Identifiability analysis of material identification using nonlinear VFM

Abstract. The success of inverse material model identification depends on the interaction between the adopted material model, the design of the heterogeneous specimens, the quality of the full-field measurements and the employed inverse identification method. Although inverse identification with full fields usually uses either FEMU or nonlinear VFM algorithms, a range of specimen designs and heterogeneity indicators have been proposed to assess the quality of the measured field and specimen design. While many studies investigate the effects of strain field heterogeneity on material model identification, few of them address the comprehensive interaction of all the above features and investigate their interactions during inverse identification through identifiability analysis. In this study, we analyze the identifiability of the parameters of the YLD2000-2d model used to describe the plastic anisotropy of steel sheet DC04 using a perforated biaxial specimen with the nonlinear VFM method. For this purpose, we performed a virtual DIC experiment with known material parameters by simulating the test in ABAQUS/Standard, generating synthetic images and reconstructing the strains via stereo DIC. Before inverse identification with a nonlinear sensitivity-based VFM, we analyzed the sensitivity of the virtual work to parameter changes and performed an identifiability analysis.