Generating 3D virtual microstructures and statistically equivalent RVEs for subgranular gamma-gamma’ microstructures of nickel-based superalloys

Machine Learning-Aided Parametrically Homogenized Crystal Plasticity Model (PHCPM) for Single Crystal Ni-Based Superalloys

“…For calibrating the two-phase dislocation densitybased crystal plasticity model, a SERVE containing 50 precipitates is generated for the SC-16 superalloy, 26,27 for which the loading conditions follow the experiments. The GA with SVR-assisted acceleration optimization strategy is applied to the 13-D material parameter space.…”

“…The data extraction process uses a highthroughput experimental system that couples a focused ion beam (FIB) and a scanning electron microscope (SEM) to extract serially sectioned c À c 0 micrographs from a single crystal of René 88DT. 26,27 The section images are stacked and registered for reconstructing a 3D voxelized microstructure, and a sequence of image processing methods is applied to the data set to segment the precipitates. 27 The 3D volume thus constructed is statistically characterized by quantifying microstructural distributions of important descriptors such as precipitate size, shape, position and c channel width.…”

“…26,27 The section images are stacked and registered for reconstructing a 3D voxelized microstructure, and a sequence of image processing methods is applied to the data set to segment the precipitates. 27 The 3D volume thus constructed is statistically characterized by quantifying microstructural distributions of important descriptors such as precipitate size, shape, position and c channel width. With these morphological and spatial statistics, microstructure generation algorithms have been developed to create SERVEs in.…”

“…With these morphological and spatial statistics, microstructure generation algorithms have been developed to create SERVEs in. 26,27 The SERVEs represent the optimal computational domain of the c À c 0 microstructure for a range of morphological statistics and mechanical responses. II.…”

“…Example statistical distributions that distinguish c 0 precipitate morphology include the size, aspect ratio and degree of sphericity, while characterization of c 0 precipitate spatial configurations include the two-point correlation function, local volume fraction distribution, channel-width distribution and distance-to-edge distribution. 27 For parametric homogenization leading to the PHCPMs, it is prohibitive to incorporate detailed microstructural representation to the higher scale. It is important to have an optimal set of reduced parameters for representing the microstructure, while retaining the accuracy of microstructural influence on response prediction.…”