Computational Homogenization Modeling of Dynamic Fragmentation in Additive-Manufactured Porous Rings

The fragmentation behavior of porous metal rings subjected to rapid radial expansion largely depends on the microstructural pore morphology. Incorporating this microstructural information, as characterized by micro-CT scans, into large-scale simulations is essential to predict the fracture behavior. However, this process presents challenges, as explicitly resolving the microstructure is computationally intractable while accessible damage models do not account for the microstructure-specific information. This study therefore uses a novel methodology to create synthetic microstructures from a micro-CT scan of an additively-manufactured AlSi10Mg alloy and subsequently convert each of these microstructures to a single initial porosity value. The Abaqus finite element solver is then used to model a rapidly-expanded ring which is radially seeded with an initial porosity distribution. The porosity evolution and failure criterion is governed by a Gurson-Tvergaard model. We perform a parametric study investigating the effect of the number of synthetic microstructures and the failure porosity value on the fragmentation behavior. We find that prescribing a spatially-varying initial porosity produces fracture patterns that more closely match real-world results compared to a spatially-homogeneous porosity. Additionally, for a spatially-uniform failure porosity, higher values correlate with fewer fragments and larger fragment lengths. A random, spatially-varied failure porosity negligibly affects the fracture behavior. The study’s methodology is then compared against experimental results and we find that we generally predict the number of fragments and fragment length. However, our method under-predicts the thickness reduction observed in experiments.