Modeling The Microstructure Of Slag Blended Cements Using Artificial Neural Network

Abstract

This paper proposes the artificial neural network (ANN) as a numerical technique to simulate the microstructure of slagblended cements. The ANN model adopted in this research consists of three neurons in the input layer which represent contents of both water-cooled slag (WCS) and OPC and WCS finenesses and three neurons in the output layer which represent calcium silicate hydrate (C-S-H), Portalndite (CH), and capillary porosity. Back Propagation algorithm was employed for the ANN training in which a Tansig function was used as the nonlinear transfer function. Thermogravimetric analysis and de-sorption approaches were performed to study the microstructure of the different OPC/slag pastes. The results obtained from experiments agreed with that predicted by ANN, where, the prediction ANN model gives very close estimates of C-S-H, CH, and capillary porosity of OPC/WCS pastes. Therefore, the developed ANN model can be used as an alternative approach to evaluate the microstructure of slag-blended cements.