Prediction of Carbonation Capacity of SCMs Using Ensemble Learning Method

The utilization of supplementary cementitious materials (SCMs) subjected to carbonation processing represents a viable strategy to mitigate anthropogenic CO2 emissions associated with concrete production, potentially contributing to the achievement of carbon neutrality. However, existing studies have limitations in effectively predicting the varying carbonation capacities of different SCMs, a gap that this research aims to address. Recent research efforts focused on the carbonation of waste-material-sourced SCMs are reviewed, along with a comparative discussion on diverse carbonation methods. A detailed data set encapsulating the properties of SCMs, and carbonation configurations was compiled. At the same time, six ensemble learning models were developed and evaluated, with a particular emphasis on the CatBoost model due to its exemplary performance in predicting the carbonation capacity of SCMs. This study suggests a promising direction for optimizing carbonation processes across different types of SCMs, underscoring their potential in sustainable concrete production.