The cement used for the SCC production was High Early Strength Portland cement, which can contain up to 5% of limestone. The cement had a specific gravity of 3.06 g/cm³. The mineral addition used was all fine material of CDW (mortar filler) obtained by manual sieving, which passed through the 75 μm sieve. This material was obtained by crushing the mortar residue from several demolition sites and clandestine disposal points in the city of Feira de Santana/BA, Brazil, which underwent a gravimetric analysis to separate the inert phases (concrete, mortar, ceramic, natural stones, among others), and removal of impurities. The CDW filler had a specific gravity of 2.46 g/cm³ and its chemical composition is shown in Table 1 and the grain size curve of the CDW filler and the Portland cement used in this study is presented in Figure 1.

The natural fine aggregates used were a fine quartz sand (FS), with a maximum dimension size of 0.6 mm, fineness modulus of 1.50; specific gravity of 2.62 g/cm³ and bulk density of 1.48 g/cm³ and medium quartz sand (MS), with a maximum dimension size of 2.4 mm, fineness modulus of 2.86; specific gravity of 2.57 g/cm³ and bulk density of 1.42 g/cm³. The natural coarse aggregates were

a. LI—loss on ignition.

granitic in origin with a maximum dimension size of 19 mm (NCA2) and 9.5 mm (NCA1). The NCA2 presented a fineness modulus of 6.68, 2.76 g/cm³ of specific gravity and bulk density of 1.50 g/cm³. The NCA1 presented a fineness modulus of 4.28, 2.73 g/cm³ of specific gravity and bulk density of 1.40 g/cm³.

A third-generation superplasticizer based on polycarboxylate, MasterGlenium® SCC 160, manufactured by BASF with a specific gravity of 1.07 g/cm³ and 42% of solids content was used for the production of the SCC. The recommended dosage of the superplasticizer (SP) is 0.2% to 1.2% relative to the weight of the binder.

The SCC mix design was established according to Gomes [26]. It is an experimental procedure to obtain self-compacting concrete of high performance that meets a minimum compressive strength of at least 50 MPa at 7 days. Once the materials have been chosen, the proposed method separately optimizes the composition of the cement paste so as to obtain the desired viscosity and fluidity, and the aggregate granular skeleton, which guarantees the properties of the SCC in the hardened state. At the end, the lowest void index found was 31.33% and the bulk density was 1.84 g/cm³.

For the SCC-R, the following criteria were used: maximum cement paste volume limit of 40%, according to Gomes [26]. As far as fluidity is concerned, the result of the flow test found must be between 660 mm and 750 mm, as this is the adequate spread for most applications, besides being a parameter that is easier to control the resistance of segregation [17] [27] [28]. In what concerns the passing ability, the mixtures should also meet the limits defined by NBR 15823-3 [28], where the difference between the flow obtained with and without the use of the J ring must be at least 0 and at most 50 mm.

The mix proportions for SCC-R production were defined after adjustments in pulp volume, w/c ratio and SP content (Table 2). The mixtures were evaluated in the fresh state for flowability using the flow test [27], plastic viscosity by T500 [27] passing ability using the J-ring [28], resistance to segregation that was evaluated by the visual stability index—VSI [27] and by the sieve method [29] and the specific gravity that was defined by the gravimetric method [30]. In the hardened state, it was evaluated axial compressive strength [31] and splitting tensile strength [32], at 7 and 28 days, and specific gravity, water absorption and void index [33].

Table 3 shows the results of the tests performed on fresh Recycled Self-Compacting Concrete (SCC-R) mixtures. The flow results obtained for the SCC-R mixtures with different filler contents at time 0 (zero) minutes indicate that the mixtures up to the 10% of CDW filler achieved compatible flow with the SCC of reference (SCC-Ref). For the mixtures 15%CDW and 20%CDW there were larger flow reductions of 6% and 8%, respectively. It can be seen that the addition of the CDW filler, in the case of the 0 min time, did not affect the spreading of the

Note: Ref.—reference; w/c—water/cement ratio; C—Cement; CDW—filler of CDW; FS—natural fine aggregate; MS—natural medium aggregate; NCA1—natural coarse aggregate 9.5 mm; NCA2—natural coarse aggregate 19.00 mm.

a. instant 0 (zero) minutes.

concretes because all the mixtures were in the SF 2 spreading class, these concretes could be applied in most structures such as walls, columns, beams and slabs, according to NBR 15823-2 [27].

Checking the results of Figure 2, it is noticed a decrease of the flow over time. It is observed that, for the 15 min instant, the SCC-R mixtures obtained lower losses when compared to the SCC-Ref. At the instant 15 min the lowest losses were presented for the 5%CDW and 10%CDW mixtures, decreasing 9% and 19%, respectively. The other mixtures lost around 23% of their spreading when compared to their respective initial spreads in the instant 0 min.

At 30 min, the highest flow loss was presented in the 10%CDW mixture, which decreased 37% of the initial flow followed by the REF, 20%CDW, 15%CDW and 5%CDW mixtures, which presented a 34%, 33%, 30% and 26% reductions, respectively, when compared to the flow of their own mixtures at 0 min. It is further observed in Figure 2 that the flow loss over time for the 5%CDW and 10%CDW mixtures showed a better performance when compared to the REF mixture. For the 15%CDW and 20%CDW mixtures, the loss of spread over time was compatible with the REF mixture. The loss on the slump flow was measured over the time till the mixtures reached a 500 mm spread. So, at 45 minutes, the 5%CDW mixture was the only one that could have its slump flow measured. So, the test had stopped at 30 minutes for REF, 10%CDW, 15%CDW and 20%CDW mixtures.

The literature indicates that the SCC-R has a tendency to flow decrease with increasing CDW content when it is used as aggregates, and that this decrease intensifies over the time [21] [34] [35] [36]. This behavior is justified by the high absorption rate of the recycled aggregates [37]. However, although the CDW filler has a high-water absorption due to its porous and rough surface texture [38], the CDW filler also presents a smaller specific surface when compared to the cement used, these factors until the 10% of CDW filler content had no relevance in terms of fluidity when they replaced the cement (Table 3). As in the course of time the effect of water absorption did not overlap the effect of the chemical reaction of the cement that explains the mixtures containing CDW filler had less

loss of flowability over the time until 15 min. Above 10% of CDW filler content it is observed that the effect of water absorption is compatible with the effect of the chemical reaction of the cement over time.

By evaluating the apparent plastic viscosity, by means of the t500 test, it is possible to verify that all the mixtures were below the period of 2 seconds, falling within the class VS 1, for which the SCC is suitable for structural elements with high steel rate, according to NBR 15823-2 [27]. An increase in relative viscosity is also noted with increasing the CDW content.

Analyzing Figure 3, an increase of the apparent viscosity over time is observed. It is also found that the increase in viscosity is greater with the increase of the CDW filler content in the mixture. This behavior was also observed in the studies of Carro-Lopez et al. [34], Tang et al. [36] and González-Taboada et al. [39]. Perius [25] found that the CDW filler was efficient as a viscosity agent, increasing the viscosity of the cementitious matrix. Apparent viscosity loss over time for mixing 15%CDW and 20%CDW, up to 15 minutes, is compatible. However, after 15 minutes, the 20%CDW mixture exhibits a different behavior from the 15%CDW mixture, not reaching a spreading greater than 500 mm.

In this work, it was also observed that the decrease in fluidity, measured by the flow test, is smaller than the increase in apparent plastic viscosity, measured by the t500 at time 0 (zero) minutes. While the flow test value of mixtures 5%CDW and 10%CDW remain equal to REF mixture, the mixtures 15%CDW and 20%CDW show a decrease of only 6% and 8% in relation to REF mixture. When it is observed apparent plastic viscosity note that there is an increase of 13%, 130%, 245% and 358% for mixtures 5%CDW, 10%CDW, 15%CDW and 20%CDW, respectively, compared to REF mixture. The literature shows that mixtures with high flow values do not have a relationship between time dependent methods (t500, V-funnel, L-box), nor with flow difference methods (J-ring), locking ratio (L-box) and segregation [40].

By verifying the behavior of the mixtures in the J-ring test, it is noted that all

SCC-Rs were below the maximum limit required per standard, which is at most 50 mm, i.e. all SCC-R mixtures have the ability to flow without causing blockages or losing uniformity, according to NBR 15823-3 [28]. The SCC-R mixtures were classified as PJ 1, which are SCC suitable for structural elements with reinforcement steel spacing from 60 to 80 mm, except for the 10%CDW mixture, which was rated as PJ 2, which are suitable for structural elements with reinforcement spacing from 80 mm to 100 mm, according to NBR 15823-3 [28].

In the test performed by the J-ring method, Carro-López et al. [34] observed that the increase in CDW content decreases passing ability. Manzi et al. [41] did not realize the relationship between the passing ability and CDW content, a fact that was also observed in this study because there is no clear trend related to passing ability with increasing CDW content. This can be noted by checking the 10%CDW and 20%CDW mixtures, which had a spreading difference greater than the difference of the REF and 5%CDW mixtures. The 15%CDW mixture did not present differences in the flow and this difference was smaller than the difference of the REF mixtures and 5%CDW. González-Taboada et al. [39] state that the results obtained by J-ring do not correlate with the results of other empirical parameters.

Evaluating the visual stability index—VSI, during the visual observation in the flow test, it was possible to verify that all SCC-R produced had a homogeneous distribution of the mixture. Thus, they were placed as VSI 0, i.e., there was no evidence of segregation or exudation in the mixture, in conformity with NBR 15823-2 [27], as shown in Figure 4.

Concerning the evaluation of the resistance to segregation by means of the sieve test (Table 3), it was observed that all the SCC-R mixtures were classified as TP2, since the material retained in the sieve was less than or equal to 15%. As well as the passing ability it was also not found in this study a clear trend related to the segregation rate with the variation of the CDW filler content. It was verified that the 5%CDW mixture obtained the same segregation rate of the REF mixture and the other mixtures obtained the difference of less than 3%, 1% and

7% for mixtures 10%CDW, 15%CDW and 20%CDW, respectively, when compared to the segregation rate of the REF mixture.

Khayat and De Schutter [42], Tang et al. [36] and González-Taboada et al. [39] also found that the great majority of mixtures of different types of SCC had a segregation rate below 15%. According to Perius [25], the presence of the CDW filler has a positive influence against segregation when compared to the limestone filler. According to the hypothesis raised by the author, the CDW filler, even though it possesses larger diameter grains than the limestone filler, it has a greater irregularity in the form of its particles, besides the greater porosity of the material, thus helping in the reduction of the possibility of segregation of the SCC mixtures.

As regards the specific gravity in the fresh state, evaluated by the gravimetric method, it was observed that there is a decrease in the results with the increase of the CDW content in 0.84%, 1.37%, 2.43%, and 3.17%, for mixtures 5%CDW, 10%CDW, 15%CDW and 20%CDW, respectively, as compared to the SCC- Ref. This result was expected since the specific gravity of the CDW filler (2.46 g/cm³) was lower than that of the Portland cement (3.06 g/cm³) and the increase of the CDW filler content in the mixture would decrease the specific gravity of the SCC-R.

Table 4 shows the average results of axial compressive strength, splitting tensile strength at 7 and 28 days, absorption rate, void index and specific mass in the hardened state, at 28 days, as well as the standard deviation and coefficient of variation of the studied SCC mixtures.

At 7 days, the compressive strength results of REF, 5%CDW, 10%CDW, 15%CDW and 20%CDW mixtures were approximately 91%, 76%, 77%, 79% and 77% of the compressive strength at 28 days, respectively. That is, the reference concrete presented the highest initial strength gain. This behavior is due to the type of cement used in the study, high early strength Portland cement, which

Sd—standard deviation; CV—coefficient of variation; fc—compressive strength; ft—splitting tensile strength; A—Water absorption rate; V—voids index; Dd—dry density; Dsat—saturated density; D—density.

promotes high initial strengths. In addition, in general, the literature indicates that the SCC present greater compressive strength gains at initial ages, such as 3 and 7 days [43] [44]. The mixtures of SCC-R showed lower gains up to 7 days, since the filler, which is an inert material, replaces the Portland cement. Thus, the initial reactivity in these mixtures is less than in the reference mixture. At 7 days, the mixtures containing the CDW filler showed a reduction in the compressive strength results when compared to the average compressive strength of the reference mixture (REF) of about 16% for 5%CDW mixture, 17% for the 10%CDW and 15%CDW and 18% for the 20%CDW mixture. In addition, it is important to note that there was no great variation in the average results of compressive strength between the mixtures containing CDW.

At 28 days, it is observed that the average compressive strengths are similar. It is possible to perceive a higher gain of compressive strength with curing time for mixtures containing CDW (approximately 30%) when compared to the REF mixture (10%).

In order to confirm if the effects of the influence of the studied variables on the axial compressive strength results were statistically analyzed using the analysis of variance (ANOVA) using Statistica software. The variables verified on the compressive strength were: CDW (0 - 5 - 10 - 15 - 20%) and age (7 - 28 days). The ANOVA result is shown in Table 5.

The results of ANOVA show that all the studied variables, as well as the interaction between the CDW content and the age have significant effect on the average resistance to the axial compression of the studied mixtures. The CDW content was indicated as the main influence factor in the variation of the mean compressive strength to SCC-R, since it presented the highest calculated F value (368.68). This fact is possibly due to the average compressive strength at 7 days.

As shown in Figure 5, and by performing a multiple comparison analysis of the means of the results using the Tukey test, it was observed that at 7 days there is a significant difference between the average compressive strength for the REF mixture and the other mixtures which use any CDW’s content. However, among the average strength results of mixtures containing CDW filler this difference was not observed. At 28 days, it was observed that, for all the mixtures evaluated, there is no difference between the average compressive strengths.

Note: DF—Degrees of freedom; MS—Mean Square; F—F-distribution; F0.05—F-distribution value at the 5%; Sg—statistical significance; S—statistically significant; NS—statistically not significant.

The behavior of higher compressive strengths with increasing curing age was also observed by Silva et al. [45]. In this study the cement was partially replaced by CDW filler from masonry residue from construction and demolition sites. The authors found that the masonry residue affected the compressive strength in the first days of curing, reducing it as the replacement level increased. This behavior was justified by the increase in water/cement ratio, because when the residue replaced the cement, there was no change in the water/binder ratio, resulting in a lower compressive strength. The low reactivity of the masonry residue also affected the compressive strength, because as the cement is replaced in larger quantities and there is less formation of hydration products. These arguments may also explain the behavior of the SCC-R in the early ages of the present study.

A hypothesis for the average compressive strength of the mixtures produced in the present study did not show differences at 28 days, due to the fact that the particles of the CDW filler acted as filler, aiding in the packaging effect and possibly acting as nucleation points of the crystals of CH (calcium hydroxide) [46]. Another hypothesis may be related to the possible presence of cement (hydrated and non-hydrated) in the CDW fines, which may increase the total amount of cement in the mixture, thus compensating the substituted cement, as was also identified by Katz [47] in his study with CDW’s fine aggregate.

Based on the average results of splitting tensile strength at 7 and 28 days, it was observed that the results decreased with an increase of the CDW filler content, both at 7 and 28 days. At 7 days, the average splitting tensile strength of the mixtures studied reached 97%, 94%, 76%, 98% and 90% of their tensile strength at 28 days for REF mixtures, 5%CDW, 10%CDW, 15%CDW and 20%CDW, respectively.

At 7 and 28 days, SCC-Ref presented the highest results of mean tensile strength. For 7 days, the average splitting tensile strengths of the SCC containing the CDW filler were lower than the SCC-Ref by about 24%, 27%, 13% and 14% for the 5%CDW, 10%CDW, 15%CDW and 20%CDW, respectively.

At 28 days, the mixtures 10%CDW and 20%CDW showed the same average splitting tensile strength. The lowest mean tensile strength was observed for the 5%CDW mixture, which showed a reduction of about 21% in relation to the average splitting tensile strength of the REF mixture. The mixtures 10%CDW, 15%CDW and 20%CDW showed reductions of about 7.5%, 14.1% and 7.5%, respectively, compared to the average splitting tensile strength of the REF mixture. The highest mean tensile strength gains were observed for the mixtures 10%CDW (31%) and 20%CDW (11%).

Santos et al. [48] also noticed in the revisions of the studies that deal with SCC with CDW that there is a decrease in the value of the average splitting tensile strengths with the incorporation of this material. Silva et al. [45] observed that at 28 days of curing, the mixtures containing mansory waste residue showed a reduction in average splitting tensile strengths of 1.5% to 15.3% when compared to the reference mixture. The authors justify this behavior due to the slowness of the pozzolanic reaction of the residue. However, after 360 days the authors noticed an increase in the average tensile strength compared to the reference mixture up to the replacement content of 37.5% of masonry residue filler.

It is observed that the relationship between tensile strength and compressive strength at 28 days for all mixtures was less than 10%. As in this study the mixtures are for high strength SCC-R, this behavior was already expected. Although there is a relation between the tensile strength and the compressive strength, the higher the concrete compressive strength, the lower the relation [49].

The analysis of variance (ANOVA) was also performed to verify if there was significant effect of the partial replacement of the cement by the CDW filler and age on the results of average splitting tensile strength. The variables verified on tensile strength were: CDW (0% - 5% - 10% - 15% - 20%) and age (7 - 28 days). The ANOVA result is shown in Table 6.

Note: DF—Degrees of freedom; MS—Mean Square; F—F-distribution; F0.05—F-distribution VALUE at the 5%; Sg—statistical significance; S—statistically significant; NS—statistically not significant.

By means of the analysis of variance, it was verified that all the studied variables have a significant effect on the splitting tensile strength of the studied mixtures. Age was the most relevant influence factor on the results, since it presented the highest value of calculated F (32.19).

The results were also submitted to a multiple comparison of means, using the Tukey test, which evaluates whether there is a significant difference between the means of the results, and it was observed that:

● At 7 days (as can be noticed in Figure 6), there is a significant difference between the average splitting tensile strength results of the REF mixture relative to all other mixtures containing CDW filler. However, the Tukey test showed that the 5%CDW mixture showed no significant difference between the mean tensile strength results when compared to 10%CDW, 15%CDW and 20%CDW mixtures. And for the mixtures 15%CDW and 20%CDW there was no significant difference between them;

● At 28 days, the REF mixture showed a significant difference between the mean results of tensile strength with the average values of the 5%CDW and 15%CDW mixtures. The 5%CDW mixture also showed a significant difference with the 10%CDW and 20%CDW mixtures, with splitting tensile growth for these mixtures. However, the 10%CDW and 20%CDW mixtures showed no significant difference between the mean tensile strength results with the REF mixture or between them, showing that these average results are similar. That is, the use of 10% and 20% of CDW filler instead of cement does not promote a decrease in splitting tensile strength.

In what concerns absorption rate and void index, it is observed that, in general,

the value of the water absorption rate and the voids index had similar behaviors: increasing with the increase of the content of CDW up to the mixture 15%CDW. For the absorption rate, there was an increase of 5.4%, 12.5%, 23.2% and 19.6%, while the voids index showed an increase of 5.6%, 11.1%, 18% 2% and 15.9% for mixtures of 5%CDW, 10%CDW, 15%CDW and 20%CDW, respectively, when compared to the REF mixture. It is observed that for the 20%CDW mixture there is a small decrease in both the water absorption rate and the void index when compared to the 15%CDW mixture.

The results of absorption rate and voids index were analyzed statistically (Table 7), through analysis of variance (ANOVA), in which the effect of the variable content of CDW filler on these properties was investigated.

By means of the analysis of variance of the results of the water absorption rate and the voids index it was possible to notice that the CDW content was significant for these properties. There is a small difference in both the water absorption rate and the void index between the REF and 5%CDW mixtures and between the 15%CDW and 20%CDW mixtures (Figure 7(a) and Figure 7(b)). Therefore, we performed a multiple comparison analysis of the means of the results, using the Tukey’s test, and it was observed that the behavior for both the water absorption

Note: DF—Degrees of freedom; MS—Mean Square; F—F-distribution; F0.05—F-distribution tables at the 5%; Sg—statistical significance; S—statistically significant; NS—statistically not significant.

rate and the void index is the same: there is no difference of the mean results between the REF and 5%CDW mixtures, and between the 10%CDW, 15%CDW and 20%CDW mixtures.

Regarding the results of density, it is noticed that the mixtures present closer values. The REF mixture and the 5%CDW mixture obtained similar values of density. There was a decrease of 0.7%, 1.6% and 1.1% in the density for 10%- CDW, 15%CDW and 20%CDW mixtures, respectively, compared to the REF mixture.

The result of the SCC density was also analyzed statistically (Table 8), through analysis of variance (ANOVA), and the investigated variable was the content of CDW. The results show that the content of CDW has a significant effect on this property.

An analysis of Figure 8 together with a multiple mean comparison analysis using the Tukey’s test show that cement substitution by up to 10% of CDW filler has no significant influence on the density mean value. Furthermore, the mean density results for the 20%CDW mixture are statistically equivalent to those of the 10%CDW and 15%CDW mixtures.

Note: DF—Degrees of freedom; MS—Mean Square; F—F-distribution; F0.05—F-distribution tables at the 5%; Sg—statistical significance; S—statistically significant; NS—statistically not significant.