Radiation shielding properties of magnetite-MgO heavy density concrete: Experimental study and machine learning analysis

This study addresses the development and performance evaluation of heavy density concrete (HDC) incorporating magnetite fine aggregates and 5% magnesium oxide (MgO) for moderate temperature radiation shielding (up to 300°C) in water-cooled reactors. Various HDC mixtures were prepared using different ratios of magnetite aggregates (25%, 50%, 75%, and 100%) in addition to a reference mixture of standard weight concrete (without magnetite). Comprehensive experimental testing was conducted to evaluate the properties of HDC samples, including density loss, mass loss, compressive strength, rebound number, ultrasonic pulse velocity, scanning electron microscopy, and radiological properties, including linear and mass attenuation coefficients, half-value layer (HVL), tenth-value layer (TVL), and mean free path (MFP). These parameters were evaluated pre and post exposure to temperatures ranging from room temperature to 300°C. Three machine learning (ML) models, namely XGBoost, CatBoost, and LightGBM, were used to predict the compressive strength of HDC specimens. Among these models, XGBoost performed best with R 2 values of 0.971 for training and 0.981 for testing. The experimental results of magnetite-MgO developed HDC demonstrated a significant improvement in radiation shielding efficacy and mechanical properties, in contrast to concrete samples without MgO.