Analytics for microstructure datasets produced by phase-field simulations

摘要: Abstract Phase-field simulations have become valuable tools in explorations of the effects processing parameters on internal structure evolution materials a broad range advanced materials. Simulations conducted high performance computers enabled resolution relatively large material volumes and generated big datasets. Although such computations captured faithfully many features observed corresponding experiments, there has not yet been broadly adopted framework to quantitatively analyze predicted structures compare them rigorously with experiments. This paper demonstrates that recently developed for quantification structure, based concepts 2-point spatial correlations principal component analyses (PCA), can address this critical need. It is further demonstrated adoption rigorous help establish objectively modeling choices made simulations.