The accurate prediction of concrete composites’ fracture behaviour requires precise meso-structural characterization, appropriate fracture modelling, and pivotal uncertainty assessment. Towards this goal, we adopt a dynamics approach with CT images to generate numerical concrete models using real aggregates with 30%-60% contents. An image slicing method is then developed to obtain a large number of realistic models in 2D, whose heterogeneity is characterised by aggregates, mortar and interfaces. Monte Carlo simulations of uniaxial tension are performed, incorporating cross-scale fracture evolution through a phase-field cohesive zone model. Statistical analyses reveal that the stochasticity of crack patterns and stress-displacement curves, especially post-peak softening responses, is inherently dependent on the random meso-structures. It is observed that increasing aggregate content accelerates the deterioration rate of peak stress, decreases the softening curve, and leads to more tortuous crack paths, considering the ITZ crack channels. On the other hand, the tensile strength of mortar has significant impacts on the load-carrying capacity, while the fracture energy primarily influences the ductility and has negligible effects on the peak stress and the pre-peak nonlinear part. Besides, the ratios of mortar-ITZ tensile strength and fracture energy are found to affect the crack paths. The proposed numerical framework holds promise to reveal complex fracture mechanisms of concrete with more insights into other loading or environmental conditions.
Read full abstract