An explicit approach to stochastically modeling fatigue crack formation

Michael Veilleux, Cornell University

Recent advances in computational and experimental capabilities have provided the opportunity to model accurately fatigue damage at multiple length scales, ranging from microns to meters. This presentation is part of a larger project where the ultimate goal is to develop such computational capabilities for metallic air and space vehicle components. Specifically, this project is developing a multiscale, mechanistic approach encompassed within a geometrical, probabilistic, hierarchical procedure. Statistically accurate geometries and physics are being represented at two length scales: the microstructural scale, on the order of 10-6 to 10-3 meters, and the component scale, on the order of 10-3 meters and larger.

The main thrust of the work presented here is toward the creation of a computational framework that explicitly models fatigue crack formation at the microstructure scale, with the test-proof material being aluminum alloy 7075-T651. Computational methods are presented that generate and discretize statistically accurate microstructure geometry models, and explicitly simulate fatigue crack formation using physics-based criteria. These methods are validated through direct comparisons to experimental observations.

Abstract Author(s): M G Veilleux, J D Hochhalter, J E Bozek, P A Wawrzynek, and A R Ingraffea