Prediction of Setting Time and Strength of Mortar Using Soft Computing Technique

Soft computing techniques, i.e., linear regression, artificial neural network, genetic expression programming, etc., are being practiced for the prediction of data. In this study, artificial neural network model predicted the consistency, setting time, and compressive strength of mortar at various curing time. The eighteen distinct mix proportions of cement mortar consisting of accelerators, i.e., calcium nitrate and triethanolamine as additives and stone powder as replacement of cement were selected for the prediction of various parameters. The accelerators are used to fasten the stiffening of cementitious materials and speed up the construction work. Stone powder was used to minimize the consumption of cement and problems associated with waste to the ecosystem. The laboratory data set was used for the prediction model. The appropriate artificial neural network model constitutes mix constituents as input parameters, i.e., cement, sand, water, and additional materials. The results from ANN training in multilayer feedforward neural network were evaluated and compared with the experimental results. A graphical representation between predicted and experimental results was also drawn. Results showed that artificial neural network technique was found effective for the prediction of various parameters of cement mortar with high correlation coefficients and low values of mean absolute error and root mean squared error.