The following Advanced features are available.
| Advanced |
Description |
|
PCC |
Process Capability Calculations |
|
GRR |
Gauge Repeatability & Reproducibility |
|
PCA |
Partial Component Analysis |
|
NSQC |
Nested SQC
Charts |
| Online |
Online Version
Setup |
Partial Component Analysis (PCA).
Statistical process control, SPC, is being implemented across the industry
without consideration to the multivariable nature of the measurements.
The statistical process control, SPC, and statistical quality control, SQC,
foundations are based upon the detection of a statistically significant change
in a given process. When a process can be described by a single measurement,
such as making an object with a desired length, implementation of these
techniques becomes an easy task.
However in the majority of the industries, a given process such as a reactor
can not be described completely by a single measurement. Implementation of SQC
and SPC technology will yield hundreds of control charts, which will identify
some causes, but not the status of the process.
The multivariable nature of processes makes it difficult to provide the unit
operator with the information with respect to the health of the process based on
SQC and SPC technology due to the number of charts generated.
The PCA technique implemented in the Advanced SQC for Excel™ reduce a
large multivariable system (i.e. 10 tags with 200 measurements) to a manageable
system (i.e. 2 tags with 200 measurement) while preserving most of the
information in the original system. Suppose that you have system with 5
measurements as it is shown in the figure below.
Application of normal SQC charts will yield to out of control charts.
However, with 3 simple clicks.
- Select the data from the 40th to 140th measurement
and start SQC for Excel™
- Click OK for the data type
- Click OK for the data headers
- Select PCA as the output.
- Check the charts ( PCA chart is condensed chart of all variables plotted
as normalized and is plotted as +- 3 which means +- 3 Sigma).
The SQC chart is in control.
Then select all data (40th to 200th ) and re-do the
PCA. Observe that the ***
Look at the SQC PCA chart, it takes off after 141 sample, and checking the
data shows the first line is draping rapidly.
If you add this functionality to you alarm systems, you can detect run away
measurements very easily.
