Comparison of reconstructed spatial microstructure images using different statistical descriptors

The objective of this research is to examine the use of statistical descriptors in the microstructure reconstruction process and the associated effect on microstructure properties. In this work, two-point correlation functions and lineal path functions were used to study the efficiency and accuracy of microstructure reconstruction. A polar plot of the two-point correlation functions is used to quantitatively evaluate the agreement with the target microstructure. In general, as more microstructure descriptors are added for the microstructure reconstruction process, the computational time increases and the reconstructed microstructures quantitatively agree better with the target microstructure. However, in some cases, adding lineal path functions did not always result in a better microstructural representation. Last, finite element analysis of thermal loading in the reconstructed microstructures of loaded nuclear waste glass is used to evaluate the property predictions with those of the target microstructure. Identifying which statistical descriptors are best for different microstructures may provide insight into optimizing the microstructure reconstruction process from limited statistical descriptors.