1. Introduction
Reinforced concrete (RC) is the most popular composite material in structures since the 20th century owing to its economic efficiency, availability, and mechanical properties. However, exposure to a marine environment may rapidly degrade the durability of RC structures as the reinforcing bars are constantly corroded by chloride attacks [
1,
2]. To prevent such chloride attacks, certain design regulations are followed to maintain the minimum cover depth for RC members, and the void between concrete and reinforcing bars is minimized via effective construction.
Furthermore, to coat the surface of reinforcing bars with epoxy, a technology was developed [
3,
4] to prevent the chloride ions from corroding the bars; however, this technology was applied only to a limited number of bridges owing to the decreased coating effect and the occurrence of local corrosion when the coating layer is peeled off or torn during concrete pouring [
5]. Moreover, the surface epoxy coating may lead to bond problems between the bars and concrete [
6].
Since the 1970s, several studies considered the application of fiber-reinforced plastics (FRPs) with a strong resistance to corrosion when used as the reinforcement for concrete structures, rather than using reinforcing bars, to prevent chloride attacks on RC structures [
7]. FRP reinforcement was primarily applied to bridge slabs in the form of bars, using glass, carbon, and aramid fibers as reinforcing fibers [
7,
8]. Recently, the development of grid-type FRP reinforcement [
9] and FRP tendons [
10] led to the expansion of applying FRP reinforcement to various structures, including building walls [
11], external strengthening of structures [
12,
13], and bridge foundations.
Among them, the grid-type FRP reinforcement is a type of textile reinforcement for textile-reinforced concrete (TRC) structures [
14] that uses fibers woven in grid form as the reinforcement for concrete. Typical textile reinforcement is produced by only weaving fibers, such as glass, aramid, carbon, and basalt fibers, in grid form without impregnating them with resin. Although the grid-type FRP reinforcement exhibits higher tensile strength than reinforcing bars, the absence of plastic deformation capacity results in rapid brittle fracture after reaching the tensile strength [
7]. Several researchers [
15,
16] reported that when fibers are impregnated or coated with resin such as FRP reinforcement, the tensile behavior of TRC members is improved in terms of smaller crack spaces, enhanced stable crack geometry, and increased maximum load.
The flexural behavior of TRC members differs from that of RC members [
17] because textile reinforcement with physical properties different from reinforcing bars are used as the reinforcement for concrete (
Figure 1). In other words, a linear elastic behavior is observed, similar to that in RC members, before the occurrence of cracks. However, the continuous occurrence of multiple cracks results in the transformation of tensile stress to textile reinforcement at the crack cross-section; no new cracks occur after the crack formation stage, where the load hardly increases. As cracking is stabilized and textile reinforcement resists the load, the load gradually increases owing to the increase in deformation, leading to failure because of the non-homogeneous pull-out between the fibers that constitute the textile reinforcement or the fracture of the textile reinforcement.
The flexural behavior of TRC members was analyzed considering the textile reinforcement that uses fibers not impregnated with resin. However, only a few studies evaluated the flexural behavior of TRC members reinforced with grid-type FRP reinforcement that impregnates fibers with resin [
18,
19,
20]. These studies also used grid-type FRP reinforcement with a small strand spacing of less than 50 mm in the weft and warp directions and a small strand cross-sectional area because of the high tensile strength of FRPs.
In this study, the flexural behavior of TRC one-way slabs was experimentally evaluated by applying grid-type carbon fiber-reinforced plastic (CFRP) as a reinforcement for concrete, hereafter referred to as carbon grid, manufactured by impregnating carbon fibers with resin. Nine types of carbon grids, manufactured by two different methods and with different geometric and physical properties, were used. The flexural behavior of TRC slabs was also directly compared to that of RC slabs. To this end, one-way slab specimens were prepared by applying carbon grids using two different manufacturing methods with varying cross-sectional areas, spacings, tensile strengths, and elastic moduli of longitudinal strands and reinforcing bars. Their flexural behaviors were evaluated using a three-point bending test.
3. Results and Discussion
3.1. Crack and Failure Geometry
Figure 6 presents the final failure geometry of the carbon grid-reinforced specimens.
The specimens with both biaxial warp-knitted and cross-laminate structured carbon grids demonstrated the occurrence of cracks at the center of each specimen. Subsequently, new cracks occurred from the center of the specimen to both ends owing to the applied load. At the center of the specimen, the mortar was crushed and separated.
In specimens G1, G2, and G3 reinforced with the biaxial warp-knitted structured carbon grids, 11, 17, and 11 cracks occurred, respectively (
Table 5 and
Figure 7). The average crack spaces ranged from 45 to 63 mm, wider than the transverse strand spacings of the carbon grids. Furthermore,
Figure 7 indicates that a bond crack between the carbon grid and mortar was observed in the case of specimen G2.
Conversely, the specimens reinforced with the cross-laminate structured carbon grids had fewer cracks, ranging from four to ten, compared with those observed in the specimens reinforced with the biaxial warp-knitted structured carbon grids. Moreover, the average crack spaces ranged from 81 to 106 mm. The average crack space for specimen J2 was 106 mm, which was more than twice the transverse strand spacing of 50 mm. On the other hand, the average crack spaces of other specimens were smaller than the transverse strand spacing of 100 mm.
Bond cracks were also observed in specimen J2 (
Figure 7); as in specimen G2, the average crack spaces in specimens J2 and G2 were considered to be more than twice the transverse strand spacings because of the poor bond between the carbon grids and mortar.
By the end of the experiment, no strand fractures were observed in the specimens reinforced with the cross-laminate structured carbon grids. Furthermore,
Figure 8 indicates that only a few fibers of the strands were disconnected in specimen J5. However, in the case of specimens reinforced with the biaxial warp-knitted structured carbon grids, the longitudinal strands were fractured in specimens G3 and G1 when the mortar was removed from the center of the specimens after the experiment (
Figure 8). Additionally, residual deformation of the carbon grid was observed in specimen G2. In the case of biaxial warp-knitted structured carbon grids, several voids occurred in the cross-section of the strands as they were manufactured using the hand lay-up method (
Figure 3a). The strands appeared to be fractured owing to the increase in the applied force and the concentration of this force on the voids.
3.2. Flexural Behaviors of Specimens with Carbon Grids
3.2.1. Load–Deflection Relationship
In all specimens reinforced with the biaxial warp-knitted and cross-laminate structured carbon grids, cracks were initially observed at the center of each specimen. Furthermore, multiple cracks continuously developed toward the ends of the specimens, thereby significantly reducing the stiffness due to the transfer of tensile stress to the carbon grid (
Figure 9). Subsequently, the maximum load was reached by mortar crushing in the mid-span of the specimen. Although the load in most specimens was reduced in stages after reaching the maximum load, specimens G2 and G3, which were reinforced with the biaxial warp-knitted structured carbon grids, exhibited a sharp load reduction.
In the case of specimens reinforced with the biaxial warp-knitted structured carbon grids, specimen G2, with the highest tensile capacity, exhibited the largest crack formation and maximum loads (
Table 6). The load of the specimens reinforced with the biaxial warp-knitted structured carbon grids decreased to approximately 11 to 14% of the maximum load upon completion of the experiment owing to the fracture of the longitudinal strands and residual deformation of the carbon grid (
Figure 9a).
In the case of specimens reinforced with the cross-laminate structured carbon grids, the crack formation load did not significantly differ depending on the type of carbon grid. Furthermore, the maximum load did not increase in proportion to the tensile force by the longitudinal strands (
Table 6). In addition, specimen J2 had a smaller longitudinal strand spacing (50 mm) than specimen J3 with the 100 mm spacing, and exhibited an increase in stiffness after crack formation and maximum load due to the increase in the number of strands (
Figure 10).
3.2.2. Behavior of Crack Formation and Stabilization Stages
Table 7 summarizes the stiffness of the pre-cracking, crack formation, and crack stabilization stages, and
Figure 11 illustrates the stiffness of the crack formation and stabilization stages. The stiffness of the pre-cracking stage indicates the stiffness between the beginning of the experiment and the initiation of crack formation. The stiffness of the crack formation stage represents the stiffness between the initiation of crack formation and the point at which cracks stop occurring continuously. The stiffness of the crack stabilization stage indicates the stiffness between the point at which cracks stop occurring continuously and the peak.
The stiffness of the crack formation and stabilization stages of the carbon grid-reinforced specimens ranged from 4 to 8% and 5 to 8% of the pre-cracking stiffness, respectively, indicating that the stiffness of the specimens was rapidly reduced by cracking, and the stiffness hardly recovered even after the stabilization of crack formation.
Figure 11g–i indicate that among the specimens reinforced with the cross-laminate structured carbon grids, specimens J5 and J6 with the largest cross-sectional areas of the longitudinal strands and specimen J2 with a small strand spacing immediately exhibited the behavior of the crack stabilization stage without the crack formation, unlike the behavior observed in other specimens; in the case of these three specimens, the load significantly increased, and the mid-span deflection increased only marginally because of cracking. The stiffer carbon grids with larger cross-sectional areas or smaller spacing of the strands used in specimens J5, J6, and J2 might affect the behavior of crack formation stage.
The elastic modulus of the biaxial warp-knitted structured carbon grids was more than twice as high as that of the cross-laminate structured carbon grids. However, the stiffness of the specimens reinforced with the biaxial warp-knitted structured carbon grids in the crack formation and stabilization stages was not significantly different from that observed in the same stages of specimens J1, J3, and J4, which were reinforced with the cross-laminate structured carbon grids (
Table 7). Moreover, the stiffness was lower than that observed in the crack stabilization stage of specimens J2, J5, and J6 because the stiffness of cracked section is associated with the effective moment of inertia [
7], which is affected by both the total cross-sectional area and elastic modulus of the longitudinal strands. As shown in
Figure 12, the stiffness of the specimens in the crack formation and stabilization stages tended to be proportional to the values obtained by multiplying the total cross-sectional area of the longitudinal strands by the elastic modulus of the longitudinal strands.
3.2.3. Post-Peak Behavior
Table 8 lists the ratios of strain to the tensile strain of the longitudinal strands in the carbon grids under and after the maximum load.
Figure 13 illustrates the relationship between the mid-span deflection of the carbon grid-reinforced specimens and strain of the longitudinal strands; here, the strain indicates the value measured using the strain gauges installed on the strands (
Figure 3).
Specimens G1, J1, J2, J3, J4, J5, and J6 exhibited a stage-wise load reduction tendency after reaching the maximum load (
Figure 9) as the mid-span deflection of the specimens increased. The strains of the longitudinal strands under the maximum load ranged from 38 to 65% of the tensile strain (
Table 7), indicating the occurrence of tensile stress that corresponded to 38 to 65% of the tensile strength.
In the case of specimen G1, the load decreased after reaching the maximum load, and a certain amount of the load was recovered as the carbon grid resisted the load. However, the load rapidly decreased again as a strain of approximately 80% of the tensile strain was measured. Finally, the load sharply decreased and failure occurring as a strain of approximately 83% of the tensile strain was measured. In the case of specimen J5, a certain amount of the load was recovered as the carbon grid resisted the load after reaching the maximum load; nevertheless, the load rapidly decreased again as a strain that corresponded to approximately 47% of the tensile strain was measured. The load gradually decreased, and the final failure occurred because of the sharp reduction in load when a strain of approximately 93% of the tensile strain was measured. In the case of specimen J6, the load rapidly decreased after reaching the maximum load as a strain corresponding to approximately 83% of the tensile strain was measured. Subsequently, the load sharply decreased, and final failure was reached as a strain of approximately 102% of the tensile strain was observed.
In the case of specimen G1, J5 and J6, a difference was observed between the strain measurement results and the damage status of the longitudinal strands. The strain gauges were practically installed on the resin on the strand surface and recorded the deformation of resin and some fibers only at the position where the strain gauge was attached; thereby, the measured values could be affected by the strand surface condition. The biaxial warp-knitted structured carbon grid used in specimen G1 used epoxy with relatively high physical properties as resin and had the small cross-sectional area of the strands and the cross-sectional voids generated during manufacturing. Therefore, the strands were considered to fracture by the occurrence of strain corresponding to 83% of the tensile strain. Conversely, the cross-laminate structured carbon grid used in specimens J5 and J6 used vinyl ester, with lower physical properties than epoxy [
23] and a larger cross-sectional area. Consequently, it was considered that the strands were not fractured despite the occurrence of strain that corresponded to approximately 93 and 102% of the tensile strain.
The load in specimens G2 and G3, with the biaxial warp-knitted structured carbon grids, tended to reduce sharply after reaching the maximum load (
Figure 9). In the case of specimen G2, a strain corresponding to approximately 64% of the tensile strain was measured under the maximum load. However, the load rapidly decreased after reaching the maximum load, and final failure occurred owing to the expansion of the bond crack generated between the carbon grid and mortar, unlike the behavior observed in specimens G1, J1, J2, J3, J4, J5, and J6. In other words, it was considered that the load in specimen G2 rapidly decreased immediately right after reaching the maximum load, although approximately 64% of the tensile stress was applied to the longitudinal strands under the maximum load because of the poor bond between the carbon grid and mortar. The strands were not fractured, but the residual deformation on the carbon grid occurred at the end of the experiment (
Figure 8).
Among the specimens, the largest tensile stress was observed in specimen G3 at a strain corresponding to approximately 82% of the tensile strain was measured when the maximum load was reached. Although a certain amount of the load was maintained after reaching the maximum load, the load suddenly dropped and final failure occurred as the cracks at the center of the specimen expanded and the longitudinal strands were fractured under the application of a load higher than the tensile strength because of the brittle fracture characteristics of CFRP composites [
24].
3.2.4. Idealized Load and Mid-Span Deflection of Specimens with Carbon Grids
Based on the experimental results,
Figure 14 illustrates the idealized relationship between the load and mid-span deflection of the carbon grid-reinforced specimens.
The experimental results indicate that after cracking, the stiffness decreased to approximately 8% or lower than the stiffness before cracking, except for the three specimens reinforced with relatively stiffer carbon grids. Additionally, the maximum load was reached when the stiffness was only marginally recovered, even at the crack stabilization stage. In other words, the load gradually increased owing to the increase in the mid-span deflection with a similar stiffness at the crack formation and stabilization stages even though the stiffness decreased sharply because of the crack occurrence, unlike the behavior observed in flexural members that used textile reinforcement without resin (
Figure 2). Therefore, the load and mid-span deflection relationship might be modeled as a bilinear curve in which the stiffness changes around the crack initiation point (
Figure 14).
In accordance with the experimental results, when the strain of the longitudinal strands was less than approximately 80% of the tensile strain under the maximum load, the load was found to decrease in stages as the load resisted by the mortar was transferred to the carbon grid when the bond between the carbon grids and mortar was secured (ii in
Figure 14). Conversely, it was found that the load resisted by the mortar could not be transferred to the carbon grid after reaching the maximum load, while the load decreased rapidly when the bond between the mortar and carbon grid was not secured, or strain corresponding to approximately 80% of the tensile strain or higher occurred on the longitudinal strands under the maximum load (i in
Figure 14).
3.3. Comparison between the Experimental and Calculated Results
The maximum load did not increase as the tensile capacity by the longitudinal strands of the carbon grid-reinforced specimens increased (
Figure 15). Particularly, in the case of specimens reinforced with the cross-laminate structured carbon grids, the maximum load did not increase in proportion to the tensile capacity by the longitudinal strands because all specimens reached the maximum load by mortar crushing. As shown in
Table 7, the strains of the longitudinal strands under the maximum load ranged from 38 to 82% of the tensile strain. Particularly, the longitudinal strands of the cross-laminate structured carbon grids exhibited a smaller strain than the longitudinal strands of the biaxial warp-knitted structured carbon grids. On the other hand, the maximum load tended to be higher in the specimens reinforced with small cross-sectional areas and spacings of the longitudinal strands but high tensile strength compared with the specimens reinforced with large cross-sectional areas and spacings of the longitudinal strands but low tensile strength.
Figure 16 presents the crack and flexural strength ratio of experimental and calculation results, considering the tensile force by the longitudinal strands.
The crack strength ratio ranged from 0.77 to 1.35, and the flexural strength ratio ranged from 0.71 to 1.33. The maximum load was generally higher than the flexural capacity. Particularly, in the case of specimens reinforced with the biaxial warp-knitted structured carbon grids with a relatively low tensile capacity, the maximum load was more than 1.0 times higher than the flexural capacity. In the case of specimen G2, which exhibited a poor bond between the carbon grid and mortar, the maximum load was 1.3 times higher than the flexural capacity, indicating that the flexural capacity was conservatively evaluated.
Flexural stiffness is usually calculated using the moment of inertia related to the section states: the gross moment of inertia,
Ig, for the uncracked section and effective moment of inertia,
Ie, for the cracked section. It is known that in the case of TRC members, the effective moment of inertia becomes lower compared with that of RC members because of the different material properties of FRP reinforcement. Therefore, researchers [
7,
25] proposed the formation of an effective moment of inertia which modified Branson’s equation for RC members [
26] based on experimental results using glass fiber-reinforced plastic (GFRP) bars, as presented in Equation (1).
where,
Icr is the cracked transformed moment of inertia (mm
4) of the following Equation (2),
Mcr is the cracking moment (Nmm), and
Ma is the maximum load moment for the deflection calculation (Nmm).
where,
b is the width (mm),
d is the effective depth (mm),
n is the elastic modulus ratio of the carbon grid and concrete,
ACFRP is the longitudinal strand area (mm
2), and
k is the normalized neutral axis depth of the cracked section.
In all specimens, the calculated stiffness of the cracked sections using
Ie in Equation (1) was revealed to be 1.5 times higher than the stiffness of the crack formation stage from the experimental results (
Table 9 and
Figure 17). Here, the calculated crack and flexural moments in
Table 4 were used for
Mcr and
Ma, respectively. Meanwhile, the calculation (
Table 9) showed that
Ie and
Icr were similar, and the calculated stiffness of the cracked sections using
Icr in Equation (2) was also shown to be 1.2 times higher than the stiffness of the crack formation stage from the experimental results. Equation (1) was proposed based on GFRP bars with an elastic modulus of 40 GPa and tensile strength of 690 MPa as reinforcement. These differences in physical properties between carbon grids and the GFRP bars are considered to affect the calculation results; the calculation values are higher than the experimental results because of the higher elastic moduli and tensile strengths of the carbon grids.
3.4. Comparison of Flexural Behaviors with RC
The tensile capacity of the carbon grids in the carbon grid-reinforced specimens was 0.6 to 5.0 times higher than that of the steel in specimen RC (
Table 1); however, the maximum load of the carbon grid-reinforced specimens ranged from 60 to 110% of the maximum load of the steel specimen RC (
Figure 18), because specimen RC reached its maximum load owing to the yielding of steel, whereas the carbon grid-reinforced specimens reached their maximum load because of mortar crushing. In this case, only tensile stress corresponding to 38–82% of the tensile strength was observed on the longitudinal strands, as indicated in
Table 7. In the case of specimens J4 and J5 reinforced with cross-laminate structured carbon grids FTG-CR13 and FTG-CR16, which were calculated to have a flexural capacity similar to that of specimen RC, the maximum load ranged from 60 to 79% of the maximum load of specimen RC.
The stiffness of specimens J4 and J5 was significantly lower than that of specimen RC after cracking (
Figure 19) because the elastic modulus of the longitudinal strands was 0.5 times lower than that of steel, whereas the total cross-sectional areas of the strands in the carbon grids arranged in specimens J4 and J5 were similar to or larger than that of steel by 0.91 and 1.40 times, respectively. Consequently, the mid-span deflections of specimens J4 and J5 under the maximum load were 2.00 and 1.35 times higher than that of specimen RC, respectively (
Table 5). Furthermore, in the case of specimen RC, the load gradually decreased even after reaching the maximum load owing to the ductile behavior after the yielding of steel; approximately 81% of the maximum load was maintained at the end of the experiment. However, in the case of specimens J4 and J5, the load sharply decreased to approximately 67 and 61% of the maximum load, respectively, as the load resisted by the mortar was transferred to the carbon grids after reaching the maximum load due to the heterogeneous bond between the mortar and carbon grids [
27]. The load finally decreased to approximately 53 and 36% of the maximum load for specimens J4 and J5, respectively, without carbon grid fracture at the end of the experiment.