The experimental full-field method (EFFM) for parameter calibration applied on an anisotropic constitutive model

Abstract

Abstract. Accurate characterization of material models is essential to ensure a higher prediction quality in Finite Element Analysis (FEA) under general loading conditions in sheet metal forming. Achieving accurate material model data frequently involves intricate inverse analysis techniques and numerous experimental tests. To overcome the complexities associated with calibration processes, the adoption of optical measuring systems like Digital Image Correlation (DIC) is widespread in material model calibration. The rich information obtained from DIC measurements is often used by material model calibration strategies to calibrate the values of material or model parameters, such as extrapolation of stress-strain curves or Lankford parameters that are not necessarily constant over the entire range of plastic deformation. This study presents the Experimental Full-Field Method (EFFM) as an innovative iterative approach for the calibration of material properties. As a special implementation of Finite Element Model Update (FEMU) [1], the EFFM uses the whole deformation field gained from DIC as boundary conditions [2]. This is then used in an inverse optimization procedure to determine parameters of complex material models. In this research, the EFFM is applied to an anisotropic constitutive model [3] to optimize three flow curves in 0°, 45° and 90° directions w.r.t. the rolling direction and the yield surface exponent, which reflects the polycrystal structure of the sheet material, to define the shape of the evolving yield locus in stress space. This is achieved by a modified tensile test specimen with L-shaped cut-outs which allows a distribution of higher strain values over a wider range of triaxiality values. With the direct use of the experimental deformation field in the FE simulation, displacements and strains are not any more objects of the optimization but only stresses. This also eliminates the step of mapping between experiments and simulations. Moreover, by using implicit time integration the inversion of the stiffness matrix becomes redundant, as positions of all nodes are already predetermined at each time step. These aspects make EFFM faster and more accurate than conventional FEMU.