Predictive model for carbonation depth of limestone filler concrete using a deep learning algorithm

Carbonation of hardened cement paste is one of the main damages causes in the corrosion of steel reinforcement, which may endanger the concrete structure, leading to deterioration and loss their integrity. For this purpose, the carbonation depth of concrete should be predicted. The current study investigates to predict the impact of partially replacing Portland cement with limestone filler on the carbonation depth of concrete by a deep learning algorithm. Therefore, an optimizer algorithm (Adam) and a Huber loss function were used to train this model. The developed model demonstrated excellent predictive performance, achieving a coefficient of determination (R2) exceeding 98%, a root mean square error (RMSE) of 1.81, and a mean absolute error (MAE) of 1.25. Therefore, a parametric analysis was performed to study the impact of the main factors that influence this phenomenon. Finally, a deep learning model, such as a Convolutional Neural Network (CNN), was created, and a parametric study was performed. The results demonstrated that the utilization of CNN drastically enhanced the accuracy of the model, lending it a high level of validity as a reliable tool for accurately simulating and predicting the carbonation depth of concrete. Accordingly, the proposed model is capable of predicting carbonation-induced corrosion and can serve as a fundamental tool for predicting the service-life of concrete structures.