A Predictive Model Based ANN for Compressive Strength Assessment of the Mortars Containing Metakaolin

Abstract

A predictive model based on the artificial neural network (ANN) was generated to assess the compressive strength of Mortar incorporating metakaolin (MK). For this purpose, a database was gathered from different works of literature for use in the ANN model. Therefore, five predictive variables as the inputs of the ANN model were considered, including the age of the samples, the ratio of MK replacement, the ratio of water to the binder, the ratio of superplasticizer, and the ratio of binder to sand. Using the constructed ANN model, a new formula has been presented, which can predict the compressive strength of the mortars incorporating MK. Then, the performance of the presented formulae was examined. The obtained conclusions indicated that the evaluated formula can predict the compressive strength of the mortars containing MK. Also, in the end, Garson's algorithm as a sensitivity algorithm was employed to examine the effect of each predictive variable on the compressive strength of the mortars incorporating MK. The results reveal that the binder-sand ratio is a more important parameter in determining the compressive strength of the mortars incorporating MK.