Predicting the Impact of Chemical and Physical Variability in Binary and Ternary Cementitious Blends

To reduce the quantity of CO2 emitted within the construction industry, cementitious by-products will need to be implemented on a larger scale. In relation to the use of by-products, one of the biggest disadvantages is that not only from source to source, obtaining a by-product from the same source could result in a variation in the chemical and physical properties daily which will then impact the mechanical properties. Therefore, the paper reviewed binary and ternary cementitious pastes that were produced from 7 different by-products, to analyse and predict the impact of variation in the chemical and physical properties on the 14-day compressive strength. The predictions and analysis were done with the use of artificial neural networks (ANN). Overall, ANN successfully derived an accurate prediction which correlated with the trends that were expected. This study noted that if parameters of the overall mix were taken into consideration, the increase in SiO2 will have a negative impact while increase in CaO would have a positive impact on the 14-day strength. The most accurate form of understanding the impact that chemical and physical variability of cementitious replacements, took into consideration both Ca/Si ratio and the average particle size.