe corrosion due to the extensive use of de-icing salts and severe environmental conditions. As glass fiber-reinforced polymer (GFRP) reinforcement has shown to be an effective alternative to conventional steel rebars as flexural and shear reinforcement, the present research work is exploring the performance of GFRP reinforcements as shear transfer reinforcement between precast and cast-in-place concretes. Experimental testing was carried out on forty large-scale push-off specimens. Each specimen consists of two L-shaped concrete blocks cast at different times, cold joints, where GFRP reinforcement was used as shear friction reinforcement across the interface with no special treatment applied to the concrete surface at the interface. The investigated parameters included the GFRP reinforcement shape (stirrups and headed bars), reinforcement ratio, axial stiffness, and the concrete compressive strength. The relative slip, reinforcement strain, ultimate strength, and failure modes were reported. The test results showed the effectiveness and competitive shear transfer performance of GFRP compared to steel rebars. A shear friction model for predicting the shear capacity of as-cast, cold concrete joints reinforced by GFRP reinforcement is introduced.

1. Introduction

Composite construction is an economical construction method used in bridges. It combines precast girders and cast-in-place slabs while maintaining the continuity and the efficiency of a monolithic concrete. It allows the utilization of the overall properties of the built-up section, which permits the use of lighter and shallower beams. Composite construction method has proven to save time and cost as well as minimize traffic disruption. Consequently, it became a broadly used practice for accelerated bridge construction.

Since the girder and the slab are cast at different times (cold-joint condition), transferring the longitudinal shear stresses along the interface becomes of the most interest to ensure full composite action (Figure 1). The overall strength and stiffness of the built-up girder-slab section can only be utilized if the composite action is achieved at their connection. Up to date, conventional steel stirrups extended from the top of the prefabricated girders are being used across the shear interface between the girders and slabs to achieve the desired shear transfer resistance. However, deterioration of the deck slab caused by environmental and traffic conditions results in an extensive corrosion of the steel reinforcement, especially when de-icing chemicals are used on bridge decks. Corrosion deterioration would lead to a gradual loss of monolithic action and reduce the strength of composite concrete beams. To remedy this serious problem, epoxy-coated steel reinforcement (ECR) was proposed as a substitute for conventional steel at the concrete girder-slab joints. Yet the use of epoxy-coated steel reinforcement faces many challenges. In addition to its high cost relative to conventional steel, it was found that as the epoxy-coating adhesion deteriorates, the damage to the coating increases, resulting in exposure of the steel bars. Therefore, ECR was found to be ineffective in improving composite action during the service life of girder-slab systems [1].

Glass fiber-reinforced polymer (GFRP) bars and stirrups have proven to be an effective alternative to steel as flexural and shear reinforcement in concrete beams, columns, and slabs. The non-corrodible nature, high tensile strength, superior bond, and lightweight characteristics of GFRP have encouraged its application in reinforced concrete structures [2]. In addition, Al Omar and Abdelhadi (2024) reported that the long-term benefits of using GFRP—such as reduced material quantities and lower environmental impact—can yield substantial life-cycle cost savings, including approximately 17% lower CO₂-equivalent emissions compared to steel and mass reductions of 77–85% in certain configurations [3]. Similarly, Younis and Ebead (2020) found that GFRP reinforcement achieved about 40–50% lower net present cost than conventional black steel reinforcement, further highlighting its economic advantage over traditional materials [4].

The design and construction requirements of concrete members reinforced with fiber reinforced polymer (FRP) reinforcement are available in several design codes and guidelines such as the Canadian standard for the design and construction of building structures with fiber-reinforced polymers, CAN/CSA S806 [5] and the American Building Code Requirements for Structural Concrete Reinforced with Glass Fiber-Reinforced Polymer (GFRP) Bars, ACI 440.11 [6] The Canadian highway bridge design code, CAN/CSA S6 [7], was the first design code to incorporate design model for cold-joint concrete interfaces reinforced with FRP reinforcement based on the work of Alkatan, 2016 [8]. The objective of this research is to evaluate the performance of GFRP as shear transfer reinforcement across concrete-to-concrete cold-joint interfaces and study the associated shear transfer mechanisms. This research program provides more insight into the effect of different reinforcement types, concrete strength, and reinforcement configuration on the shear transfer performance of GFRP-reinforced concrete cold-joints; it also presents an analytical model for predicting the shear transfer strength.

Research Significance

The shear friction model is the simplest and the most popular framework that describes the shear transfer mechanism along precast and cast-in-place concrete interfaces. In its original form, the shear friction theory assumes that the shear stresses, parallel to the concrete joint interface, are transmitted by friction only [9]. If a crack is expected along the shear plane, the roughness of the crack faces would force the interconnected concrete elements to separate when a slip occurs (Figure 2). This tendency for separation, referred to as dilatancy, would place the steel across the interface in tension, which would, in turn, generate clamping forces on the crack faces. Owing to the roughness and irregularity of concrete cracking at the interfaces, the clamping force provides a substantial shear frictional resistance. Hypothetically, if the amount of transverse steel reinforcement is sufficient and well, full anchorage is provided on both sides of the shear plane; the ultimate state of frictional shear resistance is reached when the steel rebars yield. Mattock and Hawkins (1972) proposed a modification to the original shear friction theory, where cohesion stress, $c$, and external clamping stresses,${\sigma }_n$, were combined [10]. Walraven et al. (1987) considered that the load transfer along a cracked interface face is due to the local roughness (cohesion) and general roughness (friction) [11]. Zilch and Reinecke (2000) concluded that the shear strength of concrete-to-concrete interfaces is a combination of the following three load-carrying mechanisms: adhesion bond, shear friction, and dowel action of the transverse shear reinforcement [12]. When the maximum shear capacity of the concrete surface is reached, cracking of the interface occurs, and the shear stress is then transferred by friction only if sufficient clamping stresses are provided. The dowel action contribution of the steel was found to be insignificant until a relatively high shear slip occurs [13,14].

Most of the research on steel-reinforced concrete interfaces concluded that the shear transfer mechanism between different concrete layers is directly proportional to the reinforcement ratio and yield strength; concrete compressive strength and aggregates interlock play a major role in the shear transfer mechanism. It was also postulated that the ultimate shear stress is achieved at the yielding of the steel across the shear plane [15,16,17,18]. Harris et al. 2012 suggested that the clamping stresses at the ultimate load should be a function of the steel’s modulus of elasticity of steel rather than its yield strength [19].

Table 1. Shear transfer models.

Research	Expression (SI Units)	Limits/Notes
Birkeland and Birkeland (1966) [9]	$v_u={\rho }_v{f}_y\mu$	$\mu = 1.7$ for monolithic concrete. $\mu = 1.4$ for artificially roughened joints. $\mu = 0.8$–1 for ordinary construction joints. $f_c^{\prime } \ge 27.6 \mathrm{M}\mathrm{P}\mathrm{a} \mathrm{a}\mathrm{n}\mathrm{d} v_u \le 5.52 \mathrm{M}\mathrm{P}\mathrm{a}.$
Mattock (2001) [18]	For concrete placed against hardened concrete not intentionally roughened: $v_u = 0.6\lambda {\rho }_v{f}_y$	$v_u \le (0.2f_c^{\prime } \mathrm{a}\mathrm{n}\mathrm{d} 5.52 \mathrm{M}\mathrm{P}\mathrm{a})$
Khan and Mitchell (2002) [17]	$v_u=0.05f_c^{\prime }+1.4{\rho }_v{f}_y$	$v_u \le 0.2f_c^{\prime }$
Harries et al. (2012) [19]	For cold-joint interfaces: $v_u = 0.04f_c^{\prime }+0.002 E_s{\rho }_v$	$v_u \le 0.2f_c^{\prime }$
CSA A23.3-19 [21]	For concrete placed against hardened concrete not intentionally roughened: $v_u = \lambda \left(c + \mu \sigma \right) + {\rho }_v{f}_y\cos {\alpha }_f$ $\sigma = {\rho }_v{f}_y\mathit{sin}{\alpha }_f + N/A_{cv}$	$c = 0.25 \mathrm{M}\mathrm{P}\mathrm{a};$ $\mu = 0.60$ $v_u \le 0.25 f_c^{\prime }$ ${ \rho }_{v } \ge 0.06\sqrt{f_c^{\prime }}/f_y$ MPa
AC1 318-25 [22]	For concrete placed against hardened concrete not intentionally roughened: $v_u = {\rho }_v{f}_y \left(\mu \sin {\alpha }_f + \cos {\alpha }_f\right)$	$\mu =0.60\lambda$ $v_{u }\le (0.2f_c^{\prime } \mathrm{a}\mathrm{n}\mathrm{d} 5.52 \mathrm{M}\mathrm{P}\mathrm{a})$ ${\rho }_{v }\ge (0.06\sqrt{f_c^{\prime }}/f_y \mathrm{a}\mathrm{n}\mathrm{d} 0.34/f_y \mathrm{M}\mathrm{P}\mathrm{a})$

gives some of the developments of shear friction models that were proposed to evaluate the shear transfer strength of concrete joints reinforced with steel. Most shear transfer models are linear and were empirically derived; those models are mostly calibrated to test results of the push-off specimens. Push-off specimens, which exhibited strong similarity with the interfaces in composite concrete beams are extensively used [20] The shear friction theory was adopted in many reinforced concrete design codes, including the Canadian standard of the design of the concrete structures, CAN/CSA A23.3 [21], the Canadian highway bridge design code, CAN/CSA S6 [7], the American building code requirements for structural concrete, ACI-318 [22], and AASHTO LFRD Bridge Design Specifications [23].

Recent experimental programs concluded that GFRP stirrups used as shear friction reinforcement provide a meaningful contribution to interface shear capacity compared with unreinforced joints—particularly when sufficient anchorage is provided. It is also concluded that existing steel-based shear friction provisions underpredict or mispredict FRP behavior as FRP has lower stiffness, different bond/slip behavior, and different failure modes; therefore, there is still a need for more experimental and analytical investigations to fully understand the behavior and be able to accurately calculate the strength of concrete interfaces reinforced with FRP for the friction shear transfer. The Canadian highway bridge design code, CAN/CSA S6 [7] was the first design code to incorporate design model for cold-joint concrete interfaces reinforced with FRP reinforcement based on the work of [7,8,24,25,26,27]. However, there is still need for more experimental data on the performance of GFRP as shear friction reinforcement.

2. Experimental Program

2.1. Test Specimens

The experimental testing was carried out on forty large-scale push-off concrete specimens. This method of testing shear friction is preferred as it exhibits shear-slip behavior comparable to the slab-girder interfaces of composite concrete beams [19,20]. Figure 3a shows a schematic drawing of a push-off specimen used in this study. Each specimen consists of two L-shaped concrete blocks which were cast at different times to form cold-joint conditions at the interface. The width of the interfacial shear plane between the connected blocks is fixed at 250 mm, where different values of 300, 400, and 500 mm were used for the length of the interface to change the reinforcement ratio ${\rho }_v$. The depth of each specimen was fixed at 250 mm to ensure enough anchorage length of the reinforcement. The flange of the L-shape is 250 mm wide, 250 mm thick, and 500 mm long. The main purpose of the flange part is to allow the application of the load concentrically along the shear plane. A 25 mm gap was left between the two parts in the longitudinal direction to allow for a free relative slip between the two blocks of the specimen. Accordingly, the total length of the specimen was 1040 mm. Longitudinal and transverse steel bars and stirrups were used to reinforce both the web and the flange of each L-block to prevent any premature failure of the individual block. Reinforcement in the form of a steel cage was placed in the formwork of the first block. The shear transfer reinforcement bars were placed and secured in their position with the steel reinforcement before casting the first block (Figure 3b,c). The top surface of the first block, the interface between the two blocks, was left as cast without any special treatment (Figure 3d). The second L-shape block was then cast in a similar manner after moist curing the first block for three days, as shown in Figure 3e. The three-day delay period was specifically chosen to guarantee the formation of a complete “cold joint”, thereby ensuring no co-hydration occurred across the concrete interface [28,29]. This methodology was implemented to accurately simulate common field applications in bridge deck construction, such as those where an extended time elapses between the casting of structural girders and the subsequent placement of the deck slab components [28].

A total of forty push-off specimens were cast and tested. The test parameters included the shape, axial stiffness, and reinforcement ratio of the GFRP friction transverse reinforcement and concrete strength. Two different shapes, regular stirrups and headed bars, of the sand-coated GFRP reinforcement were used, as shown in Figure 4a. The GFRP stirrups have embedment length of 220 mm in each block of the test specimens. The GFRP headed bars have about 240 mm embedment length at the smooth end in one block and 200 mm, including the head, at the headed end in the other block. The provided embedment lengths ensure the development of the full tensile strength of the GFRP stirrups and headed bars. Figure 4b shows the general layout of test specimens reinforced with GFRP stirrups and headed bars. The axial stiffness parameter of the GFRP reinforcements, $E_f{\rho }_v$, depends on the reinforcement ratio across the interface, ${\rho }_v$, and the modulus of elasticity of the GFRP reinforcement, $E_f$. Different values of the GFRP stirrups and headed bars reinforcement area across the interface as well as different interface surface area resulted in GFRP reinforcement ratios, ${\rho }_v$, in the range from 0 to 1.35%. It is to be noted that the headed reinforcement bars have a modulus of elasticity of 60 GPa (grade III) whereas the stirrup reinforcement was of grade II with a modulus of 50 GPa [30]. Consequently, the axial stiffness parameter of the test specimens ranged between 0 and 811 MPa. Moreover, three concrete strengths of 50 and 35 and 30 MPa were used to evaluate the effect that the concrete strength may have on the shear transfer. Accordingly, the test specimens were divided into the following three series: series I, II, and III for test specimens with 50, 30, and 35 MPa concrete, respectively. Specimens of series I and II have a shear plane that is 500 mm long. However, specimens in series III are divided into two groups; group A1 includes specimens with a 400 mm shear plane and group A2 involves specimens with a 300 mm long shear plane. By reducing the interface area in group A2, an increase of 33.3% of $E_f{\rho }_v$ was achieved for the same transverse reinforcement content as compared to group A1.

One specimen of series I and II and two specimens of series III were made without any transverse reinforcement across the interface ($E_f{\rho }_v$ = 0) and are used as control. Two specimens of series I and III were reinforced with conventional steel stirrups and were used to compare the behavior of the GFRP reinforcement. Some specimens were replicated to test the reliability of the test setup.

Table 2. Test matrix.

Series	Specimen ID	${\mathit{f}}_{\mathit{c}}^{\mathit{\prime }}$ (MPa)	Reinforcement Type and Shape	${\mathit{A}}_{\mathit{v}}$ (mm2)	${\mathit{A}}_{\mathit{c}\mathit{v}}$ (mm2)	${\mathit{\rho }}_{\mathit{v}}$ (%)	$\mathit{E}{\mathit{\rho }}_{\mathit{v}}$ (MPa)
I	C0-50	50	NA	0	125,000 (250 $\times$ 500)	0.00	0
	SS1-50		Steel stirrup	200		0.16	320
	SS2-50			400		0.32	640
	FS1-50		GFRP stirrup	253.4		0.20	101
	FS2-50			506.8		0.41	203
	FS2-50 *			506.8		0.41
	FS2-50 **			506.8		0.41
	FS3-50			760.2		0.61	304
	FH2-50		GFRP headed bar	253.4		0.20	122
	FH3-50			380.1		0.30	182
	FH3-50 *			380.1		0.30
	FH5-50			633.5		0.51	304
II	C0-30	30	NA	0		0.00	0
	FS2-30		GFRP stirrup	506.8		0.41	203
	FS3-30			760.2		0.61	304
	FH3-30		GFRP headed bar	380.1		0.30	182
	FH5-30			633.5		0.51	304
III	C0-35A1	35	NA	0.0	100,000 (250 $\times$ 400)	0.00	0
	SS2-35A1		Steel stirrup	400.0		0.40	800
	FS2-35A1		GFRP stirrup	506.8		0.51	253
	FS3-35A1			760.2		0.76	380
	FS4-35A1			1013.6		1.01	507
	FH3-35A1		GFRP headed bar	380.1		0.38	228
	FH4-35A1			506.8		0.51	304
	FH5-35A1			633.5		0.63	380
	FH6-35A1			760.2		0.76	456
	FH8-35A1			1013.6		1.01	608
	C0-35A2		NA	0.0	75,000 (250 $\times$ 300)	0.00	0
	SS2-35A2		Steel stirrup	400.0		0.53	1067
	FS2-35A2		GFRP stirrup	506.8		0.68	338
	FS3-35A2			760.2		1.01	507
	FS4-35A2			1013.6		1.35	676
	FH3-35A2		GFRP headed bar	380.1		0.51	304
	FH4-35A2			506.8		0.68	405
	FH5-35A2			633.5		0.84	507
	FH6-35A2			760.2		1.01	608
	FH8-35A2			1013.6		1.35	811

* refers to the 1st repeated specimen for the reliability of data. ** refers to the 2nd repeated specimen for the reliability of data.

presents the details of the test matrix. In this table, the following nomenclature is used to identify the thirty-seven specimens. The first letter of the specimen identification represents the reinforcement type (S = steel and F = FRP), the second letter stands for the reinforcement shape (S = stirrup and H = headed bar), the third character indicates the number of the used reinforcement across the interface, and the number following the hyphen indicates the concrete strength of the specimen. The two groups of specimens of series III are identified by A1 and A2, respectively. For example, specimen FS2-35A1 refers to the test specimen with two GFRP stirrups friction shear reinforcements across the interface with a 35 MPa concrete compressive strength and a 400 mm long interface plane.

2.2. Materials Properties

Three different normal weight concrete mixes were used for two concrete strengths of 50, 35, and 30 MPa. The 50 and 30 MPa concrete was mixed in the structural lab using general-use Portland cement. The 35 MPa was a ready-mixed concrete. For each specimen, two standard concrete cylinders, 102 mm diameter $\times$ 203 mm height, were prepared from the corresponding concrete batch and were cured in the same conditions. On the same day each specimen was tested on, the corresponding cylinders were tested following the ASTM C39 standard [31]. The average compressive strengths of the cylinders were 49.7, 35, and 30.5 for series I, II, and III, respectively.

The steel stirrups were No. 10M bars with a 100 mm² cross-sectional area, and an estimated elasticity modulus of about 200 GPa and a yield strength of 400 MPa. The GFRP reinforcement was No. 4-12M sand-coated V-ROD bars with a nominal cross-sectional area of 126.7 mm² [30] Two grades of V-ROD GFRP bars (Pultrall Inc., Thetford Mines, QC, Canada) were used. The stirrups were made of grade II GFRP bars based on the manufacturer information, the modulus of elasticity was 50 GPa. The headed bars were made of grade III GFRP with a 60 GPa modulus of elasticity. The mechanical and physical properties of the GFRP transverse reinforcement, according to the supplier information, are summarized in

Table 3. Properties of the used shear friction reinforcement.

Reinforcement		Nominal Cross-Sectional Area (mm2)	$\mathbf{Ultimate} \mathbf{Tensile} \mathbf{Strength}, {\mathit{f}}_{\mathit{f}\mathit{u}}$ (MPa)	Tensile Modulus Ef (GPa)	Average Ultimate Tensile Strain, ${\mathit{\epsilon }}_{\mathit{f}\mathit{u}}$ (%)
Type	Shape
GFRP	Stirrup	126.7	1140	50	2.17
	Headed bar	126.7	1312	60	2.00
Steel	Stirrup	100	fy = 400	200	εy = 0.2

.

2.3. Test Setup and Instrumentations

The test specimens were moist cured until the targeted concrete strength was reached. Afterward, each specimen was placed on its testing position using stiff steel supports on each side. A load concentric along the shear plane was applied through a stiff steel plate using a hydraulic jack. Specimens of series I and II were tested in the vertical position. As some test specimens reached the full capacity of the steel frame, series III testing was tested in horizontal position against a rigid concrete wall. Test specimens rest on rollers to eliminate friction with the floor. The two different testing setups are illustrated in Figure 5a,b. The specimens were subjected to monolithic increasing loading until failure. The relative slip along the shear plane between the two L-shape blocks of the specimen was closely monitored using two linear variable differential transducers (LVDTs), one on each side of the specimen (Figure 5c). The strain in the transverse reinforcement across the shear plane was evaluated using 10 mm electronical foil strain gages glued on the reinforcement surface at the level of the interface (Figure 5d).

3. Test Results and Discussion

3.1. General Behavior

A summary of the test results, including the measured ultimate shear transfer load and stress ($V_u$ and $v_u$), the slippage along the interface, and the transverse reinforcement strain for each specimen at the ultimate load, are presented in

Table 4. Summary of test results.

Specimen ID	${\mathit{f}}_{\mathit{c}}^{\mathit{\prime }}$ (MPa)	$\mathit{E}{\mathit{\rho }}_{\mathit{v}}$ (N/mm2)	At Ultimate
			${\mathit{V}}_{\mathit{u}}$ (kN)	${\mathit{v}}_{\mathit{u}}$ (MPa)	Slip (mm)	Reinforcement Strain $\left(\mathit{\mu }\mathit{\epsilon }\right)$
C0-50	50	0	296	2.37	0.66	-
SS1-50		320	334	2.67	0.14	-
SS2-50		640	477	3.81	0.85	1904
FS1-50		101	334	2.67	0.34	22
FS2-50 *		203	402	3.21	0.48	3881
FS3-50		304	617	4.94	0.37	402
FH2-50		122	336	2.69	0.31	389
FH3-50 *		182	323	2.58	0.28	2260
FH5-50		304	569	4.55	0.77	2953
C0-30	30	0	332	2.65	0.34	-
FS2-30		203	385	3.08	0.58	4466
FS3-30		304	384	3.07	0.64	4847
FH3-30		182	362	2.89	0.44	1472
FH5-30		304	433	3.46	0.94	4973
C0-35A1	35	0	343	3.43	0.41	-
SS2-35A1		800	281	2.81	0.39	2486
FS2-35A1		253	268	2.68	0.34	2209
FS3-35A1		380	489	4.89	0.72	5962
FS4-35A1		507	429	4.29	0.11	3976
FH3-35A1		228	328	3.28	1.00	5993
FH4-35A1		304	294	2.94	0.37	2383
FH5-35A1		380	497	4.97	0.67	6988
FH6-35A1		456	547	5.47	0.80	5752
FH8-35A1		608	550	5.50	1.00	5889
C0-35A1		0	217	2.89	0.28	-
SS2-35A2		1067	189	2.52	0.83	2343
FS2-35A2		338	172	2.29	0.41	3113
FS3-35A2		507	275	3.66	0.92	8828
FS4-35A2		676	484	6.46	0.94	3619
FH3-35A2		304	213	2.84	0.97	6563
FH4-35A2		405	218	2.90	0.30	2308
FH5-35A2		507	264	3.52	1.00	9010
FH6-35A2		608	304	4.05	0.60	3543
FH8-35A2		811	397	5.30	0.97	6081

* average value of replicated specimens.

. The load-slip relationships of each group of specimens having a similar reinforcement shape and concrete strength are shown in Figure 6.

Figure 6 shows all the shear load-slip behaviors of the tested specimens. Despite the variation in investigated test parameters among the test specimens, including the parameter, shape and type of the shear reinforcement, and the concrete compressive strength, the general shear load-slip behaviors of all test specimens can be described as successive tri-linear phases, representing the pre- and post-cracking phases I and II, as well as the post-peak phase III, as explained in the following sections and Figure 6.

3.1.1. Phase I: Pre-Cracking Behavior

Phase I represents the behavior of the specimen at loading levels below the cracking load, which is the load at which a crack along the interface is initiated. During this phase, the load is carried mainly by the shear resistance of the uncracked interface alone. A bi-linear load-slip relationship can be used to describe phase I, as two shear transfer mechanisms are involved. The first mechanism is due to the chemical adhesion bond between the concrete on both sides of the shear interface. Once the shear stresses at the interface exceed the adhesion strength (point 1 in Figure 6), the second mechanism, due to the mechanical aggregate interlocking, is activated and the load-slip behavior continues at a reduced stiffness until a longitudinal crack is developed along the shear plane (point 2, Figure 6). Test specimens without shear friction reinforcement failed suddenly at the end of this stage.

The shear friction reinforcement has no effect on the load-slip response during phase 1 and the cracking load. Previous studies have also shown that pre-cracked behavior is independent of the steel shear friction reinforcement [11,19,32]. The cracking shear stress of all the tested cold-joint interfaces of the present study was found to be in the range from 2 to 2.7 MPa. It is worth noting that this cracking strength is consistent with previous findings for steel-reinforced cold-joint interfaces [16,19].

3.1.2. Phase II: Post-Cracking Behavior

Owing to the general roughness and irregularities of the cracked interface surface, the slip along the interface in phase II is accompanied by widening of the crack along that interface. Thus, the reinforcement crossing the cracked shear plane would be subjected to tension forces. When transverse reinforcement is adequately provided across the interface, it applies a compressive clamping stress along the cracked interface which, in turn, increases friction resistance. Consequently, the post-cracking shear resistance continues to increase linearly beyond cracking up to the ultimate load, $V_u$ (Point 3 in Figure 6). This phase II is characterized by a reduction in the stiffness in the load-slip behavior, larger and visible crack widths along and across the interface, and a rapid increase in the slip and reinforcement’s strains, as shown in Figure 7 and Figure 8, compared to phase I. It was observed that only specimens reinforced with a GFRP stiffness parameter, $E_f{\rho }_v$, higher than 203 N/mm², exhibited post-cracking frictional shear transfer resistance. Post-cracking friction was absent in specimens containing GFRP reinforcement with stiffness parameter less than 203 N/mm², such as FS1-50, FH2-50, and FH3-50. In the latter case, those specimens did not show a significant increase in the shear strength after cracking and they failed at the cracking point or shortly afterwards (Figure 8a). It is evident that the frictional shear resistance after cracking is directly related to the amount of the transverse reinforcement represented by the stiffness parameter ($E_f{\rho }_v$).

The relative slip values at the ultimate load of the adequately reinforced GFRP interfaces were in the range of 0.37 to 1.00 mm. Both phase I and II confirm the well-established shear friction theory, where comparable load-slip behaviors and values were reported for steel-reinforced concrete interfaces in this study and the available literature [13,19].

3.1.3. Phase-III: Post-Ultimate Behavior

After the ultimate load is attained (point 3 in Figure 6), any attempt to further increase the applied load resulted in widening the interface crack and hence a significant reduction in the friction between the interconnected blocks. The post-ultimate behavior is characterized by a sudden drop in the load resistance with increased slip and reinforcement strain, as shown in Figure 7 and Figure 8. For all steel-reinforced specimens, the peak load corresponded to the yielding of the reinforcement across the interface and therefore no residual shear friction resistance was observed beyond the peak; the failure was brittle and accompanied with a loud bang. However, in the case of adequately GFRP-reinforced specimens, a little reduction in the peak load was noticed, but the specimens were able to sustain a significant post-peak residual resistance afterwards at higher relative slips (Figure 6, point 4). This behavior can be attributed to the linear elastic behavior of GFRP bars, as well as the excellent bond performance of GFRP compared to steel bars. The ultimate load in the adequately GFRP-reinforced specimens was noticed at around 5000 microstrains.

3.2. Effect of the Reinforcement Stiffness Parameter (${\mathit{E}}_{\mathit{f}}{\mathit{\rho }}_{\mathit{v}}$)

Table 4 shows that in the cases of GFRP-reinforced specimens with stiffness higher than $203\hspace{0.17em}\mathrm{N}/\mathrm{m}{\mathrm{m}}^2$ ($E_f{\rho }_v\ge 203$ N/mm²) increasing the axial stiffness parameter led to the enhancement of the ultimate shear transfer strength. For example, the shear strength of specimen FS3-50 is twice that of specimen FS1-50, for an increase in the reinforcement stiffness of three times. Increasing the axial stiffness parameter of specimen FH5-50 by 67%, compared to FH3-50 (from 182 N/mm² to 304 N/mm²), resulted in an increase in the ultimate load by about 79%. However, the same increase in stiffness led to only a 20% increase in the capacity for specimens with a concrete strength of 30 MPa in series II. The influence of the reinforcement stiffness appears to be more pronounced for higher concrete strength. The effect of the reinforcement stiffness can be observed among the specimens of series III as well. For example, raising the stiffness from 253 MPa for FS2-35A1 to 676 MPa for FS4-35A2 (66% increase) leads to a 41% increase in the shear transfer strength.

Figure 8 shows the load-strain behavior of the GFRP-reinforced specimens. It can be noticed that the strain and therefore the stress in the GFRP shear friction reinforcement before the cracking load was negligible, varying from 22 to 620$\mu \epsilon$. This indicates that there was no practical role of this reinforcement before the cracking onset. Figure 8a illustrates the load-strain behavior of the GFRP under-reinforced specimens of series I and II (i.e., $E_f{\rho }_v <203$ N/mm²). It can be noticed that there is no increase in the interface load resistance after cracking. However, specimens with adequately transverse GFRP reinforcement ($E_f{\rho }_v \ge 203$ N/mm²) exhibited a significant increase in the load-carrying capacity and strains were noticed after cracking. Specimens of series III were all sufficiently reinforced and showed a similar trend.

The strain rapidly increased after the cracking load and reached the range of about 4000–6000 microstrains at the ultimate load. The sufficiently GFRP-reinforced specimens sustained a significant residual load after the peak load, not less than 80% of the ultimate load. It was also noted that a more residual sustained strength was delivered as higher reinforcement stiffness was provided. As GFRP bars do not yield but can sustain higher stress, they can still provide additional shear transfer resistance through clamping effect and dowel action. On the other hand, steel reinforced specimens, SS2-50, SS2-35A1, and SS2-35A2, reached their ultimate strengths at deformation 1904, 2486, and 2343 microstrains, respectively, which were near the yielding strain. Specimens FS4-35-A1 and FS3-35-A2 have the same stiffness, concrete strength, and spacing between reinforcement. The only difference is the length of the concrete interface—400 vs. 300 mm. Figure 9 shows very similar load-slip behaviors. Specimens FS4-35-A1 showed peak shear capacity of 429 kN compared to 275 kN for FS3-35-A2. This can be attributed to the larger interface area resulting in a more direct friction force.

3.3. Effect of the Reinforcement Type (GFRP vs. Steel)

Figure 10 compares the load-slip behaviors of series I test specimens reinforced with steel and GFRP stirrups (SS1-50, SS2-50, FS1-50, FS2-50, and FS3-50). As presented also in Table 4, SS1-50 failed just after cracking at an ultimate load 13% higher than that of unreinforced specimen C0-50. However, the specimen with two steel stirrups, SS2-50, developed a 61% increase in the ultimate strength, compared to C0-50, after cracking of the shear plane. On the other hand, FS1-50 exhibited almost identical behavior to that of SS1-50, as both specimens have inadequate shear reinforcement. FS2-50, with reinforcement stiffness $E_f{\rho }_v$ equal to 203 N/mm², developed about a 36% additional frictional shear resistance after cracking. Also, FS2-50 showed a remarkable deformable behavior after the peak load, with about 80% residual capacity up to about a 6 mm slip compared to SS2-50. Specimen FS3-50 developed almost double the ultimate strength of C0-50. Increasing the reinforcement stiffness by 50%, from 203 N/mm² for FS2-50 to 304 N/mm² for FS3-50, increased the shear transfer strength by about 53%. It should be noted that during the testing of FS3-50, the cracks along the shear plane developed at about 617 kN, which was near the capacity of the testing frame, so the test had to be stopped before complete failure. These observations indicate a major role of the GFRP reinforcement stiffness in the shear transfer strength of a concrete interface when it is sufficiently provided across that interface. Also, GFRP shear friction stirrups showed comparable and even superior behavior compared to the steel ones. The stiffness parameter of FS3-50 is less than 50% of SS2-50; however, FS3-50 developed 30% more ultimate strength than SS2. This would be owing to the perfect elastic and bond characteristics of the GFRP reinforcement.

When comparing the behavior of GFRP reinforcement with conventional steel stirrups, it is to be noted that the performance of GFRP is very competitive, with steel as a shear transfer reinforcement. Specimens FS2-50, FS2-35A1, and FS2-35A2 have stiffness parameters that are about one-third of that of SS2 specimen of the same group; they developed shear transfer strengths of 84, 95, and 91% of the corresponding SS2 specimen, respectively.

3.4. Effect of the Reinforcement Shape (Stirrups vs. Headed Bars)

Figure 11 compares the load-slip behaviors for series I test specimens FS3 and FH5 which have the same stiffness parameter of 304 MPa. Specimen FS3-50 developed about only 8% and 14% higher shear capacity than FH5-50. After cracking, the sufficiently reinforced specimen FH5-50 recovered almost its ultimate capacity and exhibited an outstanding deformable behavior up to about a 6 mm slip. This can be attributed to the excellent bond characteristics of the headed bars, which could prevent any slippage of the reinforcement. In conclusion, the influence of the GFRP reinforcement shape, stirrup, headed bar, and angles on the ultimate strength is not significant. However, headed bars were found to provide a remarkable deformable behavior after the ultimate load was achieved. For the same concrete strength and stiffness parameter, GFRP-reinforced specimens with different shapes developed a comparable shear capacity. Specimens FS3-50 and FH5-50 of series I have an identical stiffness parameter as that of 304 MPa. The strength of FS3-50 was only 8% higher than that of FH5-50. Similarly, the shear capacities of FS3-35A1 and FH5-35A1, with a stiffness parameter of 380 MPa, were about 1.6% apart in terms of ultimate capacity.

While all sufficiently GFRP-reinforced specimens showed adequate post-ultimate resistance, those with headed bars were superior. For instance, specimens FH5-50 recovered almost its ultimate capacity and exhibited an outstanding deformability behavior up to about a 6 mm slip. This behavior can be attributed to the high capacity of GFRP as well as the superior bond characteristics of the GFRP headed bars, which could prevent any slippage of the reinforcement. It could be noted, however, that the transverse reinforcement, either stirrups or headed bars of the same stiffness parameter, has comparable shear resistance and the difference between the two reinforcements is marginal.

3.5. Effect of the Concrete Compressive Strength (${\mathit{f}}_{\mathit{c}}^{\prime }$)

To study the influence of the concrete strength, the load-slip responses of specimens of series I and II with the same reinforcement stiffness parameter and shape are reported in Figure 12. From Figure 12a, it is noted that FS2-50 and FS2-30 have almost the same load-slip behavior with a very marginal difference in the ultimate capacity. The ultimate capacity of FS2-50 is barely higher than that of FS2-30, by a margin of 6.5%. The influence of the concrete strength on the shear transfer capacity appears to increase in significance when higher transverse reinforcement content is used, as when comparing specimens FS3-50 and FS3-30. Figure 12a shows that specimen FS3-50 exhibited a significant 61%, increase in the ultimate capacity compared to FS3-30. The same observations are also noted for headed bars specimens of series I and II. For example, the difference in the ultimate strength between FH3-50 and FH3-30 was less than 12% compared to an increase of 32% between FH5-50 and FH5-30, when the concrete capacity increased by 66%. However, the last two specimens exhibited the same load-slip trend (Figure 12b). Examining Figure 12a,b, it is noted that up to a slip of 1 mm, specimens with sufficient reinforcement (FS2, FS3, and FH5) and higher concrete strength also resulted in much stiffer pre-cracking responses. This was not the case for specimens with low transverse reinforcement such as FH3-30 and FH3-50, which had similar stiffness prior to cracking. These observations suggest that it is better to combine high concrete strength with a higher transverse stiffness to achieve a substantial increase in the shear capacity. The limit of stiffness of $203\hspace{0.17em}\mathrm{N}/{\mathrm{m}\mathrm{m}}^2$ is suggested from the present study.

3.6. Failure Modes

The observed failure mode of the push-off specimen depends mainly on the axial stiffness parameter of the shear friction reinforcement and to a lesser extent on the shape and type of this reinforcement. All the specimens with low transverse reinforcement failed in a typical mode like the unreinforced control specimens. Test specimens C0, FS1, FH2, and FH3 in series I, II, and III developed a first crack along the shear plan; a sudden failure followed. No other cracks or spalling of the concrete cover were observed around the shear plan, as shown in Figure 13. Meanwhile, test specimens reinforced with two or more GFRP stirrups sustained a higher relative slip after cracking and exhibited extensive spalling of the concrete cover (Figure 14a). Specimens of steel stirrups in all series exhibited similar cracking and spalling patterns (Figure 14b).

When GFRP headed bars were used, no concrete spalling was observed except for specimens with four or more headed bars across the shear plan, in which the thickness of the concrete cover was relatively small (Figure 14c). Generally, the concrete spalling was associated with high shear slips. In heavily reinforced specimens, some of the GFRP headed bars ruptured at advanced loading stages, as shown in Figure 14d. This indicates the ultimate state is controlled by the GFRP bars capacity.

4. Proposed Shear Friction Equation

According to the shear friction theory, the clamping force provided by the reinforcement across the interface controls the shear transfer strength after the initiation of the crack along the interface. It is well established that the clamping stress is directly related to the available transverse reinforcement intersecting the shear plane. For steel transverse reinforcement, the classical shear friction theory postulates that when the interface cracked and the steel-reinforcing bars are properly anchored on both sides of the interface, their stresses can reach the yielding strength. In the case of the GFRP transverse reinforcement, no yielding occurs. Therefore, an alternative ultimate limit state must be introduced for GFRP-reinforced concrete-to-concrete interfaces. The axial clamping force in the given by:

$$P={\epsilon }_f{E}_f{A}_v$$

(1)

where ${ϵ}_f$, $E_f$, and $A_v$ are the strain, the elastic modulus, and $A_v$ is the total area of the GFRP bars. By normalizing the clamping force with the interface area $A_c,$ the clamping force can be replaced by the equivalent clamping stress:

$${\sigma }_f={\epsilon }_f{E}_f{\rho }_v$$

(2)

where ${\rho }_v =A_v/A_c$ is the transverse reinforcement ratio.

As it was illustrated in the test results, the shear resistance of GFRP-reinforced interfaces is mainly dependent on the axial stiffness parameter, $E_f{\rho }_v$, and then on the concrete compressive strength. The tests show that when an adequate amount of transverse reinforcement is provided, GFRP bars can sustain substantial clamping resistance for strain level up to $5000 \mu \epsilon$.

Accordingly, it is proposed that at the ultimate state, the clamping stresses developed by the GFRP are estimated as follows:

$${\sigma }_f=0.005E_f{\rho }_v$$

(3)

Figure 15 shows the analytical model best-fitting linear analysis of the test results of GFRP-reinforced specimen; the unreinforced control ones were excluded. In this figure, clamping stress and the tests shear stresses were normalized with respect to the concrete strength (${0.005 E}_f{\rho }_v /f_c^{\prime })$ along the horizontal axis and the normalized ultimate shear stress at the interface ($v_u/f_c^{\prime })$ for the GFRP reinforced specimens. The vertical axis intersection at a value of 0.07 represents the cohesion factor of uncracked concrete interfaces. The slope of the linear relationship is 0.48, which would represent the friction coefficient of cold concrete joints not intentionally roughened. It is worth noting that this is close to the ACI-319 recommended value of 0.6, as shown in Table 1, for concrete placed against hardened concrete not intentionally roughened. Accordingly, the average shear transfer strength can be estimated in a way similar to the current steel-reinforced interfaces as follows:

$$v_u = 0.07f_c^{\prime }+0.48 {\epsilon }_f{E}_f{\rho }_{v }$$

(4)

Using ${ϵ}_f=5000 \mu \epsilon$, the following equation is suggested to calculate the ultimate shear transfer strength of GFRP-reinforced composite constructions:

$$v_u = 0.07f_c^{\prime }+0.0024 E_f{\rho }_{v }$$

(5)

It is not recommended to use GFRP reinforcement content with a stiffness parameter less than 203 N/mm². Having an adequate GFRP reinforcement across the concrete interface increases the predictability of shear transfer behavior and strength and avoids brittle post-cracking rapid failure modes. A conservative lower bound expressions can be used for all reinforcement ratios, as shown in Equation (6)

$$v_u=0.04f_c^{\prime }+0.0024{ E}_f{\rho }_{v }$$

(6)

Additionally, if a conservative design criteria implied by the shear friction theory is adopted, as in the ACI 318 and CAN/CSA A23.3, the $0.07f_c^{\prime }$ cohesion term in Equation (5) could be neglected and Equation (7) could be used to evaluate the shear transfer strength for pre-cracked concrete interfaces (Figure 15).

$$v_u=0.0024{ E}_f{\rho }_{v }$$

(7)

5. Conclusions

This study proposes a new application of the glass-FRP as transverse shear reinforcement across concrete joints. To explore the feasibility and effectiveness of this application, an experimental study and a parametric analysis were conducted on forty push-off test specimens. The studied parameters included the reinforcement stiffness parameter ($E_f{\rho }_v$), the shape of the GFRP reinforcement, and the concrete compressive strength ($f_c^{\prime }$). Based on the conducted analysis of the test results, the following conclusions can be drawn:

• A minimum GFRP transverse reinforcement stiffness parameter ($E_f{\rho }_v$) of 203 N/mm² is required to activate the reinforcement contribution in proving additional frictional shear resistance after the interface cracking phase.
• The axial reinforcement stiffness of the GFRP reinforcement was found to be the dominant shear transfer parameter after the initiation of the interface cracking.
• Specimens with a low GFRP axial stiffness parameter, below the limit of 203 N/mm², failed suddenly at the cracking load. The reinforcement provided was not capable to carry any additional load.
• Interfaces with GFRP headed bars and stirrups having similar stiffness were found to exhibit similar shear transfer strength. Both reinforcement types could be used.
• Like steel-reinforced interfaces, the shear stress-slip ($v-∆$) response of as-cast cold-joint GFRP-reinforced concrete interfaces involves three successive phases, which are pre-cracking, post-cracking, and post-ultimate.
• Higher concrete compressive strength results in higher shear transfer strength and stiffer pre-cracked load-slip response, on the condition that the reinforcement stiffness is higher than the minimum of 203 N/mm².
• The strain limit value of 5000 $\mu \epsilon$ is a reasonable approximation for the strain of the GFRP bars across the interface at the ultimate load.

GFRP-reinforced specimens showed remarkable post-ultimate load-carrying capacity compared to specimens with steel reinforcement.