230
N. Wade et al.
were chosen to demonstrate the utility of modeling processes for error estimation
and to allow for the incorporation of many different parameters to be analyzed
within a reasonable time frame. Nonetheless, the results presented in the following
sections can offer some insights into important trends and will serve to demonstrate
the potential of the framework.
2.3 Additional Notes on Methodology
One goal in developing the framework was to create a simple tool which can
help inform the choice of data collection parameters, even in an in situ manner.
It provides an outline for analyzing how changes to the data collection and data
processing of microstructural data sets, and to do so it was necessary to develop
simple models relying on several simplifying assumptions. However, additional
capabilities or refined models can easily be implemented to meet the desires of
the user. If users feel model assumptions may not generally hold for their specific
application, models can easily be calibrated and/or refined to more accurately model
the process in any given system. For example, the data collected as part of Sect. 2.2.3
is instrument specific, and users might want to collect data to develop a model for
their specific instrument setup. Similarly, an improved physics-based model of the
interaction volume (Sect. 2.2.2) that accounts for chemical compositions could also
be implemented. Ultimately, the model framework was designed to allow for the
variation of individual input variables. As such, any doubt surrounding the effects of
a certain quantitative input (e.g., the size interaction volume) can be studied through
direct variation of said parameter. By varying a parameter of interest over a range of
reasonable values, and examining the output, a sensitivity to said parameter can be
developed. Model assumptions relating to parameters with high sensitivity can then
be refined as needed.
Additionally, the framework is intended to be able to incorporate computational
models in its evaluation of the error between true and measured microstructures. For
this a third step is added where the computational model is applied. An example of
such an evaluation is included as part of Sect. 4.
3 Individual Parameter Variation Examples
One benefit of the framework is that it is possible to quantitatively analyze the
effect of various characterization parameters. This allows for the sensitivity to be
quantified and compared for any parameter, allowing for a more detailed study
of the evolution and propagation of error in the sample. In this section, several
simulations of EBSD data collection are conducted over a range of typical values
for the parameters of interest. Each simulation was analyzed and compared directly
to the phantom, using only the interior non-biased features. The exact size of the
N. Wade et al.
were chosen to demonstrate the utility of modeling processes for error estimation
and to allow for the incorporation of many different parameters to be analyzed
within a reasonable time frame. Nonetheless, the results presented in the following
sections can offer some insights into important trends and will serve to demonstrate
the potential of the framework.
2.3 Additional Notes on Methodology
One goal in developing the framework was to create a simple tool which can
help inform the choice of data collection parameters, even in an in situ manner.
It provides an outline for analyzing how changes to the data collection and data
processing of microstructural data sets, and to do so it was necessary to develop
simple models relying on several simplifying assumptions. However, additional
capabilities or refined models can easily be implemented to meet the desires of
the user. If users feel model assumptions may not generally hold for their specific
application, models can easily be calibrated and/or refined to more accurately model
the process in any given system. For example, the data collected as part of Sect. 2.2.3
is instrument specific, and users might want to collect data to develop a model for
their specific instrument setup. Similarly, an improved physics-based model of the
interaction volume (Sect. 2.2.2) that accounts for chemical compositions could also
be implemented. Ultimately, the model framework was designed to allow for the
variation of individual input variables. As such, any doubt surrounding the effects of
a certain quantitative input (e.g., the size interaction volume) can be studied through
direct variation of said parameter. By varying a parameter of interest over a range of
reasonable values, and examining the output, a sensitivity to said parameter can be
developed. Model assumptions relating to parameters with high sensitivity can then
be refined as needed.
Additionally, the framework is intended to be able to incorporate computational
models in its evaluation of the error between true and measured microstructures. For
this a third step is added where the computational model is applied. An example of
such an evaluation is included as part of Sect. 4.
3 Individual Parameter Variation Examples
One benefit of the framework is that it is possible to quantitatively analyze the
effect of various characterization parameters. This allows for the sensitivity to be
quantified and compared for any parameter, allowing for a more detailed study
of the evolution and propagation of error in the sample. In this section, several
simulations of EBSD data collection are conducted over a range of typical values
for the parameters of interest. Each simulation was analyzed and compared directly
to the phantom, using only the interior non-biased features. The exact size of the