Machine learning can predict setting behavior and strength evolution of hydrating cement systems

Setting and strength development of ordinary portland cement (OPC) binders is a complex process that involves multiple interacting chemical reactions, which result in the formation of a solid microstructure. A long-standing yet elusive goal of the cementitious materials community has been to establish a basis for prediction of the properties and performance of concrete using knowledge of the chemical and physical attributes of its components - OPC, sand, stone, water, and chemical admixtures - together with the environmental conditions under which they react. Machine learning provides a data-driven basis for the estimation of properties, and has recently been applied to estimate the 28 d (compressive) strength of concrete simply from knowledge of its mixture proportions [1]. Building on this success, the current work uses a diverse dataset of different ASTM C150 cements, the chemical composition and other attributes of which have been measured. Machine learning (ML) estimators were trained with this dataset to estimate both paste setting time and mortar strength development as a function of the OPC composition and fineness. The ML estimation errors are typically similar to or lower than the measurement repeatability of the relevant ASTM test methods. ML therefore can be used to estimate the influence of binder composition and fineness on the engineering properties of cementitious systems. This creates new opportunities to apply data intensive methods to optimize concrete formulations under multiple constraints of cost, CO₂ impact, and performance attributes.