Predicting the mechanical properties of concrete incorporating metakaolin, rice husk ash, and steel fibers using machine learning

The demand for sustainable concrete is a priority in the current era. The construction industry is actively working to mitigate the environmental impact of cement production in concrete by incorporating alternative and supplementary cementitious materials and reducing carbon emissions. Metakaolin (MK) and rice husk ash (RHA) have become popular in this context due to their pozzolanic properties. This study aims to predict the compressive strength (CS), split tensile strength (STS), and flexural strength (FS) of concrete incorporating MK and RHA with the addition of steel fibers. It examines the influence of several factors, including cement, MK, RHA, water, fine aggregate, coarse aggregate, steel fiber, superplasticizer usage, and concrete age. To achieve this, the study combined experimental work with existing data from the literature. Various machine learning (ML) models, including linear regression, support vector machines, Gaussian process regression, and artificial neural networks (ANN), were employed to analyze and assess the impact on strength characteristics including CS, STS, and FS. The dataset was divided into training and testing subsets, and statistical analyses were conducted to explain the relationships between the input parameters and CS. The ANN model demonstrated superior precision in predicting CS, STS, and FS compared to the other ML models. The findings suggest that the ANN model, utilizing the identified input parameters, can accurately estimate the CS, STS, and FS of concrete with MK, RHA, and steel fibers. The application of such technologies in the construction sector can enable the rapid and cost-effective assessment of material properties and the influence of various input parameters.