Concrete precast rejects – treatment, characterization, and application in new concrete as recycled aggregate

This article evaluates the feasibility of using concrete waste from precast production as recycled aggregate in concrete. The processing of concrete waste employed a jaw-hammer crusher and a sieve, producing three types of Recycled Concrete Aggregate (RCA). After their characterization, RCA was incorporated in two types of concrete used in the precast factory: a flowing concrete (FC), slump 220 mm, employed in columns and beams; and the second one, a no-slump extruded concrete (NSC), used for hollow core slabs. X-Ray diffraction, thermogravimetric analysis and scanning electronic microscopy showed phases from the hydrated cement paste and the original aggregate. Results of mechanical performance showed that RCA did not influence the compressive strength but influenced other properties such as water absorption and modulus of elasticity. Lastly, it was concluded that RCA obtained in the precast factory showed great potential to be used within the factory production process.


INTRODUCTION
The use of recycled aggregates (RA) from construction waste brings environmental and social benefits to society. One of the direct benefits is to prevent this type of waste to be deposited in illegal areas, which tend to affect mostly the poorest of the population. In addition to the known benefits, the demand for RA research is increasing for several reasons, such as: (1) the growth of the world population, which raise the demand for construction materials; (2) the scarcity of natural aggregates around the world (MENDES, 2012;MUKHARJEE;BARAI, 2014; World Business Council for Sustainable Developmen, 2009); (3) lack of areas for legal disposals, such as landfills (TAFFESE, 2018).
Therefore, a feasible solution to this issue is to reuse these wastes again in applications such as aggregate or filler for cement and concrete. In some countries of Europe (SALESA et al., 2017;SOARES et al., 2014), RA has been used in structural concrete, replacing natural aggregate, and, in 2018, RA represented only 8% of the total amount of aggregates produced in the EU (UEPG, 2018). This occurs due to the existence of standardization and the scarcity of natural resources, which is likely to happen soon in most countries of the world. Thus, research with recycled aggregates from different sources of waste such as construction and demolition, precast concrete, bricks, granite, marble, agroindustry ash, among other types aimed to make feasible the incorporation of these wastes in mortar and concrete. Although the use of recycled aggregate still seems small in Europe, it is relevant when compared to other regions such as South America, which has no surveys on this subject. Several factors justify resistance to the use of RA, which is explained by some authors (ISMAIL; HOE; RAMLI, 2013; SILVA, R. V.; DE DHIR, 2014).
Particularly in Brazil, there is no standardization for regulating the use of recycled aggregate in structural concrete. Because of the plenty of mineral resources in the country, the construction business has not given due importance to this issue in recent decades.
However, this scenario has been changing owing to the scarcity of aggregate reported in important urban centers, such as São Paulo and Curitiba (MENDES, 2012). In the case of the Northeast region, it was reported that the rocks along the Borborema Plateau have a high potential to develop the alkali-silica reaction (PRADO, 2008), and that it was a potential source of obtaining aggregates until then. Therefore, the use of recycled aggregate is a issue that has been gaining importance in Brazil in recent years.
The origin of the waste directly interferes with the properties of the recycled aggregate. Recycled concrete aggregate (RCA) is a class of RA originating from concrete waste, which is obtained from concrete producing and concrete precast rejects. The recycling in concrete plants has a tremendous advantage when compared to a purchase of RA from an external recycling plant, due to the elimination of the purchase and freight charges. Yet, when compared to natural aggregate (NA), RCA presents lower density (up to 10%)(ARORA; SINGH, 2016;MARINKOVIC et al., 2012;CHUNG;ASHOUR, 2008) et al., 2017). This is mainly due to the high porosity of the attached mortar (JAYASURIYA et al., 2021;SOARES et al., 2014), which decreases the quality of RCA and also influences the crushing process and the size of the particles produced.
Consequently, these characteristics are transferred to the concrete matrix, resulting in concrete with high porosity (WANG; YU; LI, 2020). Some authors reported an improvement in the mechanical properties of concrete after removing the attached mortar from RAC (SARAVANAKUMAR; ABHIRAM; MANOJ, 2016;SHI et al., 2016).
With respect to the fresh state of concrete with RCA, many researchers have reported a decrease in workability as the amount of RCA increases (KHATIB, 2005;RAO;JHA;MISRA, 2007). These authors concluded that the irregular shape and high porosity of RCA particles lead to an increase in the water amount, considering the workability kept. However, other researchers had different conclusions. Galvín et al. (GALVÍN et al., 2014) reported that the incorporation of coarse RCA did not decrease the value of the slump test. Güneysi et al. (GÜNEYISI et al., 2016) verified an increase on flowability with the increase of fine RCA amount in self-consolidating concrete. Soares et al. (SOARES et al., 2014) concluded that RCA did not lead the water/cement ratio to vary, at the slump range.
In general, RCA amount affects the mechanical properties of concrete. It is a consensus that the use of RCA in concrete increases its porosity and decreases its modulus of elasticity (CABRAL et al., 2010;EVANGELISTA;DE BRITO, 2007;KHATIB, 2005;SALESA et al., 2017;DA SILVA;SILVA, RUI VASCO;DHIR, 2016;CHUNG;ASHOUR, 2008). However, the same does not occur with compressive strength. Many authors have found that the incorporation of RCA results in a decrease of compressive strength as the amount of RCA increases(ARORA; SINGH, 2016;CABRAL et al., 2010;POON;KOU;LAM, 2007). On the other hand, an increase in the compressive strength of concrete with RCA is reported by Domingo-Cabo et al. (2009) andSalesa et al. (2017).
In addition to the works cited above, it can be found studies in which no clear trend of increase or decrease is shown when the amount of RCA increases(EVANGELISTA; DE BRITO, 2007;PÉREZ-BENEDICTO et al., 2012;SOARES et al., 2014). Through an extensive literature survey, Jayasuriya et al. (2021) concluded that the incorporation of 20% recycled aggregate can lead to an increase in compressive strength.
Thus, this paper aims to evaluate the technical feasibility of the employment of RCA from precast rejects as aggregate in concrete mixes used in a precast factory. Due to the inconsistencies found in the literature regarding the influence of RCA on compressive strength, a protocol of statistical analysis was used to interpret the results. This is one of the papers that are part of research on the influence of the use of recycled concrete aggregate for precast manufacturing. Another paper (ESTOLANO et al., 2018) was published focusing on the analysis of the relationship between the modulus of elasticity and the compressive strength.
We hope our findings will encourage the construction technical society to create standardization allowing the use of recycled aggregate in structural concrete, especially in developing countries like Brazil. Besides, this paper will help the construction business to know the characteristics of the recycled aggregate and evaluate how it can be used safely in structural concrete.

MATERIALS
RCA with different sizes were produced from a concrete precast plant located in the Suape Port Complex (Cabo de Santo Agostinho, Brazil). This plant had three sites of waste generation: the line production, where it was generated waste by the leftover concrete, and the faulty concrete pieces, which were rejected by the quality control staff; and the concrete lab, where the waste was generated by crushing the specimens and the rejected mixes in the fresh state assessment. RCA used in this research came from crushed precast elements of strength classes of 35 up to 45 MPa, corresponding to prestressed hollow core slabs, produced by the process of extruding dry concrete, as well as prestressed columns and beams and centrifuged piles. It was not possible to evaluate the influence of RCA by the type of concrete, because waste was not sorted. Part of the waste was used as a landfill to expand the factory space, however not enough to avoid the accumulation in piles (Figure 1). Destining the waste to controlled landfills was not a viable solution due to the high cost demanded. During the research period, it was estimated that approximately 49 m³ of waste was generated per month and a total of more than 800 m³ of accumulated waste. This waste had been acumulated for 5 years. The crushing process used to obtain the RA consisted of a jaw-hammer crusher.
Two different fractions of RA have been originated from the crushing of the precast concrete elements: a powder-like material named in this study as recycled fine sand (RFS) and one other, which resembled as gravel and named R6.3, according to their maximum diameter size. However, R6.3 corresponded to about 70% of all material produced. Therefore, to use R6.3 as a fine aggregate, a sample of it was sieved through a 4.8 mm mesh sieve, and the through fraction was named R4.8. Figure 2 shows the recycling process used in the experimental program.  Table 1 and Figure 3, respectively, which were performed according to Brazilian standards.
The cement used to produce concrete was a CP-V ARI (high initial strength cement Portland), commonly used for precast concrete plants, and a polycarboxylate-   Besides the standardized tests, a microstructural assessment was conducted on RAC.
Further those tests, RAC was assessed through optical (OM) and scanning electronic (SEM) microscopy. Also, energy dispersive spectroscopy (EDS) was employed with SEM to identify the chemical compounds in determined phases. OM was performed using an Olympus BX51 microscope (lens magnification: 5x) and the software AnalySIS imager. The SEM/EDS was carried out utilizing a TESCAN MIRA3 electronic microscope, equipped with a tungsten filament operated an accelerating voltage of 10 kV, with a working distance of 40 mm. EDS point analysis was acquired through a OXFORD (X-Maxn) detector.
The concrete mixtures chosen for the experimental program have been based on the procedures from the precast concrete plant. Two concrete mixes were used: a flowing concrete (FC), which was used in the manufacture of beams and columns, and a no-slump concrete (NSC), that was used in hollow core slabs made by extrusion. The control concrete FC-control and NSC-control followed the characteristics below: Before the production of control and recycled concrete mixes, some tests have been performed to define the content of the replacement of recycled aggregates, which would be used to evaluate the mechanical performance of the mixes. Due to the high absorption of RFS, it was concluded that in mixtures with RA it would be unfeasible to keep a water/cement ratio close to the reference mixture (FC-ref). Therefore, the solution adopted was to decrease the specific surface of the fine aggregate, adopting the composition of 50% RFS + 50% AR4.8 for the M1 mixtures.
Then, it was decided to analyze the replacement ratios (in mass) of fine natural aggregate (FS) by fine recycled aggregate (50% RFS and 50% RG4.8) in 30%, 50%, and 100%. The mixes with 30% and 50% reached a slump of 230±20 mm, according to the established by FC-ref, while with 100% replacement ratio of recycled aggregates showed a total lack of workability (slump values close to zero), Figure 4. It was not worth adding water to attempt to reach the desired consistency, because such a large increase in the water/cement ratio would lead to a great decrease in the recycled concrete strength. An attempt to improve the workability of the concrete was by adding a superplasticizer content greater than 1.0%. However, the concrete mixture showed a high viscosity, making its use unfeasible. It occurred because the FC-control mixture contained the optimum content of superplasticizer, therefore exceeding the amount of 0.6%, which led to an effect contrary to the desired one. Concerning NSC, besides the control mixture, a mix with 100% recycled aggregate (NSC-100) was produced. The replacements carried out in NSC-100 were natural gravel (G12 and G19) with RG 6.3, and NFS with RFS. The compaction of FC mixes was performed applying 12 strokes in each of 2 layers by a metal bar. On the other hand, compaction of NSC mixes employed a vibro-pressing desk, compacting the dry concrete in 3 layers. This method simulated the charges of the extrusion machine. Table   2 shows the concrete mixes' composition. In the fresh state, the consistency (slump test) and the fresh density of the concrete mixtures were studied. The slump test followed the procedures of the Brazilian standard NBR NM 67 (ABNT, 1998) For the evaluation of hardened properties, cylinder specimens with 10 mm diameter and 20 mm height were cast. Table 4 shows the tests performed, their respective standards, and the number of replicates for each test. Compressive strength, modulus of elasticity, and splitting tensile strength tests allowed understanding the influence of the use of RAC on the mechanical behavior of the concrete. In the modulus of elasticity test, micro-strain gages were used to measure the strain of the specimens during the load ( Figure 5). Source: author's authorship (2021) The values of the dynamic modulus of elasticity (Ed) (Eq. 1) were obtained from the ultrasonic wave velocity (V), the density of the concrete (ρ), and the Poisson's ratio (µ), which was set at 0.20.
Water absorption and hardened density allowed a better understanding of the effect of RAC on concrete porosity. To evaluate the surface hardness of the specimens, a Schmidt rebound hammer was used to apply strokes in 16 points on the side of the specimen ( Figure 6). In addition to assessing the surface hardness of the materials, it was also studied whether the rebound hammer, applied in concrete specimens, could be an alternative for destructive tests such as compressive strength. To guarantee the stability of the specimens during the application of the stroke, they were loaded to a compression of 5 tf.

STATISTICAL ANALYSIS
The minimum number of eight replicates in the compressive strength tests allowed the use of statistical tools to verify if the samples had significant statistical differences.
Initially, it was checked if the samples fit the normal distribution. After that, as each one had less than 30 elements, the Student's T distribution was used, like the normal distribution for the statistical inferences. Finally, the statistical similarity between the samples was analyzed by means of the T-tests for means and the normal analysis of variance (ANOVA). Figure 7 shows a flowchart of the protocol adopted for the statistical analysis.   (2017) found larger peaks of portlandite in a cement paste powder for using as filler.
The samples were cured underwater for 90 days, preventing carbonation. Park et al. (2020) found similar results with this work after carry out an accelerated carbonation on RAC.     Through SEM, hydrated phases of cement paste were identified on the adhered mortar of RCA. Figure 12 displays the presence of ettringite and calcite phase. Ettringite is easily identifiable due to its characteristic form of needles. Interestingly, though ettringite is a metastable phase, it was identified on the coarse aggregate even after at least two years from the initial hydration. The presence of ettringite was also identified by XRD ( Figure 8). Visually, calcite was the most prevalent phase in the sample, followed by portlandite. This proves once again that the recycled aggregate contained a strongly carbonated hydrating cement paste. The process of carbonation can be also seen in Figure 14. This picture shows an agglomerate of carbonated portlandite surrounded by a fractured quartz particle. Note that there are several structures where portlandite and calcite may appear. Therefore, the utilization of SEM with EDS/EDX is crucial for the identification of these compounds. The results of the standardized aggregate's characterization are displayed in Table   2. The particle density value of the RCA was on average 6.5 % lower than that NA, while the water absorption was 24 (AFR) and 14 (AR 4.8) times higher. It has been reported that, due to the porosity of the mortar, RCA density tend to be slightly lower compared to natural aggregate (CABRAL et al., 2010;KURDA;SILVESTRE, 2017;SOARES et al., 2014). The powder material showed also higher value in comparison to NA, as expected. These RCA properties are a challenge to make this material feasible as an aggregate for concrete, as it is immediately clear that concrete with RCA will have a lower density and higher water absorption.
In a microstructural perspective, the micropores and cracks were identified in the structure of adhered mortar of RCA are the main responsible for the tremendous increase in the water absorption, because of the increase on the superficial area to saturate the aggregate with water. This also influences the density of the aggregate, and consequently, the physical and mechanical properties of concrete.
However, even with a reasonable amount of mortar adhered, Los Angeles wear value was not superior to natural aggregates. The recycled aggregate RG6.3 showed a higher wear value than the natural aggregate, in the case of G12, and lower in comparison to G19, whose value was 25.67%, 20.48%, and 26.49%, respectively. Recycled aggregates usually show higher wear than natural aggregates due to the lower strength of their adhered mortar.

Fresh state properties
In this study, the slump value of concrete with RCA was kept constant, according to each control concrete (FC-control and NSC-control), to achieve desired results. Their water content changed (slightly) to obtain the target workability. The fresh concrete bulky density of the recycled concrete mixes was lower than the control concrete mixes, in both cases; almost 3% (FC-30) and 7% (FC-50) in comparison to FC-control, and 12% (NSC-100) compared to NSC-control. In general, the trend was to decrease the slump value as the RCA amount increases, due to the high value of water absorption of RCA (GHORBANI et al., 2019). However, this behavior is not a consensus in the literature. Poon, Koon, and Lam (2007) concluded that RA from construction and demolition waste improved the workability of concrete with fly ash. Soares et al. (2014) presented that RCA did not influence in concrete slump values. Further, similar results were found by Kurda, Brito, and Silvestre (2017). As shown in

Water absorption, void index, and density
RCA has a greater absorption capacity when compared to natural aggregate.
Therefore, as concrete in this research is composed of 73 -75 % of aggregate, it is expected that concrete with RCA presents a higher water absorption when compared to the control mix. Results of water absorption, void index, and density are presented in Table 6. After 72h, FC-30 and FC-50 showed a water absorption of 3.55 and 4.08, respectively 22.8 % and 41.2 % higher than the FC-control mixture. This increase in porosity is also remarkable by the increase in void index and density as the RCA amount increases. These results are similar to what has been found in the literature (GHORBANI et al., 2019;LI, 2008;OMARY;GHORBEL;WARDEH, 2016). As seen in Figure 15, the difference in results between NSC-control and NSC-100 mixtures was tremendously greater than in FC mixes. It occurred due to the total replacement of fine and coarse natural aggregate by the recycled aggregates, leading to an increase in the amount the aggregate with high water absorption. Furthermore, extruded dry concrete had a lacking quality control to evaluate its adequate consistency, and this fact made the finding of optimum water content difficult.
However,  found an increase in this property with the incorporation of RCA (Table 6). In this research, no trend of increase or decrease has been seen regarding flow concrete. FC-30 showed a slight increase compared to FC-control and FC-50 had the lowest value than the others. In NSC mixes, NSC-100 showed a lower value of splitting tensile strength compared to NSC-control, but disproportionate to the strong decrease in compressive strength. Therefore, no trend of increase or decrease was observed due to the incorporation of RCA.

Dynamic and static modulus of elasticity
Modulus of elasticity (ME) was seriously affected by the RCA incorporation. Just control showed values about 2 to 3 % higher for dynamic ME and 3 to 8% for static ME.
In the dry mixtures, the percentual difference of NSC-control to NSC-100 was 29.2% for dynamic ME and 54.7% higher for static ME.
The relationship between the compressive strength and the other properties was detailed in another article (ESTOLANO et al., 2018). The conclusions from this paper were that modulus of elasticity did not present a good relationship with the compressive strength, because, while the compressive strength was not affected by the incorporation of RAC, the modulus of elasticity decreased significantly. In this context, the increase in the porosity showed by water absorption and density tests presented a strong relationship with the modulus of elasticity. Thus, it can be concluded that RCA, mainly as fine particles, can enhance the total cement amount in concrete, and consequently decrease the water/cement ratio. Bordy et al.(2017) found a value of 24% corresponding to the amount of anhydrous cement in fine RCA for use as a filler in concrete. This value varies according to the proportion of materials used and the age of the concrete. MPa, the minimum value required for safely hoisting the concrete slabs.  Table 7 presents the results of outliers and normality analysis.  Therefore, from the statistical analysis in the flowed concrete, the replacement of natural sand by the composition RFS + AR4.8 did not result in a decrease in strength.
One hypothesis for this to have occurred is that the hydration of anhydrous cement in the RCA, which can reach significant values as shown by Bordy et al. (2017), compensated the increase in porosity and the slight increase in w/c ratio. In addition, there was no influence in the concrete compacting since the workability between the mixtures was similar. Therefore, the strategy of using an aggregate composition with a greater specific surface proved to be adequate.
In NSC mixtures, the p-values were very close to zero, confirming that the NSC-100 mixture showed a significant decrease in compressive strength.

Surface hardness
The surface hardness test was evaluated as a non-destructive alternative to the compressive strength test on concrete specimens. The rebound hammer applied to concrete parts is known to provide reasonable estimates of the mechanical properties of concrete. However, in concrete specimens, this type of evaluation is rare to find in the literature.
At first, it was found that a problem in carrying out this test would be the cylindrical geometry of the specimen, which would prevent the correct application of the blow.
However, it was found that such a limitation did not prevent the achievement of consistent results with an acceptable standard deviation. Table 9 shows that the standard deviation values were between 0.90 and 3.08, with a maximum variation coefficient (100% * standard / average) of 8.61%.
The fluid mixtures present close values, as well as the compressive strength. In noslump mixtures, the control mix showed results slightly higher than NSC-100. Except for NSC-100 at 1 day, all other mixtures have a rebound number/compressive strength (RN/CS) ratio of less than 1.00. Analyzing the point cloud between the rebound number and compressive strength data (Figure 19), there is a clear correlation between the two properties. Analyzing the data by type of concrete, there is a remarkable relationship between properties, presenting a correlation considered strong according to Dantas (1998) (R ²> 0.8). In an individual analysis by mixture, the values of R² varied between 0.6 and 0.9.  SEM evaluation presented a diversity of phases and structures such as portlandite, carbonates and C-S-H. Ettringite was identified by XRD and SEM even after at least two years from the primary hydration.  To maintain the workability between the mixtures of fluid concrete, it was necessary to subtly increase the w/c ratio for concretes with RCA, which did not significantly influence the mechanical properties. Therefore, the effect of RCA on the workability of fluid concrete was very little or none.
 In dry mixes, it was necessary to strongly increase the w/c ratio to maintain an adequate consistency for extrusion. Therefore, the total substitution of recycled sand and natural gravel has strongly increased the demand for water.
 Though there was no significant variation in the w/c ratio, the change in the packaging of the particles resulted in an increase in porosity, and consequently an increase in water absorption and voids index and a decrease in density and elasticity module.
 Regarding the results of splitting tensile strength, no trend was observed in both flowing or no-slump concrete mixtures.
 The statistical analysis showed that there was no influence of the incorporation of RCA in the fluid concrete. However, in dry concrete, the significant decrease in mechanical performance was observed. It shows that other methods of consistency analysis for extrusion must be studied to obtain an optimal w/c ratio.
 The employment of the rebound hammer test on the side of the specimens proved to be a good tool to estimate the strength of the concrete. Performing the analysis by type of concrete showed greater precision than analyzing considering all the results.