Prediction of the compressive strength of nano-titanium based concrete composites using machine learning

Abstract

Cement manufacturing process is energy-intensive and can release significant volumes of carbon dioxide and other greenhouse pollutants which leads to the severe degradation of the environment. To save the environment, the consumption of the cement must be reduced by replacing it with other materials known as Supplementary Cementitious Materials (SCMs). Utilizing SCMs in concrete can help in lowering the construction industry's overall carbon emissions. SCMs are not only helpful in protecting the environment, but they can also improve the mechanical properties of concrete, mainly when applied in the form of Nano-particles (NPs). When NPs, in particular nano-titanium (nT), are used to replace cement in concrete, it not only improves the concrete's mechanical characteristics but it also improves the durability of concrete. Apart from various advantages of using nT as SCM in the concrete, the most superior advantage is the enhanced compressive strength (CS) which is the far most important mechanical property of concrete. This property of the concrete depends non-linearly with the ingredients required to make it. So, in order to predict the influence of these ingredients on the CS, researchers must either conduct extensive experiments or use soft computing techniques. With the help of these techniques, accurate and trustworthy models were developed to predict the CS of concrete. So, in this study, two machine learning models namely Artificial Neural Network and Gradient Boosting were developed to predict the CS of nT-based concrete composites after 28 days of curing. On comparison between the two developed models, Gradient Boosting model was found to be the most effective for an accurate CS prediction with Correlation coefficient (R2) as 0.992, Mean Absolute Error (MAE) as 1.015 and Root Mean Squared Error (RMSE) as 1.385. Artificial Neural Network was deemed to be inferior with R2 as 0.972, MAE as 1.34 and RMSE as 2.601.