^{1,2,3,a}, F. Duprat

^{1,b}, T. de Larrard

^{1,c}, J. Mai-Nhu

^{2,d}, P. Rougeau

^{2,e}, L. Marracci

^{3,f}, and P. Guédon

^{3,g}

^{1}Université de Toulouse, LMDC INSA-UPS.

^{b}duprat@insa-toulouse.fr

^{c}delarrar@insa-toulouse.fr

^{2}CERIB Epernon.

^{d}J.MAI-NHU@cerib.com

^{e}P.ROUGEAU@cerib.com

^{3}Arcadis ESG, Paris.

^{a}paulo.claude@arcadis.com

^{f}louis.marracci@arcadis.com

^{g}pascal.guedon@arcadis.com

Corrosion of the steel reinforcements in concrete structures is a major cause of their deterioration. In most cases, corrosion is induced by carbonation or chlorination and only the supposedly prominent initiating phenomenon is considered. However, a combination of the two can be more uncertain because, in this case, both the chloride binding capacity and the pore microstructure of the cement paste are affected. The case of a concrete bridge subjected to carbonation and deicing salts is considered in this study. Specific Finite Element Modelling (FEM) was developed in order to estimate the time to reinforcement depassivation and the aim of the study is to estimate the probability of effective initiation of corrosion when the chloride content at the rebar surface exceeds a threshold value. The time, t, corresponding to this event, with a predefined probability, can be a significant milestone in the maintenance policy. Concrete properties, external environment (carbon dioxide pressure, chloride content, relative humidity) and concrete cover depth are considered as random variables in the study. In order to overcome the numerical weight of the FEM in the probabilistic computations, a surrogate model based on polynomial chaos expansion is employed. A Sobol sensitivity analysis is performed on the parameters to observe their influence on the result. Several locations are assumed for the structure, implying various environments and durations of the frost period. The results show that combined exposure to chlorides and carbon dioxide increases the depassivation rate and that the concrete cover thickness has the highest impact on the probability of depassivation.