Résumé:
In a context of sustainable development, sustainable construction promotes the protection of the environment, the preservation of natural resources and the reduction of construction and maintenance costs for structures. In the design phase, the use of simulation allows the assessment of long term performance of structures and the prediction of possible disorders. However, it is common to observe differences between forecast and actual performance due to the presence of uncertainties in the data and simulation methods.In this thesis, a model was developed for the analysis of the durability of reinforced concrete structures subject to chloride induced reinforcement corrosion, taking into account the random variability of exposure conditions and the presence of uncertainties in the geometry and material properties, with special emphasis on epistemic uncertainties characterized by a lack of information. The modeling was undertaken in two phases. A deterministic model of chloride penetration in concrete was first developed. Then, the effect of uncertainties and variability in exposure conditions was included for the estimation of corrosion initiation time. The chloride penetration model was based on the simulation by the finite element method of the phenomena of heat transfer, moisture and chloride transport. The epistemic uncertainties were modeled by fuzzy probability functions and the reliability analysis was performed by adopting a mixed fuzzy Monte Carlo approach. Environmental conditions were modeled stochastically to reflect their variability over time.The numerical performance of the developed model was examined through several examples aimed at investigating the sensitivity to the different material parameters, the effect of climatic conditions, the effect of uncertainties and the variability of environmental conditions.
