Thermodynamic modeling of cementitious material is an established tool for predicting the hydrated phase assemblages, pore solution pH, and pore solution composition of mixtures of various chemical compositions and water-to-binder (w/b) ratios. However, traditional thermodynamic techniques have major limitations for modeling mixtures containing supplementary cementitious materials (SCM), and when modeling reactions at non-equilibrium conditions. Thermodynamic modeling using a Gibbs Energy Minimization framework computes the phase assemblages and speciation of a system based on its bulk elemental composition when the spontaneous energy of the system is at a minimum. However, cementitious systems are inherently non-equilibrium as they react. Furthermore, crystalline components of SCM generally are non-reactive in cementitious systems, so using the bulk composition of these materials in thermodynamic calculations may produce inaccurate results. Most modern-day cementitious binders contain SCM, and SCM can influence the reaction pathways and products in the binder. Because the durability and performance of cementitious mixtures depends in large part on which reaction products form, and the pore solution chemistry and pore size distribution of the mixture at different ages, there has been a clear need to develop models to incorporate SCM at equilibrium and non-equilibrium conditions into thermodynamic calculations, so that reaction products and pore solution can be accurately simulated at any age for any mixture. The work presented in this dissertation describes several models that contribute to filling these gaps in knowledge.
A method to predict the pore size distribution of binders containing SCM is presented, based on an assumed equilibrium reactivity (DoR*) of the SCM (the pore partitioning model or PPM). The partitioning of pore sizes is accomplished based on the modification of a model (Powers-Brownyard) for ordinary portland cement using thermodynamic calculations. The ratio of small gel pores to larger capillary pores in binders contributes to strength gain and resistance to damaging processes in concrete.
Accurate predictions from the PPM require accurate input values for DoR*. Also demonstrated in this dissertation is method to measure DoR* of pozzolanic SCM such as fly ash and silica fume. The reactivity test method combines experimental measurement of heat release and the consumption of calcium hydroxide (CH) in a simplified synthetic pore solution with thermodynamic calculations of idealized systems of silica and alumina at degrees of reaction ranging from 0% to 100%. The thermodynamic calculations provide reference lines against which the measured values can be read, providing a single numerical value for overall DoR* of a given SCM.
The results of the reactivity test method confirm literature reports that SCM DoR* is highly variable, and not strictly related to the bulk elemental composition of the material. While factors contributing to the DoR* of SCM are many, the ratio of crystalline-glassy phases is known to play a major role. Also demonstrated in this dissertation is a method to determine the reactive phases in fly ash based on subtracting the ash crystalline components as measured by quantitative x-ray diffraction (QXRD) from the bulk composition of the system as measured by x-ray fluorescence (XRF). It is shown that the accuracy of thermodynamic calculations is improved when only reactive components of fly ash are modeled, particularly for ashes with high proportions of crystalline material.
While accurate measurement and modeling of DoR* of SCM at equilibrium provides important information from which to determine durability-related properties of hydrated pastes, it is also important to understand the reactivity of the individual phases in the SCM at non-equilibrium conditions. Also demonstrated in this thesis is a kinetic framework for non-equilibrium thermodynamic modeling of pastes containing silica fume and fly ash. The kinetic model (MPK model) is based on a modification of the Parrot-Killoh model initially developed for ordinary portland cement. The model is an empirical model where the rate limiting step for each phase is based on the slowest of nucleation, diffusion, and reduction in ionic transport. Empirical constants for the MPK model are determined using a non-linear numerical algorithm fit from dissolution data for individual SCM phases described in literature. Using these empirical constants and a measured DoR* for each phase based on QXRD (DoRph*), the MPK model is shown to accurately predict the formation of hydration products and pore solution for mixtures containing fly ash and silica fume at different ages.