V. L'Hostis, F. Foct and D. Féron
J. Phys. IV France 136 (2006) 151-158
Modelling of tests performed in order to evaluate the residual strength of corroded beams in the framework of the benchmark of the rance beamsA. Millard and M. Vivier
Laboratoire de Mécanique, Systèmes et Simulation, CEA, CEN Saclay, 91191 Gif-sur-Yvette, France
(Published online 22 December 2006)
The Benchmark of the Rance beams has been organised in order to evaluate the capabilities of various modelling tools, to predict the residual load carrying capacity of corroded beams. The Rance beams have been corroded in a marine environment for nearly 40 years. Different types of prestressed beams, made of different types of cement, have been subjected to four points bending monotonous and cyclic tests as well as direct traction tests. The tests have been carried on up to failure, in order to evaluate the residual carrying capacity of the beams. Different teams have participated to the blind prediction of the tests results. In this framework, the CEA/DM2S/LM2S team has performed bidimensionnal modellings which are described in details in this paper. The various constitutive elements of the beams are represented : for concrete, the isotropic Mazars' damage model is used, in a non local version, for prestressing and passive steels, an elasto-plastic strain hardening model is adopted. The corrosion effects, taken into account for the longitudinal rebars, are derived on one hand from the measurements performed on the beams after the tests, and on the other hand from the literature. They consist mainly in a reduction of the rebars cross-section, as well as in their ductility. In principle, the properties of the bond between steel and rebars are also modified by the corrosion. Here, because of the unavailability of specific data on the smooth rebars of the Rance beams, the bond has been modelled by means of specific joint finite elements. The load carrying capacity has been calculated for the monotonous as well as the cyclic tests. Moreover, a sensitivity analysis has been performed, by considering variants where either the rebars are sane, or they have only reduced sections, with their original ductility. The results are compared to the experimental database, and discussed.
© EDP Sciences 2006