A novel development of bi-level reduced surrogate model to predict ductile fracture behaviors

In this paper, a bi-level reduced surrogate model is developed and presented to identify the material parameters of ductile fracture criterion. Using this method, the identification process becomes feasible with only a limited number of experimental tests. The method assembles local critical elements associated with global models. The surrogate model of fracture strain constructed using Diffuse Approximation and the local elements, reduces the computational effort for searching identified material parameters. Global fracture simulations are preformed to update the target fracture strain and to compute the corresponding failure onset displacement. Convincing results are obtained via successive applications of Design of Experiments (DOE) and enhanced design space transformation algorithms. The proposed identification protocol is validated for the modified Mohr-Coulomb fracture criterion using DP590 steel. Robustness of this method is confirmed with different initial guess. It orients a new direction for material parameters identification based on surrogate model which can effectively be implemented to predict the fracture behaviors. (C) 2017 Elsevier Ltd. All rights reserved.