Comparison of peridynamic and phase-field models for dynamic brittle fracture in glassy materials

We report computational results obtained with three different models for dynamic brittle fracture. The results are compared against recent experimental tests on dynamic fracture/crack branching in glass induced by impact. Two peridynamic models (one using the meshfree discretization, the other being the LS-DYNA’s discontinuous-Galerkin implementation) and a phase-field model lead to interesting and important differences in terms of reproducing the experimentally observed fracture behavior and crack paths. We monitor the crack branching location, the angle of crack branching, the crack propagation speed, and some particular features seen in the experimental crack paths: small twists/kinks near the far edge of the sample. We discuss the models’ performance and provide possible reasons behind the failure of some of the models to correctly predict the observed behavior.