Fracture initiation in multi-phase materials: a systematic three-dimensional approach using a FFT-based solver

TL;DR

This study employs a fast Fourier transform-based method to analyze three-dimensional microstructures of multi-phase materials, revealing how phase arrangements influence fracture initiation and highlighting differences from 2D models.

Contribution

It introduces a computationally efficient 3D FFT-based approach for analyzing microstructure-fracture relationships in multi-phase materials.

Findings

Fracture initiates where hard phases are interrupted by soft phase shear bands.

Hard phase boundary orientation influences high tensile stress regions.

3D analysis shows differences in strain partitioning compared to 2D models.

Abstract

This paper studies a two-phase material with a microstructure composed of a hard brittle reinforcement phase embedded in a soft ductile matrix. It addresses the full three-dimensional nature of the microstructure and macroscopic deformation. A large ensemble of periodic microstructures is used, whereby the individual grains of the two phases are modeled using equi-sized cubes. A particular solution strategy relying on the Fast Fourier Transform is adopted, which has a high computational efficiency both in terms of speed and memory footprint, thus enabling a statistically meaningful analysis. This solution method naturally accompanies the regular microstructural model, as the Fast Fourier Transform relies on a regular grid. Using the many considered microstructures as an ensemble, the average arrangement of phases around fracture initiation sites is objectively identified by the correlation between microstructure and fracture initiation -- in three dimensions. The results show that fracture initiates where regions of the hard phase are interrupted by bands of the soft phase that are aligned with the direction of maximum shear. In such regions, the hard phase is arranged such that the area of the phase boundary perpendicular to the principal strain direction is maximum, leading to high hydrostatic tensile stresses, while not interrupting the shear bands that form in the soft phase. The local incompatibility that is present around the shear bands is responsible for a high plastic strain. By comparing the response to a two-dimensional microstructure it is observed that the response is qualitatively similar (both macroscopically and microscopically). One important difference is that the local strain partitioning between the two phases is over-predicted by the two-dimensional microstructure, leading to an overestimation of damage.