Identification of material parameters of the Rousselier model by non-linear optimization

This work is concerned with identification of material parameters for inelastic deformation laws. In this context, non-linear boundary and initial value problems are solved using the developmental finite element code SPC-PMHP for parallel computers. The ductile damage model of Rousselier for large elasto-plastic strains is implemented as a system of non-linear differential and algebraic equations. For solving the inverse problem, the solution of the direct problem is embedded in a gradient based method. This way, material parameters could be identified analysing inhomogeneous two-dimensional displacement fields. Deterministic optimization procedures are used to identify parameters by means of a least-squares functional. A semi-analytical sensitivity analysis was adopted to calculate the gradient of the objective function. Numerical experiments with synthetically generated displacement fields were carried out to check the algorithm. The identification procedure was successful when one material parameter is allowed to vary. Experiments with two or more unknown parameters were less successful, because in some cases only a local minimum was found.