Fracture propagation by using shape optimization techniques based on outer Riemannian metrics

TL;DR

This paper introduces a shape optimization-based method for simulating brittle fracture propagation in 2D, avoiding diffuse interface parameters and level set functions, and employing Riemannian geometry for improved analysis.

Contribution

It presents a novel fracture simulation approach using shape optimization on Riemannian manifolds, offering an alternative to phase-field models with enhanced analytical and practical benefits.

Findings

The method accurately predicts fracture paths in tension and shear tests.

It produces more realistic fracture paths by incorporating strain splitting.

Results compare favorably with phase-field simulations.

Abstract

In this work, we investigate a novel approach for the simulation of two-dimensional, brittle, quasi-static fracture problems based on a shape optimization approach. In contrast to the commonly-used phase-field approach, this proposed approach for investigating fracture paths does not require a specified `length-scale' parameter defining the diffuse interface region nor a level set function. Instead, it interprets the fracture as part of the boundary of the domain and uses shape optimization algorithms to minimize the energy in the system and therefore describes the fracture propagation directly. Embedding the problem of energy minimization in a Riemannian manifold framework formulated on a suitable shape space, together with the choice of an outer Riemannian metric, yields both advantages from an analytical as well as an applied perspective. Furthermore, an eigenvalue decomposition of the strain tensor is used to produce more realistic fracture paths (the so-called strain splitting), which only allows fracture growth from tensile loads. Numerical simulations for the commonly considered single-edge notch tension and shear test are performed and the results are evaluated in comparison to phase-field results.