40
S. P. Donegan and M. A. Groeber
• Data processing and cleanup functionalities, including robust image processing
courtesy ITK
• Statistical computations such as feature size and shape distributions, histograms,
and distribution fitting
• Surface and volumetric meshing
• Instantiation of synthetic microstructures from morphological and crystallographic statistics, provided either from generative statistical models or experimental measurements
Leveraging the above functionalities in concert with the flexibility of SIMPL, a user
can construct arbitrarily complex workflows for difficult ICME problems. In the
following section, such a problem is introduced as a case study to demonstrate the
utility of DREAM.3D.
6 Case Study: Ti-6242Si Pancake Forging
This section presents a case study for an ICME workflow concerned with quantitatively relating processing parameters in a titanium disk forging to measured
microstructure characteristics. Unless otherwise noted, all processing and analysis
steps presented were performed using DREAM.3D. This problem is a subset of the
workflow shown in Fig. 1. Specifically, we are interested in those procedures shown
in Fig. 12. A cylinder of Ti-6Al-2Sn-4Zr-2Mo-0.1Si (Ti-6242Si) with diameter
25.4 mm and height 38.1 mm was forged into a pancake with an average true
height strain of 1.07. After forging, the pancake was cross-sectioned radially
and characterized using both backscatter electron (BSE) imaging and EBSD.
Concurrently, the forging process was simulated using DEFORM ® . This forging
was part of a larger study in which additional cylinders were excised from the same
parent billet and isothermally compressed. Specific experimental and simulation
details may be found in Pilchak et al. [60].
This work was motivated to develop quantitative relationships between process
history and resulting microstructure. Specifically, microstructure features of interest
are microtexture regions (MTRs). MTRs are relatively large (i.e., millimeters, to
centimeters) regions of similar crystallographic orientation that form in near-alpha
titanium alloys [61]. These regions have been identified as prime factors implicated
in dwell fatigue debits of titanium forgings [62, 63]. In order to understand the
impact the process state has on MTRs, the model output from DEFORM ® was
fused with the characterization data. Then, the model data was zoned using an
approach from unsupervised machine learning, partitioning the forging geometry
into discrete regions of self-similar processing history. With the characterization
information colocated with these zones, an assessment can be made concerning the
types of microstructure expected for a given process. This zoning procedure has
the additional benefit of signifying to a designer which regions of a component
S. P. Donegan and M. A. Groeber
• Data processing and cleanup functionalities, including robust image processing
courtesy ITK
• Statistical computations such as feature size and shape distributions, histograms,
and distribution fitting
• Surface and volumetric meshing
• Instantiation of synthetic microstructures from morphological and crystallographic statistics, provided either from generative statistical models or experimental measurements
Leveraging the above functionalities in concert with the flexibility of SIMPL, a user
can construct arbitrarily complex workflows for difficult ICME problems. In the
following section, such a problem is introduced as a case study to demonstrate the
utility of DREAM.3D.
6 Case Study: Ti-6242Si Pancake Forging
This section presents a case study for an ICME workflow concerned with quantitatively relating processing parameters in a titanium disk forging to measured
microstructure characteristics. Unless otherwise noted, all processing and analysis
steps presented were performed using DREAM.3D. This problem is a subset of the
workflow shown in Fig. 1. Specifically, we are interested in those procedures shown
in Fig. 12. A cylinder of Ti-6Al-2Sn-4Zr-2Mo-0.1Si (Ti-6242Si) with diameter
25.4 mm and height 38.1 mm was forged into a pancake with an average true
height strain of 1.07. After forging, the pancake was cross-sectioned radially
and characterized using both backscatter electron (BSE) imaging and EBSD.
Concurrently, the forging process was simulated using DEFORM ® . This forging
was part of a larger study in which additional cylinders were excised from the same
parent billet and isothermally compressed. Specific experimental and simulation
details may be found in Pilchak et al. [60].
This work was motivated to develop quantitative relationships between process
history and resulting microstructure. Specifically, microstructure features of interest
are microtexture regions (MTRs). MTRs are relatively large (i.e., millimeters, to
centimeters) regions of similar crystallographic orientation that form in near-alpha
titanium alloys [61]. These regions have been identified as prime factors implicated
in dwell fatigue debits of titanium forgings [62, 63]. In order to understand the
impact the process state has on MTRs, the model output from DEFORM ® was
fused with the characterization data. Then, the model data was zoned using an
approach from unsupervised machine learning, partitioning the forging geometry
into discrete regions of self-similar processing history. With the characterization
information colocated with these zones, an assessment can be made concerning the
types of microstructure expected for a given process. This zoning procedure has
the additional benefit of signifying to a designer which regions of a component