Predictive modelling of concrete tensile strength using ANN and Supplementary Cementitious Materials (SCMs)

Abstract

The global construction industry's rapid growth has led to a substantial increase in construction and demolition waste, necessitating sustainable solutions. Supplementary Cementitious Materials (SCMs) derived from waste materials have emerged as eco-friendly alternatives, improving concrete's mechanical properties and durability while reducing its Footprint of carbon. The goal of this study is to use artificial neural networks (ANNs) to predict the tensile strength of concrete that contains SCMs. A robust ANN model was developed using a dataset of 440 entries, encompassing various SCM types such as waste glass powder, ceramic waste powder, and marble powder. The Levenberg-Marquardt algorithm was employed to optimize ANN architectures, achieving high predictive accuracy. The best-performing model demonstrated an excellent correlation coefficient (R = 1) and minimal Mean Squared Error (MSE), validating the efficacy of ANNs in forecasting concrete properties. The findings highlight the potential of SCMs and ANNs in advancing sustainable construction practices by reducing reliance on physical testing and accelerating mix design optimization. The approach becomes even more practical thanks to an developed application provided that shall be able to give, on-spot optimization of the mixture.