The Path from Data to Actionable Information as a Driver for the Industrial Ecosystem
value of in-process verification is much
greater than getting a more accurate OEE
metric: It is also the primary feedback
mechanism to create the predictive and
prescriptive AI models. Quality Information
Framework (QIF) and MTConnect combined
with AP-238 (STEP-NC) have demonstrated
how this can be done using standards.
be performed on a continuous basis: It
cannot be subject to a spot inspection since
the results will not be realistic.
Using data from equipment, we have found
the optimal process performance by historic
data analysis. There is additional complexity
when considering high part mix processes.
The solution to high mix is to analyze the
process down to each micro-planned step
since each part is the aggregate of many
smaller processes that are combined create
an outcome.
P ERFORMANCE E FFICIENCY
Performance [4] is the most abused of the
three factors. The numerator of the
performance equation, the planned process
time, is an engineering estimate that is often
exaggerated to increase the OEE metric. For
example, engineering may say the process
takes 30 minutes, but the actual time to
execute the process is 15 minutes. This will
lead to a 2.0 factor for this metric. If the
availability of the equipment is 60% and the
performance is 200%, then the resulting OEE
using these two metrics will be 120% since
the quality factor is often 100%.
By using AI, one can find equivalencies and
analyze similar features using historical
precedence. When combined, these can
generate
an
optimal
performance
benchmark, for even a single, one-off part.
The collection and classification of
production time occur during enrichment
stage, and the comparison to baseline will
occur in the ecosystem integration where
the current time accounting will be
compared against the target using the
benchmark created by combining the
process plan and historical information in
prescriptive analysis.
The correct way to compute performance is
to baseline the process for a period and to
determine the fastest time to execute under
the best conditions. Measurement must also
T OWARDS P RESCRIPTIVE O PERATIONAL E FFECTIVENESS
The prior discussions began leading to the improvement of OEE to be a useful KPI and allow for
correct attribution of quality, availability, and performance to each process step. By utilizing this
metric and the additional data collected, optimizations can be made to the process while still
meeting quality and schedule targets.
The methodology as described above will also allow for a more prescriptive approach to
performance and quality by combining historical observations with predictions in a continual
refinement process that can work across high part mix and variability. OEE done in this way can
be used to prescribe process flow and manufacturing tasks to get optimal effectiveness from
equipment.
IIC Journal of Innovation
- 35 -