Modelling hydration process of cement nanoparticles by using an agent-based molecular formation algorithm.

Abstract

This paper presents an agent-based modelling approach for hydration process of Portland cement taking place at the nanoscale. By following a combination of stochastic, deterministic and adaptive rules, cement system is allowed to evolve from an anhydrous separated state to a bond state with enhanced mechanical properties. We consider the two main chemical reactions producing C-HS gel and portlandite, and identify the related anhydrous cement components as two autonomous agents, while water and hydration time are viewed as activator agents. The material representation consists of a geometrical configuration based on a two-dimensional granular fluid, which shows to be suitable when studying nanostructure formation. The resulting agent-based model is tested on a 103μm x 103μm thin film with 20μm thickness containing anhydrous cement. We evaluate the modelling approach after 7 days of hydration from comparison with experimental results, and agreement is good.