Parametric sensitivity analysis and modelling of mechanical properties of normal- and high-strength recycled aggregate concrete using grey theory, multiple nonlinear regression and artificial neural networks

It is well-understood that the incorporation of recycled concrete aggregates (RCAs) in a concrete mix can lead to some impacts on the mechanical properties of the concrete due to the inferior characteristics of the RCAs. In this study, the performances of available code-based and empirical models reported in the literature on recycled aggregate concrete (RAC) mechanical properties (i.e., compressive strength, elastic modulus, flexural strength and splitting tensile strength) are first assessed using extensive experimental data collected from the literature, and the assessments indicate that these models cannot achieve a desirable accuracy for their predictions. Aiming to develop more reliable approaches for predicting RAC's mechanical properties with higher accuracy and to cover wide-range of influential parameters of RAC mixes in the model expressions, a mathematical approach, namely grey system theory (GST) is used to examine the parametric sensitivity of the mechanical properties of RACs. The results of GST indicate that the overall mechanical properties of RACs depend on the geometrical indices of aggregates and also the concrete mixture proportions. The evaluation of GST also confirms the facts that the effect of RCA is different for the concrete at normal and high strength grades due to the difference in the failure mechanism of the concrete at different strength grades. Finally, multiple nonlinear regression (MNR) and artificial neural networks (ANN) are employed to simulate the mechanical properties of RACs using the key parameters of RAC mixes identified using GST. The results demonstrate that the proposed MNR and ANN approaches can provide more accurate predictions for the mechanical properties of RACs compared to previous models reported in the literature.