User login


You are here

Numerical convergence of finite difference approximations for state based peridynamic fracture models

Prashant K. Jha's picture

Prashant K. Jha and Robert Lipton

Computer Methods in Applied Mechanics and Engineering, 2019.



1. Well-posedness of a general nonlinear state based peridynamic models.

2. A priori numerical convergence rate for finite difference approximations of state based peridynamic models.

3. Numerical verification of convergence rate for samples with growing cracks.

4. Simulations of multiple cracks for samples subject to bending load.

5. Numerical experiments demonstrating that the increase in peridynamic energy of the evolving damaged region is the same as the classical Griffith energy release rate.



In this work, we study the finite difference approximation for a class of nonlocal fracture models. The nonlocal model is initially elastic but beyond a critical strain the material softens with increasing strain. This model is formulated as a state-based peridynamic model using two potentials: one associated with hydrostatic strain and the other associated with tensile strain. We show that the dynamic evolution is well-posed in the space of Hölder continuous functions C0,γ with Hölder exponent γ ∈ (0,1]. Here the length scale of nonlocality is ε, the size of time step is Δt and the mesh size is h. The finite difference approximations are seen to converge to the Hölder solution at the rate Ct Δt + Cs hγ/ε2 where the constants Ct and Cs are independent of the discretization. The semi-discrete approximations are found to be stable with time. We present numerical simulations for crack propagation that computationally verify the theoretically predicted convergence rate. We also present numerical simulations for crack propagation in pre-cracked samples subject to a bending load.



PDF icon Preprint (To appear in CMAME)1.45 MB
Subscribe to Comments for "Numerical convergence of finite difference approximations for state based peridynamic fracture models"

Recent comments

More comments


Subscribe to Syndicate