Answers are grounded in this paper and cite their sources.
Accurate crash simulations require robust regularization strategies to obtain mesh-independent predictions of ductile damage and fracture on coarse shell meshes. Although the GISSMO damage model is widely used to model stress-state-dependent failure, its regularization is typically calibrated from tensile tests only and relies on prescribed bi- or trilinear stress-state dependence. This work proposes a numerical calibration procedure that enables stress-state-dependent regularization for shell elements over the considered range of plane-stress triaxialities assuming isotropic elastic behavior and von Mises plasticity with isotropic hardening. Constant-triaxiality loading of rectangular element blocks is employed to generate well-defined stress states, while traction boundary conditions are constructed from a preliminary displacement-driven simulation to allow localization to develop. Using Swift’s maximum-force criterion to identify the onset of diffuse necking, the mechanical work is decomposed into pre- and post-localization contributions. Regularization factors are obtained by matching the post-localization work of coarse meshes to a reference mesh, yielding stress-state-dependent failure curves. The method is demonstrated on an aluminum sheet material and compared against established GISSMO regularizations with bilinear and trilinear triaxiality dependence. Validation on constant-triaxiality blocks, tensile and Nakajima specimens as well as a profile crush test show that the proposed regularization provides comparable or improved mesh convergence, while retaining physically motivated behavior at the end points such as negligible regularization in shear and biaxial tension. The results further highlight fundamental limits of regularization for very coarse meshes, where geometric representation errors dominate the response.
No takes yet. Share an insight, caveat, or question.
Zink et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: