||NAVSO P-3679: Producibility Measurement Guidelines/Methodologies
4.5 The Characterization of Processes
As discussed in the previous section, rolled-throughput yield and
robustness can be related to the three primary sources of variation circles. To
answer the questions of how to collect such data and what indices should the
data be reduced to, the issue of process, component, and material
characterization must be examined. Process characterization is defined as a way
to measure process capability, or a description of the singular qualities,
features, or traits related to individual and/or collective, progressive steps
used to manufacture a product. Process characterization is based on any
description of the process in terms of product performance in a numerical sense.
Related tactics and tools can be applied to component and material
characterization as well.
Characterization strategy essentially consists of four primary
• Product parameter definition
• Product parameter analysis
• Process parameter optimization
• Process parameter control.
This strategy seeks to link the process to
the product in a scientific manner. The basic structure of the four phase
approach is illustrated in Figure 4-7
Using producibility measurement tool #2, the first step toward
the quantitative assessment of producibility includes the capture of process,
material, and component data. The data is then statistically summarized into
capability indices and finally, the summary information is transferred to a
computer data file.
During product design simulation, the file can be periodically
accessed by the designer to develop the most cost effective combination of items
such as parts, materials, and processes. If actual capability data does not
exist, the designer can postulate the situation. Once the baseline simulations
and engineering studies are complete, the projected parameter yields are
multiplied and subsequently normalized. In turn, the normalized yields can be
used to create an overall statement of producibility.
By applying this process across many different products, an
organization can benchmark its overall design and manufacturing ability. Through
careful analysis of such benchmarking information, a company can discover how to
capitalize on its strengths and overcome areas of limitation.
Phase I of the Strategy
In overview of the characterization strategy, the first phase
involves defining the critical product response characteristic to be studied.
Although more than one response variable may be identified and subsequently
studied, this discussion is confined to only one for simplicity of
Phase II of the Strategy
The second phase of the strategy entails
the collection of baseline product data. The baseline data is gathered on a
short-term and long-term basis. The short-term data is used to estimate the
instantaneous reproducibility of the response characteristic (Y)
. In this manner, it is
possible to establish the extent to which the process is capable of meeting the
specified product quality standards - free of unsettling influences. The
long-term data is used to estimate the sustained reproducibility of the
dependent variable over a very large number of production intervals. Such data
not only reflects instantaneous or "random" noise, but systematic, nonrandom
influences as well. From these two measures, it is possible to study the extent
to which Y is robust to perturbing influences within the related network of
causation. From another vantage point, the disparity between the short-term and
long-term metrics can be viewed as a measure of how well the process is
controlled over time.
Once the product parameter capability has
been established, the second phase is continued by identifying the key process
variables (X1...XN) within the network of
causation which exert an undue influence on Y
. This is
most often accomplished through the application of statistically designed
experiments and various diagnostic tools. The intent of such activity is to
identify leverage with the network of causation; to study competing settings
among the major causal variables; and to assess the feasibility and relative
efficacy of control with respect to the key casual variables. Of course, the
capability may need to be reassessed after such activity.
Phase III of the Strategy
The third phase involves optimizing the product response
characteristic in relation to the given product standards and specifications.
This particular phase would be exercised only if the parameter capability data
and manufacturing situation so warranted. The intent is to "fine tune" the
independent variables identified during the second phase. In addition, this
phase focuses on desensitizing the response characteristic to systematic and
non-systematic sources of variation. Because of this, the application of
advanced inferential statistics, response surface design, and computer modeling
is often required. It may again be desirable to reassess capability.
Phase IV of the Strategy
The fourth phase is related to monitoring
the dependent variable (Y) and controlling the "vital few" process
) within their
respective operating tolerances. After an adjustment period, the product and
process parameter capabilities would be reassessed. Depending upon the outcomes
of such a follow-on analysis, it may be necessary to revisit one or more of the
The process characterization strategy is a scientific
and deductive method for structuring a product response problem. As may be
noted, the strategy emphasizes deductive inquiry and repeatability, two of the
central issues associated with scientific