Smart optimization of packing density for cleaner production of cementitious composites using mineral powders

Abstract

Cementitious composites are porous materials; therefore, controlling their porosity during the design process is crucial to ensuring the quality and strength of the composite. Recently, with the growing urgency to reduce the carbon footprint of cementitious composites, novel admixtures have been introduced as substitutes for cement. Each time a novel admixture is implemented, a research campaign is required to evaluate packing density. Traditional methods used for this purpose are time-consuming, costly, and generate waste. Implementing a waste-free solution using neural networks is a promising alternative. The authors present an analysis of the effectiveness of substituting cement with fly ash and granite powder up to 30% of the cement volume in composites. Additionally, they designed a neural network model to predict the packing density of these mixtures. The practical value of this approach was demonstrated through life cycle assessment analyses.