Effect of partially replacing crushed oyster shell as fine aggregate on the shear behavior of short RC beams using GFRP rebar strengthened with TRC: Experimental and numerical studies

2.1 Test specimens

In this research, nine concrete beams were fabricated to examine the shear performance of beams that were reinforced with GFRP bars and enhanced with carbon TRC. The beams were divided into three groups according to the proportion of COSs that partially replaced river sand in the concrete mix: 0, 20, and 40%. For each concrete mix, three beams were cast. Table 1 describes the level of COS replacement and the strengthening applied to each beam.

The first beam in each group served as a control specimen and was not strengthened with TRC. The second and third beams were upgraded with two and three carbon textile layers. All beams were constructed with uniform dimensions, measuring 150 mm in width, 250 mm in height, and 900 mm in length. The span for the three-point bending tests was established at 600 mm to create a shear-critical region and accurately assess the shear performance of the beams.

Longitudinal reinforcement in each beam was provided using GFRP bars, with three Ø16 bars placed in the bottom layer and two Ø16 bars in the top layer (Figure 1). Transverse reinforcement was provided by Ø6 GFRP stirrups, positioned at 100 mm intervals along the beam’s length. This uniform reinforcement layout ensured that the only variables influencing the results were the proportion of COSs and the presence or absence of TRC strengthening. Two TRC configurations were tested: one with two layers and another with three layers of carbon textiles, applied to the two side faces and the bottom face of the beams. This setup was designed to assess the impact of textile layering on shear performance, especially in beams where COSs partially replaced natural sand. The TRC system was intended to offset any potential strength reductions from the use of alternative aggregates, thereby enhancing the overall shear resistance of the beams.

Figure 1

Details of test beams.

2.2 Identification of materials

2.2.1 OSC

The oyster shells dedicated to this study are Pacific Oyster. The aggregates, including natural sand, COSs, and 5 mm × 10 mm gravel, are washed and dried before concrete pouring. The primary source of fine aggregates is natural sand extracted from the Song Da River, with a fineness modulus (Mk) of 2.5 and a density of 2.67 g/cm³. Besides, the oyster shells are crushed so that the largest particles pass through a 4.75 mm sieve, as illustrated in Figure 2. The coarse aggregate, 5 mm × 10 mm gravel, was selected with a maximum grain size smaller than 12.5 mm. The grain composition of both fine and coarse aggregates meets the ASTM C33 [35] recommendations and is shown in Figure 3.

Figure 2

Extraction of Pacific Oyster shell at different scales. (a) Oyster shells are washed and dried before grinding. (b) Grain sizes from 0.15 to 4.75 mm of COSs.

Figure 3

The size distribution curve conforms to the requirement of ASTM C33. (a) Fine aggregate. (b) Coarse aggregate.

In this case, the chemical composition of oyster shells was analyzed using the Auto Quantitative Spectrum (AQS) method using an X-ray diffraction (XRD) spectrometer, and the experimental results are presented in Figure 4. Concretely, the percentage of CaCO₃ content is mainstream, accounting for 73.34%. Meanwhile, other compounds such as Richterite Na₂CaMg₅(Si₈O₂₂)(OH)₂ and Leucite low KAlSi2O6 occupy around 5.73 and 4.07%, respectively. The loss in calcination content of oyster shells is estimated to be less than 16.86%. This analysis confirms the optimal replacement for fine aggregate by a COS in the scope of sustainable development.

Figure 4

X-ray diffraction spectrum of Pacific Oyster shell specimens.

In this study, the fine aggregate component was partially substituted with COSs at replacement levels of 0, 20, and 40%. In contrast, other materials, such as cement, fly ash, and coarse aggregate, remained constant. Due to the higher water absorption capacity of oyster shells compared to natural sand, the water content added to the mixtures containing partial replacement of oyster shells is greater than that of the control mixtures.

This approach enables the evaluation of the effects of COS inclusion on the mechanical properties of concrete when used as a partial replacement for fine aggregate. Specifically, three concrete mixtures were prepared, each consisting of cement, fine aggregate, coarse aggregate, fly ash, and a superplasticizer admixture, with a water-to-binder ratio of 0.36. The traditional concrete mixture (0% COS) was designed to achieve a target strength of C50, based on ACI 211.4R-08 [36], and further refined through experimental adjustments. C50 is the compressive strength that meets the requirements of typical construction works. The detailed composition of the concrete mixtures is provided in Table 2.

Table 2

Composition of concrete mixtures per cubic meter

Materials	Units	Traditional concrete	20% OSC	40% OSC
PC40 fine cement	kg	454.5	454.5	454.5
Natural sand dried	kg	635.7	508.5	381.4
5 × 10 gravel dried	kg	1040	1040	1040
COSs dried	kg	0	50.5	101.0
Water	l	208.9	212.6	216.2
Fly ash (F)	kg	80.2	80.2	80.2
Superplasticizer admixture	l	5.6	5.6	5.6

2.3 Fabrication of the specimen procedure

The fabrication of the specimens was conducted with careful precision to ensure uniformity and reliability across all experimental results. All three types of concrete were mixed in a drum mixer until a homogeneous consistency was achieved, after which it was poured into formwork to cast the beams. Prior to pouring the concrete, the GFRP reinforcing cages were assembled outside the formwork and inserted afterward after attaching strain gauges to the respective rebars and stirrups (Figure 5). After casting, the beams were cured in water for 28 days under controlled laboratory conditions.

Figure 5

Fabrication of RC beams with COS aggregates.

Upon completing the fabrication process, two beams from each group were wrapped with either two or three layers of carbon textiles. In contrast, the remaining beams were kept unstrengthened as control specimens. The TRC strengthening procedure was performed in accordance with the guidelines outlined in ACI 549.4R-20 [40]. Preparation of the substrate, including the removal of corrosion debris and roughening of the surface, was conducted to ensure optimal bonding between the TRC and concrete substrate. A 5 mm-thick layer of fine-grained concrete was first applied to the beam surfaces using a metal trowel. The carbon textile was then embedded into this layer, ensuring full integration into the concrete matrix. Subsequent layers of fine-grained concrete were applied to completely cover the textile, with this process repeated until the specified number of textile layers was achieved, as illustrated in Figure 6.

Figure 6

Application of TRC strengthening layers on RC beams.

2.4 Test setup and measurement

For the three-point bending test arrangement, each beam was placed on two supports, spaced 600 mm apart, with the load applied at the midpoint to induce shear forces in the beam's critical region (Figure 7). The two supports are cylindrical with a diameter of 80 mm, ensuring that the beam can freely rotate at the contact points. The loading frame and load application were controlled by the displacement method, in which the loading rate was maintained at 0.5 mm/min, and the test was conducted until failure. Deflections were recorded using a Linear Variable Differential Transformer (LVDT) positioned at the mid-span of the beam. Additionally, strain gauges were attached to the GFRP rebar and stirrups to monitor the strain distribution during loading. This arrangement allowed for a comprehensive analysis of the load–deflection behavior of the beams, as well as the effectiveness of the TRC strengthening in improving shear capacity.

Figure 7

Test setup.

Simultaneously, several cylindrical concrete specimens measuring 150 mm × 300 mm were cast and cured in water at a controlled temperature of 22°C for 28 days. These specimens were used to assess various mechanical properties, including compressive strength and static modulus of elasticity. These tests conform to specifications ASTM C39 [41] and ASTM C469, respectively [42]. Moreover, to assess the flexural tensile strength, four-point bending tests were carried out on rectangular specimens with dimensions of 100 mm × 100 mm × 400 mm. The testing procedure adhered to the ASTM C78 guidelines [43], as outlined in Table 4. The characteristics of different mixtures are summarized in Table 4 and Figure 8.

Figure 8

The perspective of experimental work for determining the mechanical properties of traditional concrete and OSC.

3.1 Finite element analysis and boundary conditions

In this study, nine 3D finite element models were constructed using Abaqus/Explicit [44] to replicate the experimental tests carried out on the specimens, as described in Section 2. The concrete and steel plate domains were discretized using the C3D8R from the Abaqus material library, which is a fully integrated 3D brick element with eight linear nodes. Each element had dimensions of 15 mm × 15 mm × 15 mm. Depending on the type of specimen, the overall quantity of elements varies between 9,600 and 12,960. For the reinforcement, T3D2 truss elements were employed to represent the behavior of the GFRP bars. Against the boundary condition, the models were established respecting the three-point bending principle with the applying force controlled by displacement at the middle and two supporting pins, which could rotate around its symmetrical axis. For the concrete–TRC interface, a tie constraint was applied to model the interaction between the two materials. Compared to the flexural test, the shear test gives a lower deflection at the midspan of beams, leading to the more efficient strengthening of TRC (Figure 9). Thus, the assumption that all materials have a perfect bond to another one is chosen.

Figure 9

Mesh discretization and boundary conditions for the modeling of RC beam with/without strengthening.

3.2 Input parameters for the models

The input parameters for these models were primarily based on those established by Vervloet et al. [45]. To model the behavior of concrete, the well-known concrete damage plasticity (CDP) model was utilized. This model, initially introduced by Hognestad [46], has been further developed by various researchers, including Nguyen et al., Lubliner et al., and Lee and Fenves [47,48,49]. CDP is a non-associated plasticity model, wherein the plastic potential ( Φ ) is a Drucker–Prager hyperbolic function. This function combines with a distinct function of yield surface ( F ), according to Ayhan et al. [50], which controls the stress–strain relationships, as depicted in Figures 11 and 12:

Here, Ψ is the dilatation angle taken against the deviatoric stress plane; ϵ is the eccentricity parameter; σ t 0 is the uniaxial tensile stress at failure; q ̅ is the hydrostatic pressure; and p ̅ is the von Mises-equivalent effective stress, defined by Eqs. (2) and (3), respectively:

In Eq. (2), s denotes the deviatoric part of the effective stress tensor:

In Eq. (4), σ ̅ max corresponds to the maximal principal effective stress; ε pl is equivalent plastic strain; and α , β , and γ are calculated as follows:

where σ b 0 and σ c 0 are initial equibiaxial compressive yield stress and initial uniaxial compressive yield stress, respectively; and K c is the ratio of the second stress invariant on the tensile meridian to the compressive meridian at the initial yield.

This ratio affects the shape of the yield surface on the deviatoric plane. Figure 10 illustrates the contribution of the chosen K c values to evolution of yield surfaces in the deviatoric cross-section (refer to Kazemi et al. [51]).

Figure 10

Effect of different K c values to yield surfaces in the deviatoric plane under the Drucker–Prager yield criterion.

As shown in Figures 11 and 12, inelastic compression strain ε c in and tensile cracking strain ε t ck are obtained by employing Eqs. (8) and (9), in which ε oc el and ε ot el refer to the compressive elastic strain and the tensile elastic strain, respectively, determined by Eqs. (10) and (11):

Figure 11

Stress–strain curve and concerning parameters of concrete under uniaxial compression of the CDP model.

Figure 12

Stress–strain curve and concerning parameters of concrete under uniaxial tension of the CDP model.

The elastic stiffness of concrete is usually degraded under applied load compared to the initial stiffness. This diminution is characterized by the tensile and compressive damage parameters, d t and d c . According to Tawfik et al. [52], these coefficients are as follows:

The plastic strain under compression or tension can be determined using Eqs. (16) and (17):

Especially in Abaqus, the compressive damage parameter d c (DAMAGEC) and the tensile damage parameter d t (DAMAGET) under uniaxial loading are also the function of other variables, in which f tm and f cm depend on the tensile or compressive strengths of concrete, G f and G ch represent fracture and crushing energies per unit area and l eq reflect the geometric element properties of the mesh, like in Antoniou et al. [53]:

In the current study, the compressive strength and modulus of elasticity values were derived from experimental measurements, while the tensile strength values were calculated from those of flexural tensile strength, following the guidance of ACI224.2R-92 [54]. The modulus of rupture ( f r ) and direct tensile strength ( f t ) are determined using Eqs. (20) and (21), respectively:

where w c is the unit weight of concrete (kg/m³), f c ′ is the compressive strength of concrete (MPa), g r = 0.012 – 0.021 , and g t = 0.0069.

Therefore, the direct tensile strength can be taken between 33 and 58% of the flexural tensile strength. These values serve to establish the stress–strain diagram proposed by Zheng et al. [55], validated for ordinary concrete, as presented in Figure 13. Against this model, the values at the peak of ultimate strengths generating the shape of stress–strain evolution in both ascendant and descendent branches could be defined experimentally, as reported in Table 5. The tensile strain at peak ( ε t 0 ) is calculated by employing Eq. (22); meanwhile, the compressive strain at peak ( ε c 0 ) is formulated based on the modified Hognestad model for the compressive behavior of concrete (refer to Madandoust et al. [56]):

Figure 13

Stress–strain diagram of the CDP constitutive model for concrete.

The overall mechanical properties of other materials used for the model are outlined as follows (Figure 14).

Figure 14

Stress–strain responses of steel and reinforcement (GFRP/textile).

Among the multi-plasticity parameters of the CDP model described by Vervloet et al. [45], Tawfik et al. [52], and Kazemi et al. [51], the first one is the dilatation angle ψ , representing the mechanic behavior of concrete under confined stress. The recommended values range from 30° to 45° (refer to Shawki Ali et al. [57]).

The second one is flow potential eccentricity ϵ , reflecting the ratio between the tensile strength and compression strength, and the recommended value is 0.1. The third parameter is the biaxial stress ratio, σ b 0 / σ c 0 , which ranges always taken from 1.00 to 1.16. The fourth parameter is the shape factor K c , and its default values vary from 0.667 to 1.0. The last parameter is viscosity λ , which allows the enhancement of the rate of solution convergence in certain sufficiently small problem steps with chosen values from 0.0001 to 0.008 [57,58]. The Poisson’s ratio of concrete is usually taken from 0.15 to 0.25, according to Ahmed [59] or Pal [60], and its value increases when the compressive strength of concrete decreases. The details concerning the mechanical parameters of using materials in the models are displayed in Table 5.

4.1 Bearing mechanism and load–deflection relationship

First, the shear performance of the nine beams was systematically evaluated through three-point bending tests. The results provided valuable insights into the influence of crushed OSC, the effectiveness of TRC strengthening, and the structural behavior of the investigated beams.

During testing, distinct failure modes were observed, with notable differences between the control and TRC-strengthened beams. The control beams (unstrengthened) exhibited a brittle shear failure, characterized by the rapid expansion of diagonal shear cracks, typically initiating near the support and propagating toward the load application point (Figure 15). This led to a sudden and catastrophic failure once the critical shear stress was exceeded. In contrast, the TRC-strengthened beams demonstrated a more ductile failure mode. The TRC layers effectively delayed shear cracking, with beams reinforced with two layers of carbon textiles showing smaller, more distributed cracks. Beams with three TRC layers exhibited even finer crack patterns, indicating a higher degree of crack control and stress redistribution within the beams, thus preventing the sudden shear failures observed in the control specimens.

Figure 15

Crack patterns of all test beams.

Load–deflection curves for each beam group were plotted to analyze the structural behavior under increasing loads. Overall, regarding the load–deflection curves depicted in Figures 16–18, a good agreement exists between the numerical and experimental results. The response of load–deflection of all beams is approximately linear up to the first crack formation. Subsequently, the curves turned non-linear as a result of stiffness deterioration corresponding to the deflection hardening period. The deterioration develops with the load increasing until the ultimate load is obtained. After that, there was a decrease in load until the fracture point, often named the deflection softening period. The deflection softening period seems to be more important, accompanying a larger deflection value for high replacing content of crushed oyster shell cases. This explains why more ductile failure modes appeared in G3 specimens. Also, the addition of textile layers was the most effective for G3 in terms of improving bearing capacity and stiffness. The ductility index (µ) of the specimens represents the ability of structural elements to endure inelastic deformation while maintaining their load-carrying capacity before failure. It is evaluated by the ratio of the mid-span deflection at ultimate load (Δu) to the mid-span deflection at the yield limit (Δy) (refer to Maras et al. [14]). The increase of the ductility index values depicted in Table 6 indicates that the addition of TRC layers enhances the structural ductility:

Figure 16

Load vs deflection of the three beams in the G1 group.

Figure 17

Load vs deflection of the three beams in the G2 group.

Figure 18

Load vs deflection of the three beams in the G3 group.

The load–deflection responses show that the control beam TCB-R exhibits the highest stiffness and load capacity, peaking at over 217 kN with a deflection of around 2.5 mm. The beams with 20 and 40% COS replacements show reduced stiffness and load capacity, with the 20% COS beam reaching about 186 kN and the 40% COS beam just over 168 kN. The overall strength of the concrete decrease refers to the intrusion of COS, with higher COS content leading to greater reduction. The control beam TCB-R exhibits the highest stiffness, followed by the 20OSB-R beam and the 40OSB-R beam. This trend indicates that increasing the percentage of COSs reduces the stiffness of the beam. However, the 40% COS beams display enhanced ductility, as reflected in their more gradual decline in load capacity after reaching a peak load, compared to the sudden drop seen in the control beam.

The load–deflection curves in Figure 17 compare the performance of beams with 20% COS content, including an unstrengthened beam and two other reinforced beams, respectively, with two and three layers of TRC. The unstrengthened beam has the lowest load capacity, peaking at around 186 kN with a deflection of 1.9 mm, indicating limited stiffness and strength. The beam reinforced with two TRC layers shows significant improvement, reaching approximately 266 kN at a 3 mm deflection, with much higher initial stiffness and sustained load capacity over a greater deflection range. The addition of TRC layers greatly enhances the structural performance of the beams, with three layers offering the most substantial improvement, which is around 283 kN of ultimate load at a deflection of 2.9 mm. The strengthened beams exhibit significantly higher load capacity and stiffness, demonstrating the reduced strength caused by the partial replacement of aggregates with COS was compensated by the effectiveness of the TRC layer.

The load–deflection curves in Figure 18 illustrate the performance differences among beams with 40% COS content, comparing the two beams strengthened with either two or three layers of carbon textile to one unstrengthened beam. The unstrengthened beam 40OSB-R reaches a peak load of approximately 168 kN with a deflection of around 2.2 mm, showing the lowest stiffness and strength among the three, as indicated by its gradual load increase with deflection. In contrast, the beam reinforced with two carbon layers exhibits a much steeper curve, indicating improved stiffness and reaching a peak load of about 235 kN, corresponding to a deflection of 2.5 mm. The three-layer TRC beam performs the best, peaking at around 267 kN at a deflection of 2.3 mm, demonstrating superior stiffness and load-bearing capacity. The TRC reinforcement clearly enhances the structural integrity of the beams, with the three-layer TRC beam offering the highest performance. Even the two-layer TRC beam shows significant improvement over the unstrengthened beam, underscoring the effectiveness of TRC in compensating for the reduced strength associated with high COS content.

The overall shear capacity of beams includes a part of the GFRP rebar, the other part of concrete, and the contribution of the TRC material. Using shear prediction from four specifications such as JSCE 1997, CSA S806-12, ACI 440.1R-15, and BS-8110 [61,62,63,64], Tu et al. [65] found that compared to conventional concrete, OSC provides a lower nominal shear resistance. These predicted results are lower than the shear provision obtained from experiments on RC beams with GFRP rebar, ranging from 28 to 49%. As shown in Table 6, ACI 440.1R was selected for the shear provision of both reinforcement and concrete. Otherwise, the application of TRC in the shear strengthening of RC structures has been extensively studied over the past decade. However, no design codes currently exist for such structures (refer to Tran et al. [31]). The shear strength of the TRC layer can be calculated using the guidelines for FRP strengthening layers, as defined in ACI 440.2R [66].

The specimens strengthened with two layers of carbon textiles showed a significant amelioration in the shear capacity and deflection. The shear capacity of the 0% oyster shell beam increased by approximately 44% (from 108 to 156 kN), while the 20 and 40% oyster shell beams saw improvements of 43% (from 93 to 134 kN) and 39% (from 84 to 118 kN). The addition of a third TRC layer resulted in the highest observed shear capacities. The shear capacity of the 0% oyster shell beam increased by 49% (from 108 to 162 kN), while those with 20 and 40% OSC showed improvements of 52% (from 93 to 142 kN) and 58% (from 84 to 134 kN). An adjustment of the COS content clearly ameliorated the ductility index of specimens, interpreted by a more dominant deflection softening zone on G3 specimens against load–deflection responses.

These findings seem to be in good agreement with the experimental results by Tran et al. [31], where an improvement in shear resistance by TRC strengthening was observed, ranging from 36.2 to 66.4% (a/d = 2.3) and from 33.9 to 83.1% (a/d = 2.8) with one and two layers of textile, respectively. The peaks of load–deflection relationship allows the prediction of the shear strength of beams depending on the replacing ratio of oyster shell and other mechanical properties with different strengthening methods, as presented in Figures 19–22.

Figure 19

Shear strength vs replacing COS (in %).

Figure 20

Shear strength vs modulus of elasticity corresponding to different COS contents.

Figure 21

Shear strength vs compressive strength corresponding to different COS contents.

Figure 22

Shear strength vs flexural tensile strength corresponding to different COS contents.

The partial replacement of COSs for natural sand in the concrete mix had a noticeable impact on both stiffness and shear capacity. As the percentage of COSs increased, the initial stiffness decreased across all beam groups. This reduction in stiffness was reflected in the load–deflection curves in the initial portion, which became less steep with higher OSC, indicating a diminution in the elastic modulus of the concrete. Despite this diminution, the application of TRC significantly mitigated the adverse effects. Figure 23 displays the stress distribution of the GFRP rebar at the ultimate states of all beams tested, which gives the maximal stress values compared to the results obtained by the experiment, as presented in Table 7. Due to the hypothesis of the ideal bond between the different materials, the experimental values of maximal stress on the GFRP rebar slightly exceeded those of the modeling. This means that the bearing capacity of both the master beam and TRC are overestimated numerically.

Figure 23

Stress distribution on the GFRP rebar at ultimate states of nine study cases.

In general, textile layers proved their efficiency in strengthening, above all, toward three textile layers. Specifically, the maximal stress at ultimate states reveals significant differences in performance based on the level of TRC reinforcement, reflected by a more modest stress distribution on the GFRP rebar. The load–strain curves displayed in Figures 24 and 25 illustrate the effect of TRC strengthening on both stirrup and bottom reinforcement. At the same applied load, lower strain values were observed in the strengthened beams. The overall strain of the longitudinal rebar is significantly lower in the case of TRC-strengthened specimens.

Figure 24

Load–strain relationships of GFRP stirrup in nine study cases.

Figure 25

Load–strain relationships of GFRP longitudinal rebar in nine study cases.

Considering the bottom reinforcement and stirrup, the un-strengthened beams indicate early and significant stress, reflecting lower stiffness and load distribution capacity. Otherwise, the beams with two TRC layers show improved load capacity, with reinforcement stress remaining more controlled. The three-layer TRC beam performs the best, reaching a gradual increase in stirrup stress, demonstrating superior stiffness and better load management. TRC reinforcement clearly enhances both load capacity and stress control, with three layers providing the most effective improvement.

The numerical results in stress distribution proved that during the test, the fracture of GFRP rebar did not occur; meanwhile, several positions in the textile around the two supports and loading point were broken, as shown in Figure 26.

Figure 26

Stress distribution on the textile in the cases of strengthening by two and three textile layers.

4.2 Influence of crushed OSC on shear performance of beams without strengthening

Indeed, different failure modes could be identified through structural responses. The efficacy and reliability of the proposed model were verified by comparing it with test results corresponding to each case, as shown in Figure 27. As previously discussed, the control beam TCB-R demonstrated the shear failure method owing to the transmission of fine shear cracks inside the shear zone across the diagonal direction, connecting two supports and the loading point with the crushing of concrete in the compression zone. During the loading process, the fine cracks were extended and propagated intensively in the tensile zone.

Figure 27

Comparison of failure pattern given by modeling and experiment of beams without strengthening. (a) Beam TCB-R. (b) Beam 20OSB-R. (c) Beam 40OSB-R.

Against beam 20OSB-R, intensive cracks were identically notified in the tension zone, but concrete crushing was not observed. It seems that the COS content increase reduced the shear performance. Consequently, the failure pattern is less brittle in the ultimate state. In the case of 40% COS content, due to a weaker ultimate load, the shear cracks appeared sparsely, with a diagonal crack connecting the loading point and one of the supporting pins at the end of the test. This observation was confirmed by both numerical and experimental results.

4.3 Influence of TRC thickness strengthening on shear behavior of strengthened beams

As mentioned above, under strengthening by multiple TRC layers, several properties of beams were ameliorated, such as the stiffness or shear performance, leading to a different failure pattern. Figures 28 and 29 compare the experimental and numerical failure patterns, respectively, in the case of two and three textile layer strengthening. A parabolic damage zone of fine grain concrete relying loading point and two supports at two lateral faces of specimens was alerted in strengthened beams with two textile layers. Likewise, the damage zone is more inflated for the strengthened beams with three textile layers, corresponding to a higher collapse level of specimens. In several specimens belonging to G1 or G2, for example, there exists many local crushing of concrete around two supports. This occurrence justified that there is a very good match between the experiment and modeling. Besides, neither concrete crushing is noted for G3 specimens, proving an amelioration in the ductility of materials.

Figure 28

Comparison of failure pattern given by modeling and experiment of beams strengthened by two textiles layers. (a) Beam TCB-2T90. (b) Beam 20OSB-2T90. (c) Beam 40OSB-2T90.

Figure 29

Comparison of failure pattern given by modeling and experiment of beams strengthened by three textiles layers. (a) Beam TCB-3T90. (b) Beam 20OSB-3T90. (c) Beam 40OSB-3T90.

Furthermore, Figure 30 assesses numerically the collapse method of all master beams strengthened by TRC. This analysis brought a vision that was unavailable to be achieved by experimental observation. At the same replacing ratio of COS, the addition of textile layers enhances the stiffness and the ultimate load, yielding more apparition of concrete crushing in the compression zone and less flexural cracks in the tension zone. Indeed, more diagonal cracks were observed due to shear failure at the two lateral faces of specimens. On the other hand, at the same level of textile layer strengthening, as the ultimate loads and stiffness decline with the adjustment of COS, the damage level of specimens considered also seems vaguer as well.

Figure 30

Failure pattern, given by modeling of six master beams strengthened by TRC.