Alkali-silica reaction (ASR) is a reaction between the alkalis from the cement paste (i.e., Na⁺, K⁺,OH^-) and uncrystallized silica within the aggregates. ASR is one of the leading causes of early concrete deterioration in Canada and worldwide (9). Concrete expansion levels are generally used to reflect ASR-induced damage. Figure 1 shows a qualitative crack propagation model in ASR-affected conventional concrete (CC), presenting A) sharp cracks and B) onion skin type cracks (10). Sharp cracks are generally initiated within the aggregate particles at a slightly damaged degree (i.e., 0.05% expansion level). As the reaction continues to a moderate expansion level (i.e., 0.12%), the previously generated sharp cracks propagate towards the aggregate’s edges. At high expansion levels (i.e., 0.20%), the sharp cracks begin to extend into the cement paste. At very high expansion levels (i.e., 0.30%), the sharp cracks formed in the aggregates and the cement paste form an extensive cracking network and decrease the physical integrity of the material. Meanwhile, onion skin type cracks follow the aggregate’s edge, elongating with increasing expansion until entering the cement paste. Mechanical properties (i.e., compressive strength tensile strength, and modulus of elasticity) of concrete affected by ASR are therefore reduced as a function of the expansion level, regardless of the aggregate type (i.e., coarse vs. fine aggregates) and nature (i.e., mineralogy) (10, 11). De Souza et al. (12) observed that the direct shear strength of ASR-affected specimens reduces as a function of ASR expansion, observed even at low expansion levels, which was attributed to the adverse effect of ASR on the aggregate interlock of the concrete specimens.

Recycled concrete aggregate (RCA) is a multi-phase material composed of original virgin aggregate (OVA) and residual mortar (RM). RCA is often considered a lower quality material (i.e., presence of residual mortar - RM, impurities, deterioration, and high variability, among many others), resulting in lower concrete performance. Several mix-design procedures have been developed to increase the overall performance of recycled concrete. Among these mix-design techniques, the Equivalent Volume (EV) method (5) improves the hardened and fresh state properties and results in a more sustainable concrete mixture. The EV method is based on the mix-design technique developed by Fathifazl et al. (3) and its modified version (4). The EV mix-design technique aims to proportion recycled concrete mixtures with the same volume of aggregates (coarse and fine) and cement paste as a companion CC. Indeed, there are concerns regarding the use of RCA due to the presence of past deterioration such as ASR (12, 13), which can be influenced by the RCA production (i.e., aggregate crushing, washing, and storage) and the remaining reactivity potential (14-18). Trottier et al. (6, 7) found that the kinetics and deterioration process (i.e., crack width, crack length, and propagation path) are affected by the severity of initial damage and its location (i.e., OVA or RM). However, the influence of this type of aggregate on the aggregate interlock, determined through the direct shear test, remains unknown. Techniques used to assess ASR damage in concrete.

To fabricate the control CC specimens (i.e., 100 mm by 200 mm, 35 MPa concrete cylinders) and concrete specimens to be used to produce the RCA, a highly reactive coarse aggregate (i.e., Springhill - Greywacke) was combined with natural non-reactive sand sourced locally in Ottawa. All concrete specimens were mix-proportioned following the concrete prism test (CPT) as per ASTM C1293 (28); CSA general use (ASTM type 1) Portland cement was used in the mixture. To accelerate ASR development, the total alkali content of the mixture was raised to 1.25% Na₂O_eq. by cement mass after adding reagent grade NaOH. The specimens were moisture-cured in 100% relative humidity (RH) and 20°C, and then de-moulded after 24 hours. Holes of 5 mm in diameter by 15 mm in depth were then drilled in both ends of the specimens, and stainless-steel gauge studs were glued using a fast-setting cement paste slurry on both ends for the length change measurements. The specimens were left to moist-cure for an additional 24 hours before the initial reading. The specimens were then stored over a film of water in 22-litre plastic containers lined with an absorbent cloth and then placed in an environmental chamber (i.e., 100% RH and 38^oC). The abovementioned absorbent cloth was installed on the lid and inside the buckets to prevent the formation of droplets and minimize the effect of leaching. Prior to the periodic monitoring of the length change measurements, the containers were removed from the environmental chamber and cooled down to 20°C for 16 ± 4 hrs. The specimens were taken out one by one during the measurement period, and buckets were returned to the environmental chamber immediately after each measurement. Two levels of expansion representing distinct damage degrees were selected to produce the RCA: 0.05% and 0.30% (i.e., slight and severe, respectively). The specimens were then jaw crushed to produce reactive coarse RCA ranging from 4.75 mm to 19 mm in size. After crushing, the RCAs were stored in sealed buckets in conditions preventing further ASR development (12^oC). The RM content of slightly damaged RCA and severely damaged RCA were 46% and 51.5%, respectively. The RM content was determined following the procedure proposed by Abbas et al. (29), which involves five cycles of freezing and thawing. All recycled mixtures were proportioned using the EV method (5) while keeping the total cement content of the system equal to 420 kg/m³ (i.e., using the CPT mixture proportions as the companion mixture) as presented in Table 2. To raise the alkali content, the recycled mixtures also incorporated reagent grade NaOH and were cast, cured, prepared, stored, and monitored over time, following the same procedures as aforementioned for the CC. It is worth noting that to assess the effect of initial damage on the future performance of the RCA concrete, only the alkali contribution from the new cement paste was considered. The total alkalis of the system was kept constant, and the stage of the initial reaction was not considered in the total alkali content calculation.

Figure 3 illustrates ASR-induced average expansion levels as a function of time. A standard deviation of 0.01%-0.04%, 0.02%-0.04% and 0.02%-0.03% was obtained for the CC and recycled mixtures made of slightly damaged and severely damaged RCA, respectively. Expansion measurements were carried out for 150 days on the CC at which an expansion level of 0.31% was reached, whereas after 188 days, levels of expansion of 0.24% and 0.28% were obtained for the slightly and severely damaged recycled mixutres, respectively. A similar trend is observed at the beginning from 0-0.03% of expansion (i.e., 29 days), after which both recycled mixtures increase in expansion up to 0.09% and 0.06% at 51 days for the severely and slightly damaged RCA, respectively, while the CC attains only 0.05% at 52 days. After approximately 63 days, the CC swelled at a faster rate when compared to both recycled mixtures. The severely and slightly damaged recycled mixtures, as well as the CC, reached an expansion level of 0.05% at 36, 44 and 49 days, and an expansion level of 0.12% at 79, 94 and 75 days, respectively. Meanwhile, an expansion of 0.20% was observed after 94, 128, 157 days for the CC, severely and slightly damaged recycled mixtures, respectively.

The direct shear strength and reductions at distinct expansion levels of the CC and recycled concrete mixtures are presented in Figures 4a and 4b, respectively. Results indicate that the direct shear strength of all mixtures lessens as a function of induced expansion and ASR development. The CC’s direct shear strength begins to decrease only after 0.05% of expansion (i.e., initially at 8.81 MPa) which can be attributed to the cracks being present within the aggregates as presented in Figure 1; thus, cracks propagating through the interfacial transition zone (ITZ) during shear failure. Hereafter, the decrease in direct shear strength is linear for the CC, reaching 5.82 MPa at 0.20% of the expansion. The recycled mixtures on the other hand, present a lower initial shear strength when compared to the CC. The slightly damaged RCA displayed an initial direct shear strength of 7.2 MPa. As expansion advanced, at low and moderate expansion levels (i.e., 0.05% and 0.12%, respectively), the direct shear strength reduced to 5.81 MPa and 5.44 MPa (i.e., losses of 20% and 24%, respectively). At an expansion level of 0.20%, a significant loss is observed (i.e., loss of 44%), reducing the direct shear strength to 4.02 MPa. Likewise, the severely damaged RCA with an initial direct shear strength of 6.8 MPa presented a similar trend through the low and moderate expansion levels (i.e., losses of 20% and 29%, respectively). Interestingly, no significant loss was observed at a high expansion level (0.20%), unlike the slightly damaged RCA (i.e., from 29% to 34%).

The result from another study by Trottier et al. (6) with the same RCA particles and mixture proportions, was used to complement the results obtained through the direct shear test and understand the role of the RCA particles. Due to the multi-phase nature of the aggregates, the cracks observed in the cement paste were classified into two categories: a) cracks in new cement paste (NCP) and b) cracks in the residual cement paste (RCP). It should be noted that the crushing and weathering of the aggregates could result in closed cracks inside the aggregates, and these cracks are not necessarily attributed to ASR (30). However, a decrease in such cracks is a result of closed cracks in the aggregates opening due to the expansion of silica gel within these sites. As such, the CC mixture shows an increase in the proportions of open cracks in the aggregates (from 12% to 40%), as well as cracks in the cement paste (from less than 1% to 10%) up to moderate expansion level (i.e., 0.12%). Afterwards, their proportions remain constant (Figure 5a) although the number of cracks keeps increasing, as shown in Figure 5b. The severely damaged mixture (i.e., 0.30%-SPR-RCA) displays a significantly higher portion of open cracks in the aggregate with and without gel when compared to the slightly damaged mixture (i.e., 0.05% -SPR-RCA) before being subjected to secondary damage (28% and 10%, respectively). The proportion of open cracks in the aggregate without and with gel increases up to 45% for slightly damaged RCA and 36% for the severely damaged RCA concrete at 0.20% of expansion (Figure 5a). Meanwhile, the cracks in the cement paste present a significantly smaller proportion overall yet, similar proportions in both recycled mixtures are observed, increasing with expansion (i.e., from 5% to 20%).

The DRI numbers generally increase with increasing expansion, as presented in Figure 5b. The CC mixture displays the highest DRI values, followed by the severely then slightly damaged RCA mixtures. Before being subjected to ASR, only a negligible level of damage was observed for all concrete mixtures (i.e., DRI values of 138, 73, and 171, for CC, slightly and severely damaged RCA, respectively). At 0.05% and 0.12% of expansion levels, the CC mixture presents the highest DRI numbers (i.e., 307 and 465, respectively) between all mixtures, while the slightly and severely damaged RCA mixtures achieve 161 and 261 for low (i.e., 0.05%) expansion and 268 and 390 for moderate (i.e., 0.12%) expansion, respectively. A significant increase in the DRI number is found for the slightly damaged RCA concrete at the expansion of 0.20%. (i.e., 470) whereas the severely damaged RCA and CC mixtures increased to 514 and 621.

The shear strength of recycled mixtures presented interesting results when compared to the CC mixture. At low expansion levels (i.e., 0.05%), CC mixtures presented cracks mainly within the aggregates (10, 11, 21). This behaviour was well captured by the direct shear test (i.e., no reduction of the shear strength as per Figure 4) since no “aggregate interlock loss” is expected when cracks remain within the aggregate particles. Conversely, at moderate expansion levels (i.e., 0.12%), some of the cracks developed within the aggregates reached the cement paste (Figure 1), decreasing the aggregate interlock due to the splitting of the particles, which significantly reduces the shear strength of the affected material. In addition, the location of cracks (i.e., aggregate vs cement paste) and their severity (i.e., number, crack length and crack width) significantly affected the shear strength of CC (31). On the other hand, RCA mixtures showed a lower initial shear strength when compared to CC, likely due to: a) the distinct microstructure (i.e., multi-phase nature and increased number of ITZ) of RCA particles and, b) the crushing/processing inducing micro-cracks, which might have compromised the aggregate interlock capacity of RCA concrete as previously observed in the literature (22, 32, 33). Conversely to the CC mixtures where the shear strength remained constant from 0% to 0.05% of expansion, the recycled mixtures showed a 19% and 20% shear strength loss. This difference in the shear strength reduction in recycled mixtures compared to CC is likely due to the previous ASR damage, resulting in the secondary damage having a short initiation period leading towards cracks extending to the cement paste in recycled mixtures before such cracks are observed in CC (Figure 5b). Moreover, the influence of the previous ASR damaged was captured by the direct shear test where recycled mixtures made of slightly deteriorated RCA concrete displayed a greater initial shear strength when compared to RCA made of severely deteriorated particles. The DRI numbers obtained for the slightly and severely damaged recycled mixtures prior to being subjected to secondary ASR damage (i.e., 73 and 171, respectively) support the trend observed through the shear strength. Yet, a DRI number of 138 was reported for the CC, while a higher shear strength was observed at 0% of expansion. Therefore, the number of cracks does not necessarily govern the shear strength. Moreover, the open cracks in the aggregates are significantly lower in number for the slightly damaged RCA yet, similar numbers are observed for the severely damaged mixture and the CC. At 0.05% of expansion, although the shear strength in CC remains constant, the number of open cracks in the aggregate increases significantly while a smaller increase is observed in both recycled mixtures, highlighting the influence of the nature of the RCA presenting previous damage. Furthermore, the width of cracks may have played an essential role in the direct shear strength of recycled mixtures. Recycled mixtures at low expansion levels presented significantly wider cracks than CC (i.e., 0.15 mm and 0.10 mm, respectively) (7).

An interesting behaviour is observed at 0.20% of expansion for the slightly damaged recycled mixture observed through both the shear strength and the DRI number. The shear strength thus presents a higher loss for the slightly damaged mixture while the DRI number increases significantly with the number of open cracks in the aggregate being similar to that of the severely damaged mixture, further highlighting the difference between the RCA particles subjected to different levels of previous damage. In addition, at high expansions (i.e., 0.20%), cracks were widest in the slightly damaged RCA concrete reaching 0.25 mm, compared to 0.20 mm for severely damaged RCA concrete and CC as previously reported (7). The sudden loss of direct shear strength observed in the recycled concrete made of slightly damaged RCA at an expansion level of 0.20% may be linked to wider observed cracks. Nevertheless, the reactive potential of the RCA particles was captured by both tools used in this study. It is worth noting that it is generally agreed that the shear strength of concrete not only depends on the location and width of cracks, but the crack directionality could also play an important role in the shear strength of concrete specimens (34, 35), thus the impact of crack directionality on the shear strength of RCA concrete is recommended topics for future work.