The HS-C mixture proportions are presented in Table 4. The design strength of the conventional HS-C was targeted to be approximately 60 MPa at 28 days. The HS-C specimens were separated into two groups. The first group was the conventional concrete composed of river sand and crushed limestone. The second group was the RAC composed of river sand and RCA. GBA was used to substitute OPC at 20 to 50%wt of binder (RA20, RA35, and RA50). The crushed limestone and RCA were made saturated surface dry (SSD) by immersing them in a bucket of water for 24 h before mixing in the concrete. The ratio between effective water to binder of all concretes was fixed at 0.30. Following ASTM C494 (28), commercially available polycarboxylate ether polymers superplasticizers, Type A and F, were employed in the concrete mixtures to control the slump of fresh concrete at values between 150 and 200 mm.

The required superplasticizer content of HS-C is shown in Table 4. Both the CT and RA concretes required 1.5 kg/m³ of superplasticizer to maintain the slump of fresh concrete at values between 150 and 200 mm. When GBA was used to substitute OPC in HS-RAC, a higher superplasticizer dosage was needed in order to maintain the same degree of workability (150–200 mm slump). In the conditions where 20, 35 and 50% of OPC in RAC was replaced by GBA, the required superplasticizer contents were 2.3, 2.8 and 3.4 kg/m³, respectively. The increase of superplasticizer dosage was due to the high porosity and high LOI value of GBA, which led to high water absorption.

This experiment used the two-stage mixing approach proportional-2 (TSMA_p2) method suggested by Tam et al. (14), witch has the following steps. First, all aggregates (both coarse and fine) and half of the binder are mixed for one minute. Second, the rest of the binder is added, and then water is added and mixed for another two minutes. The exclusive TSMA_p2 procedure creates a thin film of cementitious slurry (substances from the cement with a relatively lower water to binder ratio) on the surface texture of the RCA, which is expected to penetrate the porous old mortar and fill the old cracks and voids. Therefore, the weaker link at the interface transition zone (ITZ) of concrete can be improved.

The density results for the high-strength CT concrete and RAC are given in Figure 5. At 90 days, the density of CT concrete was 2450 kg/m³, and the density of RA concrete was 2250 kg/m³; the difference is approximately 8% lower than the of CT concrete when the recycled coarse aggregate was used as a replacement for crushed limestone. The lower density of the RA concrete was due to the old cement paste or mortar attached to the surface texture of the original aggregate, which had a lower SG (2.45) than that of crushed limestone (2.71). Therefore, it was expected that by substituting the crushed limestone with recycled coarse aggregate, the concrete’s density would be lower. The results were similar to those in Safiuddin et al. (34) and Khoshkenari et al. (35).

The partial substitution of GBA for OPC had a slight influence on the density of the RAC. The RA20 and RA35 concretes had densities of 2265 and 2225 kg/m³, respectively, which were close to that of the RA concrete without GBA. However, the use of GBA to substitute 50%wt of binder resulted in the reduced density of 2185 kg/m³ for RA50, which was approximately 3% lower than the density of the RA concrete. It was clearly observed that the high amount of GBA replacement of OPC by weight led to an increase in paste volume, from which the density of GBA was made to be lower than that of OPC (see Table 1), resulting in the lower density of the RA50 concrete.

Figure 6 shows the volumes of permeable pore space (%void) of the CT concrete and recycled aggregate concretes. The %void of the CT and RA concretes were 6.80 and 12.69%, respectively, and it could be observed that the %void of concrete increased when the crushed limestone was replaced with RCA in the concrete. The higher %void of the RA concrete was due to the high porosity of the old mortar or cement paste adhered to the surface of the RCA, which was approximately 8 times higher than that of the crushed limestone.

In case of the RAC containing GBA, the RA20, RA35 and RA50 concretes had %void values of 12.43, 12.70 and 14.16%, respectively. Replacing OPC with 20 and 35% of GBA had a slight effect on the porosity of the RA concrete. However, when 50% GBA was incorporated, the %void of the RA50 concrete further increased. This may be related to the fact that adding a large amount of GBA led to a reduction in the cement content in the concrete, and the Ca(OH)₂ produced by the hydration reaction of the OPC was not sufficient to allow the pozzolanic reaction with the silica and alumina oxides from the GBA. Therefore, the porosity of RA50 concrete increased, and the results are analogous to those of the concrete made with other pozzolanic materials (36, 37).

The relationship between compressive strength and curing age for the CT concrete and RAC is presented in Figure 7. At 7, 28 and 90 days, the compressive strengths of the CT concrete were 57, 62 and 70 MPa, respectively. Meanwhile, the RA concrete had compressive strengths of 58, 63 and 69 MPa, or approximately 101, 102 and 99% of the CT concrete at 7, 28 and 90 days, respectively. The normalized compressive strengths of RAC compared to the CT concrete are given in Figure 8. It can be observed from Figs. 7 and 8 that RAC made with RCA obtained approximately the same strength as the corresponding CT concrete. The slightly greater strength of the RA concrete relative to the CT concrete may be attributed to a stronger bond developed at the ITZ between the old mortar and the original aggregate of RAC (which was obtained from the concrete with the compressive strength of approximately 50 MPa) (38–40); therefore, the strength-controlling link of the RA concrete was the ITZ between the new mortar and the RCA, where the interfacial bond was the weakest phase in the concrete. These results are similar to those in previous studies (41, 42). Moreover, the TSMA_p2 procedure could have improved the weaker link at the ITZ of the RAC, which led to increase in the strength of the concrete (14).

For RAC with GBA at the early ages, an increase in GBA replacement reduced the compressive strengths of those concretes. For instance, at 7 days, the compressive strengths of RA20, RA35 and RA50 concretes were 55, 50.9 and 44.2 MPa, or 96, 89 and 77% of the CT concrete, respectively, whereas the compressive strengths of the concretes at 28 days were 61.3, 56 and 49.9 MPa, or 99, 91 and 81% that of the CT concrete, respectively. At the early ages (before 28 days), the concrete strength due to the hydration reaction (between cement and water) was greater than the strength gained from pozzolanic reaction. When GBA was used as a substitute for OPC in the concrete, the amount of OPC in the mixture was reduced, leading to a reduction in the hydration reaction. However, when the OPC was partially replaced by 20% GBA, there was greater compressive strength than for the CT concrete at the later ages (90 days or more). For example, at 90 days, the RA20 concrete had a compressive strength of 70.8 MPa, or 102% of the CT concrete. The results indicate that BA with elevated fineness is a good reactive pozzolanic material and can be used as a mineral admixture to improve the strength of RAC. On the contrary, RAC mixtures having 35 and 50% replacement levels of GBA (RA35 and RA50 concretes) developed less compressive strength than the CT concretes. The lower strength development of the RA35 and RA50 concretes was attributed to the pozzolanic reaction being produced to a lesser extent due to the lower amount of calcium hydroxide as a consequence of the low content of Portland cement in the concrete (43).

The relationship between the density and the compressive strength of the CT concrete and RAC is presented in Figure 9. At the same compressive strength, the density of CT concrete was higher than that of RAC with and without GBA. For instance, at the compressive strength of 70 MPa, the density of RAC was approximately 2255 kg/m³, which was approximately 8% lower than that of CT concrete (with a density of approximately 2450 kg/m³). In addition, the compressive strength increased as when the density of the RAC increased. A prospective study indicated that the RAC had densities between 2180 to 2280 kg/m³, corresponding to compressive strengths of between 56 and 72 MPa.

Figure 10 presents the relationship between the %void and the compressive strength of the CT concrete and the RAC. In the figure, the data were divided into two groups. The first group (the conventional concrete group) was the concrete using NA, and the data were obtained from Rukzon and Chindaprasirt (24) and plotted, including a comparative analysis with the results of this group. The second group (the RAC group) was the concrete using RCA with and without GBA. The results showed that the RAC with and without GBA had higher %void than that of the conventional concretes at the same compressive strength. For the concrete compressive strengths between 61 and 72 MPa, RAC with and without GBA had a higher %void than the conventional concrete by approximately 39–50%. However, the reduction in the %void of the concrete enhanced the compressive strength of all concrete groups (conventional concrete and RAC with and without GBA). It was noted that there was a direct relationship between %void and the compressive strength of concrete for each group.

A migration test was carried out, and the chloride ion penetration depth was measured by spraying the fresh fracture surfaces of the specimen with an AgNO₃ solution. Each result represents the average of three samples. The depths of chloride ion penetration into CT concrete is presented in Figure 11, while Figure 12 shows the relationship between the depth of chloride ion penetration and the replacement level of GBA for the RA concretes. At 28 days, the depths of chloride ion penetration into CT and RA concretes were 16.3 and 21.9 mm, respectively, and decreased to 14.9 and 20.5 mm, respectively, at 90 days. Apparently, replacing the crushed limestone with RCA had a negative impact on the chloride resistance of the concretes. Crushed limestone presents impermeable inclusion and therefore is an obstacle to chloride ingress (44) (as shown in Figure 13a), while the porosities of the attached old mortar of the RCA facilitate chloride ion penetration (as shown in Figure 13b). This can explain why the reason why chloride ingress in RA concrete is high compared with CT concrete.

The use of GBA as a binder to replace OPC had a clear positive influence on the durability of RAC in terms of the resistance of chloride ion penetration. This result gave the same results with concretes containing NA (24, 26). Considering the results of chloride ion penetration into RAC (as shown in Figures 11 and 12), the depths of chloride ion penetration for the RA20, RA35 and RA50 concretes (GBA replacement levels of 20, 35 and 50%, respectively) were 20, 14.7 and 13.3 mm at the age of 28 days, which was approximately 9, 33 and 40% lower than that of the RA concrete, respectively, and decreased to 14.9, 9.2 and 7.7 mm at the age of 90 days, which was approximately 27, 55 and 62% lower than that of the RA concrete, respectively. This indicates that increasing the GBA replacement levels resulted in an increase of the resistance of RAC due to chloride ion penetration. Moreover, at high GBA content (i.e., 35 and 50% replacement), the depths of chloride ion penetration into RA35 and RA50 concretes were lower than for the CT concrete at a similar age, although the compressive strengths of RA35 and RA50 were lower. This is due to the filler effect, in which the small particles of GBA fill in the space between the cement particles, resulting in a much denser binder. In addition, the pozzolanic products (CSH and CAH) from GBA were responsible for physical binding, by which chloride ions may be physically adsorbed on the surfaces of those products (45, 46). Thus, GBA is considered to inhibit the chloride ions from penetrating deeper into concrete containing RCA.

The effect of GBA content on the percentage decrease in the depth of chloride ion penetration and the percentage increase in the compressive strength of concrete at 28 to 90 days is presented in Figure 14. It can be seen that the compressive strength had a small influence on decreasing the chloride ion penetration into RAC, while the GBA replacement value had a larger influence on reducing the infiltration of chloride ions into concrete. For example, the increase in compressive strength for the RA concrete was 9%, and the decrease in the depth of chloride ion penetration was 6%. Replacing OPC with 20, 35 and 50% (by weight of binder) of GBA yielded increases in compressive strength of 16, 15 and 14%, respectively, while the decreases in the depth of chloride ion penetration were 25, 38 and 42%, respectively. This confirmed the result regarding the improvement of durability in RAC with respect of chloride resistance, demonstrating that replacing OPC with GBA was more effective than the increase in the compressive strength.

The chloride migration coefficient (D_nssm) after the migration test was calculated using equation [3] as per NT Build 492 (31), for which the applied constant electrical potentials of 60 V with 18 h were used. The results of the chloride migration coefficient are demonstrated in Figure 15. The chloride migration coefficients at 28 days for the CT and RA concretes were 5.45x10^–12 and 7.29x10^–12 m²/s, respectively. At the age of 90 days, the CT and RA concretes had chloride migration coefficients of 5.01x10^–12 and 6.91x10^–12 m²/s, respectively. It can be seen that the low quality of the RCA, which had a higher porosity than crushed limestone (also see Table. 3), resulted in a low resistance to chloride ion penetration into RAC. This result is consistent with the findings of Ying et al. (47), who reported that the attached old mortar content had a great influence on chloride ion diffusion into RAC. As expected, the RA concrete had a higher chloride migration coefficient than the CT concrete at the same testing ages.

However, RAC with GBA showed good resistance to chloride ion penetration compared to conventional concrete using crushed limestone and an increased chloride resistance for RAC with increased GBA replacement. Moreover, the chloride migration coefficients of the RA concretes containing GBA as a cement replacement of 35 and 50% were lower than that of CT concrete at similar ages. The chloride migration coefficient at 28 days for the RA20, RA35 and RA50 concretes were 6.76x10^–12, 4.93x10^–12 and 4.54x10^–12 m²/s, respectively, and they decreased to 5.04x10^–12, 3.12x10^–12 and 2.61x10^–12 m²/s, respectively, at the age of 90 days. Similar results had been reported for studies on the non-steady-state migration coefficients of concrete (48), where the use of fly ash up to 20% of OPC replacement was found to have lower chloride migration coefficients at later ages and at higher cement replacement levels. These results confirm that the pozzolanic reaction products provide the RA concretes containing GBA with protection against chloride ion penetration.

Figure 16 present the results of the compressive strength and chloride migration coefficients of concrete. The compressive strength of concrete affected the chloride ion penetration resistance, especially the RAC with GBA. For instance, the RA and RA20 concretes at 28 days had compressive strengths of 63 and 61 MPa, and the chloride migration coefficients were 7.29x10^–12 and 6.76x10^–12 m²/s, respectively, whereas at later age (90 days), the RA and RA20 concretes had compressive strengths of 69 and 71 MPa, and the chloride migration coefficients were 6.91x10^–12 and 5.04x10^–12 m²/s, respectively. Nevertheless, the GBA replacement levels in RAC had a great influenced on the chloride ion resistance compared to the compressive strength. For instance, the compressive strength of the RA concrete at 90 days was 69 MPa and the chloride migration coefficient was 6.91x10^–12 m²/s, whereas the RA20, RA35 and RA50 concretes had compressive strengths of 71, 64 and 57 MPa and the chloride migration coefficients were 5.04x10^–12, 3.12x10^–12 and 2.61x10^–12 m²/s, respectively.