For her research, 3-D Multi-Scale Modeling Combined with Machine Learning for a Novel Structural-Prognosis Framework, mechanical engineering assistant professor Ashley Spear received a three-year $339,129 grant from the AFOSR (Air Force Office of Scientific Research).

The goal of this research is to enhance predictive capabilities for structural-materials performance by incorporating novel 3-D microstructurally small fatigue crack (MSFC) data into a combined multi-scale-modeling/machine-learning approach. This new approach will lead to quantitative, physically based models that will enable, for the first time, the ability to predict with high fidelity and in a probabilistic context the morphological evolution of 3-D MSFCs as a function of microstructurally dependent fields. The multi-scale modeling effort will leverage a novel experimental data set that was collected recently using synchrotron-based measurements (Spear et al., 2014). The recent experimental observations showed a significant variation in the rate of MSFC growth along individual crack fronts, i.e. MSFC morphology evolved in a very non-self-similar manner. This suggests that idealized descriptions of cracks (e.g. planar, self-similar) are not accurate or adequate in representing fatigue cracks at the microstructural length scale. Thus, one of the salient aspects of this work will be the incorporation of real (as-measured) MSFC morphologies into a high-fidelity, microstructure-based model to probe the fields in the neighborhood of the evolving MSFC. Machine learning will then be employed as a way to discover a quantitative model relating the local fields to the observed MSFC morphological evolution. Subsequently, the learned model can be used to predict the complex shape evolution of 3-D MSFCs.


To learn more about Dr. Spear and her research please visit her lab site, Multiscale Mechanics & Materials Laboratory.