Contribution to a better understanding of long-term hydration, structuration and mechanical properties of slag based cementitious materials: Experimental and modeling approaches

Abstract

Blast Furnace Slag (BFS), a by-product of the steelmaking process, can be used as a mineral addition in blended cement. In addition to reduce CO2 emissions in the construction field, the use of BFS in cementitious materials also contributes to the improvement of their mechanical and durability properties. However, despite the numerous studies proposed in the literature about the hydration properties of BFS-blended cement, some aspects remain unclear, such as the influence of slag substitution rate on hydration kinetics and on the stoichiometry of hydrates formed and their impacts on both microstructural and mechanical properties. For these reasons, the present study aims firstly to better understand how BFS-blended cement hydrates based on a complete hydration model that considers both kinetics and chemical reactions. The only required input data are the formulation parameters (i.e. water-to-binder ratio (W/B), substitution rate of clinker by slag (fs)), and curing. This study does not introduce the heat of hydration, which limits the validity of other existing models to short-term hydration. Analytical laws involving calibration parameters, which are assumed to vary with the substitution rate of clinker by slag, represent the hydration kinetics of clinker and slag while a stoichiometric approach is adopted to describe the chemical formation of hydrates. In order to calibrate and validate the parameters involved in the hydration model, four different cement paste mixes formulated with 0%, 30%, 50%, and 80% (by mass) of slag are studied in terms of hydration degrees and microstructure evolutions, and completed with data from the literature. The amounts of hydrated phases and those of dry densities, particularly at later ages, are well reproduced by the hydration model. Porous structure evolutions are also investigated in terms of partition between capillary and CSH gel pores by comparing the capillary porosity predicted by the model with the total porosity measured experimentally. Finally, predictive modeling of mechanical properties (compressive strength and Young modulus) of tested materials are finally proposed based on hydration model outputs. A good agreement between experimental and predicted data is observed.