Identification of material parameters of the Gurson-Tvergaard-Needleman model by combined experimental and numerical techniques

To identify material parameters from suitable experiments it is prevalent to use global informations like force-displacement or force necking curves. The quality of accordance between measured and calculated forces at given displacements can be expressed by a least-squares functional. In this contribution a non-linear optimization method will be presented, which minimizes the least-squares functional by use of a gradient based method. The gradient of this functional is calculated in a semi-analytical sensitivity analysis. To determine the derivatives of the force with respect to the material parameters, the local sensitivities on an intersection will be added together. On this intersection, the total nodal force and the external force have to be equal and the normal displacements have to be independent on the material parameters. The parameter identification is embedded in the finite element code SPC-PMHP for solving non-linear boundary and initial value problems on parallel computers. The Gurson-Tvergaard-Needleman model is used to describe the plastic deformation and damage behaviour of the ductile structural steel StE 690. The developed algorithm is applied to tensile tests with notched cylindrical bars. (C) 2004 Elsevier B.V. All rights reserved.