Mesoscale simulation of C–S–H creep and stress relaxation by discrete element modelling

Creep and stress relaxation are time-dependent phenomena that deteriorate concrete structures, primarily occurring in calcium silicate hydrate, the key binder in Portland cement. Evaluating microstructure development during creep is challenging due to long testing durations. This study employs a novel discrete element method to model creep and stress relaxation in C-S-H, enabling microstructure evolution exploration. The simulation results align well with nanoindentation tests, allowing identification of key factors influencing creep. A microstructure-induced machine learning model is developed to describe the relationship between microstructure and creep deformation to assess the importance of various microstructure indices. This study directly verifies the mechanism by which high pressure accelerates the creep in nanoindentation tests. The influence of microstructure indices on creep is quantitatively analyzed, revealing that penetration depth resulting from reduced modulus has the strongest correlation with creep. A detailed analysis of surface forces offers valuable insights for designing experiments and optimizing material properties.