Elastoplastic parameter identification by simulation of static and dynamic indentation tests

This work presents a numerical optimization procedure for the identification of elastoplastic material parameters by means of inverse analysis, through both static and dynamic indentation tests. A finite element method (FEM) modelling of the indentation test is put in place by analysing first macroscopic effects (indentation curve, residual imprint geometry) at variable constitutive parameters (elastic modulus, yield stress, hardening coefficient). The FEM solver is then linked to an optimization routine by assembling an integrated loop towards the solution of the inverse problem. Later, the FEM solver is replaced by a radial basis function network interpolation of pre-calculated data, combined to a principal component analysis, allowing the reduction on computational burden of each non-linear analysis. Next, a detailed study on the identification procedure is performed by applying it to pseudo-experimental data that is generated numerically prior to the inverse analysis, which is possibly affected by random noise with given variance. The reliability of the inverse procedure is then demonstrated for both static and dynamic indentation tests as a necessary condition for further validations with true experimental data. The information from only the imprint geometry is shown to be sufficient for consistent material parameter identification.