Cement CE II/B-LL 32.5R (Italcementi limestone Portland cement) according to EN 197-1 was used. The chemical composition of this cement is reported in Table 1.

A superplasticizer (SP, ester of acrylic or methacrylic acid monomer) having 1000 g/mol side chain length and an acid/ester ratio equal to 3.5 was used. Crushed asphalt was supplied by a local recycling company; it was not cleaned before turning it to aggregate, only air dried. The experiments aimed to reuse crushed asphalt on site, so no pre-soaking or conditioning was carried out to simulate real in-situ manufacturing. Three natural aggregates and GA were used (Table 2):

CG: coarse gravel (1 ÷ 30 mm)

The grading of natural and GA aggregates was evaluated by sieve analysis according to EN 933-1 (Figure 1). The grading of GA was similar to that of CG. However, the presence of a higher amount of fines was noticed (see Figure 1). Coarse sand, on the other hand, was significantly finer than GA. Chemical analysis of GA was also carried out according to EN 1744-1 noting that sulfate, sulfur, and soluble chloride contents were lower than the limits allowed by EN 12620. Petrographic analysis (EN 932-3) showed no gypsum, amorphous silica or pyrite in the GA. The methylene blue test (EN 933-9) indicated no mud or clay being present in the GA.

Water absorption and density of natural aggregates and GA were evaluated according to EN 1097-6 (Table 2). GA showed higher water absorption, but lower density than natural aggregates as a consequence of its bituminous binder content.

The concrete mixes were optimized (13, 14) in order to evaluate the feasibility of replacing natural aggregates by GA. Both natural and GA aggregates were combined to meet the Bolomey curve, modified to take into account the cement factor of the mix. Concretes were made by increasing the amount of GA aggregates to replace natural aggregates up to 20% (by mass). Dry GA was used in the concrete and the water content was adjusted in order to compensate for the absorption of aggregates during the concrete mixing procedure (Table 2).

All concrete mixes were prepared in a power-driven rotary mixer with a moving base (but without blades or paddles). The mixing procedure was as follows: first the dry components were mixed for 2 minutes, then 70% of the total water was added and mixed for 1 minute, then the rest of the water was added and mixed for one more minute. Superplasticizer was added to meet the target workability.

Three series of concrete mixes were prepared: the first (a) with a cement content of 330 kg/m³ and a fixed w/c ratio of 0.53 (Table 3) for both the reference concrete and GA mixes. In the second series (b), the w/c ratio of the GA mixes was reduced to 0.45 whilst increasing the superplasticizer dosage and maintaining the same cement content as in the first series (mixes GA_b5%, GA_b10%, GA_b15%, GA_b20%) (Table 4). The third series (c) was made by increasing the cement content to 350 kg/m³ and adopting a w/c ratio equal to 0.45 (mixes GA_c 5%, GA_c 10%, GA_c 20%) (Table 5).

Samples were cast in 150 mm cube steel moulds, which conformed to EN 12390-1. The moulds were cleaned and lightly coated with oil before the casting procedure. Concrete was compacted on a vibrating table. After that, the samples were covered with polyethylene film and left to set for 48 hours. Then they were removed from the moulds and cured in water (at a temperature of 20±2 °C) for 7 days and in a curing chamber (at an air temperature of 20±2 °C and relative humidity ≥95%) for another 21 days or until testing, thus conforming to EN 12390-2.

Dosage of the polycarboxylate-based superplasticizer was adjusted for all mixes to attain the same 200 mm slump at the end of mixing (S4 class, according to EN 206-1) as the reference concrete. Slump was measured immediately at the end of mixing and 30 and 60 minutes thereafter according to EN 12350-2. Entrapped air and density were also evaluated on the fresh concrete according to EN 12350-6 and EN 12350-7, respectively. The density and compressive strength of the hardened concrete at 1, 7, 14 and 28 days were measured in accordance with EN 12390-2. Finally, elastic modulus at 28 days and drying shrinkage up to 120 days were determined on the hardened concrete according to EN 12390-13 and UNI 11307 respectively.

The addition of increasing percentages of GA in the concrete mixes reduced the superplasticizer demand as evidenced in Figure 2 (first series concretes). The superplasticizer dosage was in the range of 0.2–0.4% for the mixes at 0.53 w/c ratio and 330 kg/m³ cement dosage. In the second and third series, the superplasticizer dosage was increased (vs that of the first series) as a consequence of the lower w/c ratio (0.45 vs 0.53) (see Figure 2).

Higher GA addition led to higher workability at 60 minutes (Figure 3). Despite the higher water absorption of GA with respect to natural aggregates, the GA addition produced positive effects on workability retention. This could probably be attributed to the rounded shape of natural aggregates and also to the presence of oil traces that act similarly to superplasticizer for granting higher initial fluidity and prolonged workability. This effect can be considered beneficial for prolonged transportation or placing of concrete, especially in hot climates. Flow table tests were carried out according to EN 12350-5 and confirmed the slump test results (Figure 4). The same tendency was also evidenced in the case of concrete mixes with a w/c ratio equal to 0.45 (second and third series). The addition of GA beyond 5% produced a drastic reduction in the 60 minute workability loss of concrete (Figure 5). In concretes with a cement content of 350 kg/m³, GA addition was responsible for a very low workability loss, independently of the replacement percentage of GA (Figure 6).

No bleeding or segregation tendency was detected for any of the concretes except for the mix with 20% of GA. For this reason, the maximum percentage of GA to avoid bleeding must be limited to 15%. No anomalous air entrapment was noticed for any of the mixes. The density of GA concretes was substantially similar to that of RC with only natural aggregates.

Compressive strength tests were carried out to evaluate the effect of GA addition on the mechanical properties of concrete. An increase in GA addition produced a decrease in compressive strength of concrete at early ages (Figure 7).

The compressive strength decrease is evident for the highest substitution of natural aggregates; at 20% substitution, the compressive strength decrease of GA concrete was about 50% with respect to RC. This could probably be ascribed both to the presence of oils in bitumen that adversely affect the kinetics of cement hydration and to a poor aggregate-cement paste interface, as evidenced by the failure paths during the compressive strength tests. The preferential crack propagation detected by visual observation, in fact, was through the cement paste-GA interfaces, in the zones where the presence of hardened bitumen was high. However, the reduction of the w/c ratio (second series: b), produced a positive effect on the compressive strength of concrete prepared with GA aggregates (Figure 8), as expected. Compressive strength values are similar for RC and GA concretes up to 15% substitution level; yet any further increase up to 20% produces a 15–20% reduction in strength. As expected, the same effect was evidenced also at the highest cement dosage (Figure 9). In conclusion, on the basis of the compressive strength test results, GA use should be limited to 15% replacement.

Drying shrinkage was also measured on samples made with a cement content of 330 kg/m³, w/c=0.45 and 5–15% GA replacing natural aggregates (Figure 10). Data were compared to those collected for RC with 330 kg/m³ cement content and w/c ratio equal to 0.53. Drying shrinkage of GA mixes was similar to that of the reference mix. However, it must be noted that GAC was prepared at a lower w/c ratio with respect to RC (at the same cement content). Consequently, the volume of cement paste, responsible for shrinkage, was lower for GAC vs RC. Hence, it can be concluded that the GA addition had a negative effect on shrinkage. This can probably be attributed to the lower stiffness of GA with respect to natural aggregates.

Young's modulus of hardened concrete was measured on cylindrical samples at 28 days (UNI-EN 6556). The higher the GA replacement the lower the Young's modulus, see Figure 11. However, reduction in w/c ratio (0.45 vs 0.53) allowed achievement of a Young's modulus for the 15% substitution of natural aggregates similar to that of RC.