Identification of hardening behavior using inverse modeling and image correlation

The basic principle of an inverse modeling procedure as it is used for parameter identification, is the generation of a complex and heterogeneous deformation field that contains as much information as possible about the parameters to be identified. One way of obtaining such a non-homogeneous deformation is by making the geometry of the specimen less regular. Another possibility is to make the loading conditions more complex. In this paper both options are actually combined by using a biaxial tensile test on a cruciform specimen in order to identify the parameters of a Swift isotropic hardening law. The yield criterion is modeled by the isotropic Von Mises criterion. The optimization technique used is a constrained gradient based Newton-type routine, which means that in every iteration step, a sensitivity calculation has to be performed in order to indicate the direction in which the parameters are to be optimized. The functional to be minimized is a least-squares expression of the discrepancy between the measured and the simulated strain fields at a certain load. The numerical routines as well as the identification results, based on simulated strain fields, are discussed.