To determine the compressive strength of self-compacting recycled aggregate concrete using artificial neural network (ANN)

Abstract

Nowadays, special concrete-like self-compacting concrete (SCC) requires sustainability by introducing recycled aggregates as a partial replacement for natural aggregate. Technological development initiatives in the construction sector estimate the 28 days' concrete compressive strength before casting due to faster requirement; one method selected is an artificial neural network. From works of literature, 515 mixed design are collected and utilized for training, validation, and testing data to prepare models. Different applications of SCC require different strengths of concrete. Based on control mix compressive strength, the mix designs are grouped into three families as low, medium, and high strength, apart from a common family. The correlation between input and output variables for three different families is analyzed. ANOVA analyses are done for input parameters. Coefficient of relation (R2) is used for sensitive assessment and results for family I (R2 = 0.9299), family II (R2 = 0.824), family III (R2 = 0.8775), and family IV (R2 = 0.7991). Two further sensitivity analyses indicate that input parameters' influence varies for different families.