58
S. Ghosh et al.
The SERVE is further classified into a microstructure-based SERVE or MSERVE and a property-based SERVE or P-SERVE in [29–31]. Similar clarifications
have been proposed in [32]. The M-SERVE is defined as one for which morphological and crystallographic characteristics of the microstructure are the sole
determinants of the representative volume. Convergence tests are needed for
determining the minimum SERVE size necessary for the statistics of characteristic
microstructural descriptors to converge. The M-SERVE is the foundation of the
DREAM.3D software [33, 34]. On the other hand, the P-SERVE is determined from
convergence of material response functions and properties.
Development of the M-SERVE and P-SERVE is preceded by the creation of
3D statistically equivalent virtual microstructures or SEVMs. The statistics of
morphological descriptors in SEVMs must be equivalent to those derived from
high-fidelity datasets obtained from experiments involving electron backscatter
diffraction (EBSD) [34, 35], scanning electron microscopy (SEM), or computed
tomography (CT). Deterministic models, e.g., in [36], are not well-suited for
microstructures with significant spatial variations. Various methods of generating
polycrystalline or polyphase microstructures have been proposed in the literature,
e.g., in [37–43]. These methods account for spatial variations in microstructural
morphology and represent the microstructure with statistical distributions functions
equivalent to those obtained from experimental observations. Conversely, other
methodologies rely on a multiphase field approach to generate microstructural
statistics for both polycrystalline and polyphase microstructures [44, 45]. These
methods focus on the modeling and subsequent statistics of the formation of
microstructural geometries and complement the top-down approach of observing
and matching the microstructural statistics. In [34, 35], statistically equivalent
polycrystalline microstructures have been developed to capture the statistics of grain
shape and size, crystallographic orientations, misorientations, and their correlations.
The DREAM.3D code resulting from this work [33] is very successful in generating
SEVMs for polycrystalline materials.
This chapter is devoted to a discussion on the development of the M-SERVE
and P-SERVE of the Ni-based superalloy René 88DT [46] at multiple scales, as
depicted in Fig. 1. Section 2 focuses on the M-SERVE and P-SERVE development
for intragranular γ − γ microstructures at the subgrain scales. It involves a
sequence of tasks, viz., serial sectioning, image processing, feature extraction, and
statistical characterization, followed by micromechanical analysis and convergence
tests for establishing the M-SERVE and P-SERVE. Subsequently the M-SERVE
and P-SERVE development for the higher-scale polycrystalline microstructure,
characterized by grains containing annealing twins, is discussed in Sect. 3.
S. Ghosh et al.
The SERVE is further classified into a microstructure-based SERVE or MSERVE and a property-based SERVE or P-SERVE in [29–31]. Similar clarifications
have been proposed in [32]. The M-SERVE is defined as one for which morphological and crystallographic characteristics of the microstructure are the sole
determinants of the representative volume. Convergence tests are needed for
determining the minimum SERVE size necessary for the statistics of characteristic
microstructural descriptors to converge. The M-SERVE is the foundation of the
DREAM.3D software [33, 34]. On the other hand, the P-SERVE is determined from
convergence of material response functions and properties.
Development of the M-SERVE and P-SERVE is preceded by the creation of
3D statistically equivalent virtual microstructures or SEVMs. The statistics of
morphological descriptors in SEVMs must be equivalent to those derived from
high-fidelity datasets obtained from experiments involving electron backscatter
diffraction (EBSD) [34, 35], scanning electron microscopy (SEM), or computed
tomography (CT). Deterministic models, e.g., in [36], are not well-suited for
microstructures with significant spatial variations. Various methods of generating
polycrystalline or polyphase microstructures have been proposed in the literature,
e.g., in [37–43]. These methods account for spatial variations in microstructural
morphology and represent the microstructure with statistical distributions functions
equivalent to those obtained from experimental observations. Conversely, other
methodologies rely on a multiphase field approach to generate microstructural
statistics for both polycrystalline and polyphase microstructures [44, 45]. These
methods focus on the modeling and subsequent statistics of the formation of
microstructural geometries and complement the top-down approach of observing
and matching the microstructural statistics. In [34, 35], statistically equivalent
polycrystalline microstructures have been developed to capture the statistics of grain
shape and size, crystallographic orientations, misorientations, and their correlations.
The DREAM.3D code resulting from this work [33] is very successful in generating
SEVMs for polycrystalline materials.
This chapter is devoted to a discussion on the development of the M-SERVE
and P-SERVE of the Ni-based superalloy René 88DT [46] at multiple scales, as
depicted in Fig. 1. Section 2 focuses on the M-SERVE and P-SERVE development
for intragranular γ − γ microstructures at the subgrain scales. It involves a
sequence of tasks, viz., serial sectioning, image processing, feature extraction, and
statistical characterization, followed by micromechanical analysis and convergence
tests for establishing the M-SERVE and P-SERVE. Subsequently the M-SERVE
and P-SERVE development for the higher-scale polycrystalline microstructure,
characterized by grains containing annealing twins, is discussed in Sect. 3.