Experimental investigation and SVR model to predict the mechanical properties of RAC by enhancing the characteristic of RCA using surface treatment method along with modified mixing approach

Abstract

The present study aims to make the concrete prepared with recycled concrete aggregate (RCA) more sustainable by evaluating workability and mechanical strength, i.e., compressive, flexural, and splitting tensile strength of recycled aggregate concrete (RAC). In the present study, RCA is used as a replacement for natural coarse aggregate, and replacement is done by 20%, 40%, 60%, 80%, and 100%. In the context of making RAC more sustainable, a modified two-stage mixing approach (M-TSMA(sp-ggbs)) is adopted. In this mixing approach, cement is replaced by 10% of silica powder and ground granulated blast slag (GGBS) and mixed with water to prepare a slurry, in which RCA is added in the pre-mixing stage. It is observed that the mechanical properties of concrete are slightly improved. Further, to improve RCA characteristics, RCA treatment is done using two surface treatment solutions, HCl and Na2SiO3; (a) RCA is soaked in three different molarities, i.e., 0.5 M, 1 M, and 1.5 M of HCl solution, for 1 and 7 days, and (b) RCA is soaked in 15%, 30%, and 45% concentrations of Na2SiO3 solution for 1 and 7 days. The RCA showing the best characteristics regarding water absorption, impact, and crushing value is considered for preparing the concrete mix. Now, the treated RCA is used to prepare sustainable RAC in conjunction with M-TSMA(sp-ggbs). The experimental results show that the workability of all concrete mixes is within the desired range. Moreover, the mechanical properties are significantly improved by considering RCA treated with 45% concentrations of Na2SiO3 soaked for 7 days at a 60% replacement level along with M-TSMA(sp-ggbs). Further, this article used support vector machine regression (SVR) to predict the compressive, flexural, and splitting tensile strength based on the experimental values and the input variables. The SVR model predicted the compressive, flexural, and splitting tensile strength with 97%, 95%, and 88% accuracy.