|About this Abstract
||2016 TMS Annual Meeting & Exhibition
||ICME Infrastructure Development for Accelerated Materials Design: Data Repositories, Informatics, and Computational Tools
||Analytics on Large Microstructure Datasets Using 2-pt Statistics
||Ahmet Cecen, John W. Gibbs, Peter W. Voorhees, Surya R. Kalidindi
|On-Site Speaker (Planned)
With the rapid development of numerous experimental techniques and imaging tools, there is now a critical need for efficient strategies to distill useful insights from massive amounts of 3D and 4D materials datasets. Spatial correlations in the form of n-point statistics have been shown to be very effective in addressing this need. In this work, we present a versatile framework for the computation of 2-pt statistics for a very broad class of microstructures, while accommodating diverse boundary assumptions, types and kinds of local states, masking conditions, and efficient resource utilization especially for massive materials datasets. We demonstrate these strategies through application to a large 3D microstructure snapshot of an Al-Cu solidification obtained using X-ray tomography techniques. In particular, we extract new physical insights regarding the spatial arrangement of the solid-liquid interfaces that improve our understanding of the underlying phenomena governing the evolution of the solid-liquid mixture.
||Planned: A print-only volume