In order to develop an identification strategy for continuum constitutive models devoted to quasi-brittle materials, suited for structural analysis, often realized arbitrarily, a model based on the discrete element method has been formulated. The discrete model is used to compensate the lack of experimental data required to calibrate the continuum model. Thanks to intrinsic predispositions with respect to fracture mechanisms, the discrete model can be used easily, and its efficiency has been proved. However, only 2D simulations have been undertaken so far, mostly due to computational costs limitations.

A 2D framework reduces extensively analysis possibilites with such model, in particular for reinforced structures where 3D effects are predominant. The purpose of the present post-doctoral work is to extend to 3D the discrete approach already developped in 2D. The developments will be integrated in the FEA code CAST3M-CEA developped by DEN/DANS/DM2S/SEMT. In the mean time, the discrete model will be optimized using available tools, such as solvers, available in the CAST3M-CEA environment. Depending on the computational costs improvements, even complete structures simulations might be considered.

At the end of this work, the developed numerical tool will allow to extend the identification stragegy to constitutive models including 3D effects, such as steel/concrete interface models (confinement) and concrete model (dilatancy).