Identification tridimensionnelle des lois locales de propagation des fissures par micro-tomographie X, mesures de champ étendus et simulations par éléments finis étendus – RUPXCUBE

RUPXCUBE project aims at developing innovative link between experimental and computational mechanics dedicated to the identification of three-dimensional crack growth laws. The dramatic human and environmental risks resulting from possible failure due, for instance, to rupture in aeronautic or nuclear structures justify the amount of work dedicated to improve their safety. The ability of predicting the lifetime of a component requires both material and mechanical knowledge in experimental, theoretical and numerical fields. Many factors influence the crack growth rate under cyclic loading and lead to retardation or acceleration effects linked to crack closure and small scale yielding, combined loading, bifurcation. Recent experimental facilities like X-Ray synchrotron combined with recent advances in digital image correlation allow for in-situ and continuous monitoring of crack growth in opaque materials and offer new possibilities in the understanding of fundamental fracture mechanisms. In this respect, the European Synchrotron Radiation Facility (ESRF ' Grenoble) is one of the best places in the world to access and use such in situ experiments dedicated to three-dimensional image acquisition. Furthermore, three-dimensional image correlation is the natural interface able to connect directly experimental observations to a mechanical formalism that will in turn be used as input data for numerical simulations. This synergy to which we have recently contributed is unique and original and will lead to significant progress in the understanding of crack growth mechanisms. In this respect, in order to analyze the long-term reliability of structures, there is a need for robust and validated 3D crack propagation models and numerical procedures. This effort involves a number of specific challenges on the numerical modelling side because of the very demanding cost of three dimensional approaches. Validation itself appears to be even more challenging. However, because of the wealth of information provided by full field kinematic measurements in three dimensions, the same experimental data can be used both for the identification and validation purposes. The latter point is a remarkable feature of the proposed approach that which should allow for a fine tuning of the compromise between complexity of the description and consistency with the experimental data. The present study illustrates one of the routes that can be followed to achieve a direct link between experimental and computational mechanics.