The mathematical modeling of failure mechanisms in solid materials and structures is
a long standing problem. In recent years, peridynamics has been used as a
theoretical basis for numerical studies of fracture initiation, evolution and
propagation. In order to investigate damage phenomena numerically, suitable
material and damage models have to be implemented in an efficient numerical
framework. This framework should be highly parallelizable in order to cope
with the computational effort due to the high spatial and, depending on the
problem, temporal resolution required for high accuracy. The open-source
peridynamic framework Peridigm offers a computational platform upon which
new developments of the peridynamic theory can be implemented. Today,
isotropic material models and a very simple damage model are implemented in
Peridigm.
This paper proposes three energy-based damage criteria. The implementation
approach as well as the extension of Peridigm with these physically motivated models
is described. The original criterion of Foster et al. is adapted for ordinary
state based material. The other two criteria utilize the decomposition of
peridynamic states in isotropic and deviatoric parts to account for the failure-mode
dependency.
The original criterion is verified by the numerical simulation of two mechanical
problems. At first, a virtual double cantilever beam (DCB) experiment is performed
to determine the energy release rate. This value is the fundamental material property
required for the proposed criteria. Additionally, the DCB problem is then used to
investigate the convergence of the numerical scheme implemented in Peridigm. In a
second step, a model of a plate with a cylindrical hole under tensile loading is
compared with an extended finite element method solution. Results of both
numerical solutions are in good agreement. Finally, a fiber reinforced micro
structure model is used to analyze the effect of the different criteria to the
damage initiation and crack propagation under a more complex loading
condition.