Abstract This study investigates the thermal gradient and peak temperature as key indicators of thermal cracking risk and long‐term structural integrity in mass concrete. A comprehensive 2‐year experimental program comprising 91 laboratory tests was conducted to evaluate the adiabatic temperature rise and thermal conductivity of different concrete mixtures. The mixtures comprised diverse proportions of cement, ground granulated blast furnace slag (GGBFS), and fly ash (FA), as well as varied water‐to‐cementitious material (w/cm) ratios. The temperature rise was measured using a semi‐adiabatic calorimeter, while the thermal conductivity was experimentally determined using a TEMPOS meter. The experimental results were subsequently utilized as input parameters in a finite element method (FEM) model to simulate the core temperature evolution and the thermal gradient between the core and surface of mass concrete elements. Validation of the FEM predictions was carried out using field temperature measurements acquired from a 2 m × 2 m × 2 m mock‐up specimen of a real raft foundation with a thickness of 2 m. The results demonstrated a good agreement between the numerical and measured data. Furthermore, ANN and XGBoost models were developed to predict maximum core temperature and core‐to‐surface temperature differential using cement content, FA, GGBFS, maximum adiabatic temperature, its rise rate, and thermal conductivity as input parameters. The machine learning models demonstrated strong predictive capability, with the ANN achieving R 2 values of 0.96 for core temperature and 0.94 for thermal gradient, while XGBoost attained comparable performance with R 2 values of 0.93 and 0.92, respectively. The proposed integrated experimental–numerical–machine learning framework accurately predicts the maximum core temperature and thermal gradients in mass concrete containing cement with varying replacement levels of GGBFS and fly ash.
Muneer K. Saeed (Wed,) studied this question.