Development of a robust virtual material for the mechanical behavior, using homogeneous and heterogeneous tests

The subject of this PhD takes place within the AutoMeCal project, for Automated Mechanical Lab’ and Model Calibration, which is supported financially by the French National Research Agency (ANR), for 4 years starting in 2024. The aim of this project is to create an automated and intelligent tool where the input is a chosen new material and the output is the calibrated model, i.e., the optimized set of parameters for a given phenomenological model, that is accurately representative of the mechanical behavior of the material. This project is limited to thin sheet metal products, as used in automotive (steel and aluminium alloys) and electronic (copper alloys) industries. The research challenges are the automated mechanical testing itself, for several mechanical states, and the robustness of the automated model calibration and validation.

Developing a robust automated model calibration is the research hypothesis of the PhD, using an automated mechanical testing unit, that is also under development in the AutoMeCal project, using both homogeneous and heterogeneous tests. It is therefore related to mechanical engineering and design, material mechanical behavior, mechanical modelling, inverse methods for material parameters identification and robotization. The aim of the PhD work is to propose an automated method for model calibration based on an inverse methodology, involving strategies when dealing with many parameters like identifiability analysis, homogeneous tests versus heterogeneous ones, optimal combination of several tests, validation of the parameter set. A first stage includes an analysis of the state of the art, a selection of a mechanical model able to represent anisotropy, hardening and rupture of sheet metals, in relation to the experimental database built with automated testing and, if required, with other additional tests. Then, a robust calibration strategy must be defined, based on identifiability analysis to build a relevant experimental database, on combination of algorithms and if necessary, on knowledge-based analysis. Finally, several sets of parameters are expected to be obtained and a validation stage must be chosen and tested.