ASSESSMENT OF MODEL PARAMETERS FOR FRACTURE SIMULATION IN BRITTLE DISORDERED MATERIALS LIKE CONCRETE AND ROCK

Abstract

Different fracture models for brittle disordered materials require varying empirical c ntent in the form of model parameters that, somehow, must be measured in an experiment. The success of a model to ‘predict ’ situations hitherto unexplored by experiment depends on the correctness of the empirical content of the model, i.e. on the success in capturing material behaviour in the said model parameters. For fracture this seems only possible when the model is capable of simulating fracture mechanisms to a high degree of accuracy in the first place. Different models may be applicable at different scale levels, use may be limited by constraints from the roughness of the material structure itself (RVE), but in the end when fracture becomes fatal, specimens/structures are separated into two or more distinct parts separated by localised macrocracks of size similar to the characteristic size of the considered specimen/structure. In particular deriving correct empirical content for the localisation stage is extremely difficult due to intertwining of material and structural effects. In classical fracture mechanics these effects are elegantly taken care of in the structure of the theory, but in models based on continuum principles, the link between material and structure is lost, resulting in tremendous difficulties in determining the empirical content. The solution must be found in physics-based approaches, which are based on correctly simulating fracture mechanisms in the first place