Predicting the Compressive Strength of Mortar Incorporating Bentonite Using Artificial Neural Networks

Sustainable concrete has become more popular due to supplementary cementitious materials (SCMs) that help achieve sustainability. Despite the well-established benefits of these SCMs, the search for substitute materials continues as they become harder to find and adapt to changes with the industry. Concrete performance may be enhanced using bentonite, a commercially available clay mineral that shows promise as an SCM. In the present work, an Artificial Neural Network (ANN) model was developed to predict the compressive strength of cement-based mortar incorporating bentonite as a SCM, by training it on existing data, allowing for better performance and mix design improvement. A comprehensive experimental database comprising test specimens was established. A critical assessment of the collected experimental data suggested that there are several key parameters governing compressive strength gains. The proposed model's parameters, such as weights, biases, and transfer functions, were effectively transformed into a mathematical model that correlates the compressive strength with the key input parameters. An experimental investigation measuring the impact of treating bentonite at various temperatures on compressive strength was also included in the study.The statistical evaluation results indicated that a three-layered Artificial Neural Network model with different hidden neurons could precisely estimate the compressive strength of mortar mixtures modified with bentonite, showing strong agreement with the experimental results. The mortar's compressive strength may be increased by partially replacing cement with calcined bentonite, especially in the initial stages. The type of bentonite and the intended performance determine the appropriate replacement rate and calcination temperature.