Computational model to estimate chloride diffusivity by ternary random walk simulation on tomographic 3D microstructure of cement matrix

This study proposed a model to estimate the chloride diffusivity of the cement matrix by performing a ternary random walk simulation, which considers both connected and disconnected pores in tomographic 3D micro-structure. Ternary structures of cement paste, consisting of pore, hydrate, and anhydrous phases, were obtained from X-ray micro-CT images by calculating each phase's representative linear attenuation coefficient based on the quantitative X-ray diffraction results. A novel random walk simulation introducing the concept of hydrate retention time upon ternary microstructures was applied to cement paste. From the developed ternary random walk simulation, the values of the hydrate retention time were determined considering multi-scale porosities and the experimental data of chloride diffusivity. The results showed that the hydrate retention time ranged from 325 to 108 as w/c increased from 0.3 to 0.7. Empirical formulas for estimating hydrate retention time according to the porosity of hydrate voxels and the hydration degree was established. Based on these formulas, the proposed model performing ternary random walk simulation on X-ray micro-CT images enables the estimation of chloride diffusivity of actual cement paste with only the initial w/c as input, making it a practical and accurate tool for durability characterization.