This study presents an extended numerical approach based on GeoMicro3D to simulate the reaction kinetics and three-dimensional (3D) microstructure evolution of alkali-activated fly ash (AAFA). Dissolution experiments were conducted under varying NaOH concentrations and temperatures to formulate predictive rate functions for Si and Al release. These experimentally derived kinetic functions, alongside a thermodynamic dataset for N-(C-)A-S-H gels, were incorporated into the GeoMicro3D model to capture the chemical reactions and 3D microstructure evolution of AAFA. The model well captured reaction degree of fly ash, formation of solid products, evolution of pore solution compositions, and porosity over time. Notably, it is the first to predict the time-dependent spatial distribution of phases within the 3D AAFA microstructure by integrating kinetic and microstructural modeling. Dual validation using both dissolution data and microstructural metrics demonstrates the model's reliability and robustness. This integrated framework provides new insights into the coupled chemical–microstructural evolution of alkali-activated materials.

Previously, the lack of a thermodynamic database for N-(C-)A-S-H gel limited the application of thermodynamic modeling to alkali-activated fly ash (AAFA). This study pioneers thermodynamic modeling of AAFA using a recently developed thermodynamic dataset for N-(C-)A-S-H gel. The reaction products, pore solutions and reaction kinetics of AAFA pastes were experimentally determined. Based on the reaction kinetics, the composition of the solid phases and the pore solution of AAFA were modeled over time. The results showed that the simulated compositions of the solid reaction products and pore solution match closely with the experimental results, especially for the sodium hydroxide-activated system. Moreover, modeling results point out the potential presence of minor reaction products (e.g., C-(N-)A-S-H gel, microcrystalline ferrihydrite, Mg-containing phases) undetectable by experimental techniques. The study also demonstrated that thermodynamic modeling accurately captured the amount of bound water in reaction products, highlighting its robustness in both qualitative and quantitative analysis.