Modélisation de l’amortissement dans les analyses dynamiques non-linéaires de structures en béton armé : modèles matériaux et identification expérimentale

In addition to the new seismic zoning of the French territory in 2011, the Fukushima nuclear accident the same year prompted the French government to focus on the safety of nuclear buildings. To ensure the viability of such structures, performative models are required, so they must integrate fine physical phenomena descriptions. In the case of reinforced concrete structures, the significant difficulty comes from the lack of knowledge about the damage evolution and the ability of concrete to dissipate energy.Since the 1970s, damping has been considered a predominant element in analysing structures under dynamic loading. For reinforced concrete structures, many sources of damping act in parallel. When a structure is subjected to an earthquake, it dissipates energy by hysteresis at the material scale and other phenomena modelled by viscous damping (global scale). The bibliographical review at the beginning of the thesis aims to treat the notion of damping in the literature with more or less complex proposed models and the identification of this quantity to improve the models.Various problematic relating to the concept of damping are of interest in this PhD work: In the framework of nonlinear dynamic computations, which local and global viscous damping formulations best represent the experimental structural response? How are local and global damping energy dissipation mechanisms evolving during nonlinear dynamic computations? How could we improve the damping modelling at the local scale, on a physical basis, to reduce the requirement of arbitrary equivalent viscous damping at the global scale?A multi-fibre model is developed in Cast3M software from experimental data on reinforced concrete beams to answer these research questions. Nonlinear behaviour models for concrete are used. Experimentally, the steel reinforcement is not plasticised, so the work focuses on concrete cracking when the steel reinforcement remains in the linear domain. The numerical model is calibrated on quasi-static tests. Sixteen formulations of classical damping matrices are then studied, on dynamic tests, with damping rates varying from 0.5% to 5%. The responses of the different models are compared with the experimental responses and energy analyses. Therefore, it appears necessary to model sufficient physical phenomena at the local scale (cracking, friction, unilateral effect) to obtain results consistent with the experimental responses.In the last chapter, a damping identification method is developed on a one degree-of-freedom model equivalent to the multi-fibre model of the reinforced concrete beam. The objective is to identify the transient evolution of the viscous damping rate in parallel with the beam's damage. The main conclusion is that the viscous damping rate evolves exponentially with respect to a damage index defined from the degradation of the beam stiffness. However, if the damage value is less than 0.6, considering a constant damping rate equal to 4% seems adequate. Finally, a locally updated damping model is proposed based on the development of nonlinearities in concrete. The advantage of these models is their physical basis at the local scale of the material, in addition to their representation of the experimental results. The main influencing phenomena are crack development (damage) and friction in cracks.