Predicting the compressive strength of rubberized concrete incorporating brick powder based on MLP and RBF neural networks

Abstract

The investigations on the performance of concrete incorporating waste materials hold promise in achieving sustainable construction. Although various studies have addressed the mechanical behaviour of concrete containing waste tyre rubber (WTR) and clay brick powder (CBP), an advanced understanding of predicting the compressive strength of such concrete remains underdeveloped. In this study, predicting the compressive strength of rubberized concrete incorporating CBP using the artificial neural network (ANN)-based models is proposed for the first time. The prediction is based on multilayer perceptron (MLP) and radial basis function (RBF) neural networks. It is shown that MLP is superior in predicting the compressive strength of rubberized concrete incorporating CBP compared with RBF. However, regardless of the algorithm used, the R² and adjusted R² values are higher than 0.75. Results on Pearson’s r values greater than 0.85 illustrate higher predictive abilities of the neural networks. Moreover, the study demonstrates that it is not possible to obtain significant relationships between the individual independent variables of rubberized concrete incorporating CBP and concrete compressive strength. The ANN-based models in this research contribute towards an understanding of predicting the compressive strength of rubberized concrete incorporating CBP, which can inspire further modeling studies involving such materials.