Inverse modelling of constitutive parameters for elastoplastic problems

Abstract

To obtain an improved identification of constitutive parameters to be used in finite element method simulations of elastoplastic deep-drawing processes an inverse method was applied using an explicit finite element code to simulate material tests. This problem was addressed by formulating the constitutive parameter identification as an optimization problem. The method was to minimize the objective function defined as the error between the result from the material test and the result from the finite element simulation. The optimization technique is based on the Levenberg-Marquardt method. The objective function was established in a least-squares sense where the design variables were the constitutive parameters of the material. The inverse method was started and, when a global optimum was reached, a set of constitutive parameters were identified. This was performed for both a linear hardening model and a power-law hardening model. It is shown that the inverse method predicts two models which qualitatively show the same overall characteristics for the investigated material.