Operationalize Statistics Summaries

The Operationalize Statistics Summary tool can be opened from the Statistics menu of the Tools pane, or by using the search term Operationalize in the Tools search field. Clicking the tool opens a user interface in the tools pane.

Alternatively, the Operationalize button in the top toolbar of the Summary Report view can be used to operationalize the Statistics Summary that has been selected in the Baseline: field, as shown in Figure 2.

The fields in the User Interface populate automatically with information from the selected baseline Statistics Summary. In the example of Figure 2, the Production Reference Statistics Summary is configured for an I-MR chart with run rules WE1 to WE4.

The fields in the User Interface can be populated or adjusted using the drop-down selectors or by clicking Add defaults.

Title: The title field defaults to ‘Operationalize’ plus the date and time. Users can set a custom title for the operationalization instance by editing this field.

Select a Statistics Summary: The selected Statistics Summary will be applied to a signal for continuous monitoring. The selected summary must have been configured with a control chart pair (I-MR, XBar-R, or XBar-S). The configuration is done with the Statistics Summary tool.

Input signal: Normally, the input signal is the same signal that was used to generate the Statistics Summary. It can also be a similar signal, for instance in a second production line.

Limit output to be within a condition (optional): A user can select when to see the monitoring output. For instance, if the Statistics Summary represents operation during manufacture of a particular product grade, the monitoring output should be limited to displaying only when that product grade is being manufactured.

Select control charts to operationalize: The available charts and run rules are those that have been configured in the Statistics Summary. In Figure 3, the Statistics Summary has been configured for I-MR charts with run rules WE1, WE2, WE3, and WE4. The I- chart and MR- chart will be displayed in the Seeq Workbench, and violations of the selected run rules will be flagged for automated detection of unwanted process variation.

Figure 3a shows the default selection, while Figure 3b shows other options. Of these, the Union of run rules condition has a capsule when any of the run rules is flagged (i.e. a Boolean OR function on WE1, WE2, WE3 and WE4 in the example of Figure 3a).

Select statistics to operationalize: Selected statistics will be aggregated over each capsule of a periodic condition and displayed in the Seeq Workbench, if they have been configured in the Statistics Summary. There is more information in a later section of this page.

Open Summary Report View: Clicking this link opens the Summary Report tab in the display pane of the worksheet.

Execute: Clicking the Execute button creates the Operationalization outputs in the Trend view of the worksheet. Any previous outputs that have been subsequently been removed from the tool will also be removed from the Trend and Details panes.

Color coding: The icons for the signal and condition in Figure 3a are grey, while those in Figure 3b are colored. The icons acquire a color when the signal or condition is present in the Details pane of Workbench, typically after the Execute button has been pressed.

The template instance can be updated and reapplied.

Clicking the Reapply template pencil icon for any of the operationalized signals or scalars in the Details pane will re-open the user interface to allow further editing and adjustment.

Limit summary to a condition

The optional Limit output to be within a condition restricts SPC calculations to measurements collected under specified operating conditions. Measurements outside the capsules of the limiting condition are not displayed. Use cases include:

• Limiting calculations to production runs

• Excluding maintenance periods

• Focusing on particular operating modes or product grades

Figure 7 illustrates the effect of selecting Compressor on as a limiting condition for the Production signal. Compared to Figure 5, the I-chart and MR-chart samples and the run-rule markers are only displayed under the capsules of the Compressor On condition.

Sample alignment Condition

The Sample alignment condition is a mandatory field for the subgroup charts Xbar–R and Xbar–S. A subgroup is a run of several consecutive samples. The average (Xbar) and range (R) or Standards Deviation (S) are calculated for each subgroup. If signal has 1000 samples and a subgroup size of four, there will be around 250 Xbar values.

Sample alignment defines how the subgroup boundaries are determined. The capsules of the Sample alignment condition are operational periods such as shifts, days, batches, or production grades. There must be sufficient samples in each capsule to form subgroups.

Subgroups are anchored to the start of operational periods. In that way, the same subgroups are preserved as the worksheet display pans and zooms.

Figure 8 shows the Weekly condition at the top of the worksheet. Included here for the purposes of explanation are green point capsules below the brown bar in Lane 1 to mark the start of each Weekly capsule. New subgroups will start at these points.

The Alignment condition can be manually configured in the Advanced section of a Statistics Summary. Sample alignment defaults to the Limiting condition if one is selected, otherwise it defaults to “Weekly”. The operationalization in Figure 8 uses the default Weekly condition with no limiting of the output.

Alignment capsule boundaries

When a new alignment capsule begins, a new subgroup is started and the run rules are reset. There may be a visible gap in the output signal between one alignment capsule and the next, as seen in Figure 9. This happens when the time difference between the final group of the previous capsule and the first group of the next capsule is more than the maximum interpolation of the operationalized signal.

Behavior of subgroups and run rules

Subgroups and run rule sequences are reset at the boundaries of Alignment condition (e.g. Weekly in Figure 10).

Subgroups and run rule sequences do not reset at the boundaries of Limiting conditions (e.g. Compressor on in Figure 10).

When valid measurements resume, subgroup formation and run-rule evaluation continue using the next valid samples unless a sample alignment boundary has been reached.

The right hand side of the shaded area in Figure 10 shows that Western Electric Rule 4 is triggered on the third sample in a capsule of the limiting Compressor on condition. WE4 detects runs where eight consecutive samples are above or below the center line. Here, the WE4 rule has been violated because five sample in the previous capsule were above the center line as well as the three in the right hand capsule of the shaded area.

Statistics

The Statistics section offers the following statistics that can be operationalized. The selected statistics are calculated periodically and displayed in the Trend view of the worksheet.

Capability and performance:

• C_pk - Process capability index

• P_pk - Process performance index

• Normal Distribution p-value

Comparison tests, where the basis for comparison is the Baseline selected for operationalization:

• Equal Variance p-value

• Equal Mean p-value

These statistics are described and defined in the Statistics Definitions and Reference page of the Knowledge Base.

Statistics configuration

If Statistics are selected, further fields are displayed.

Periodic condition:

Statistics will be calculated within the capsules of the selected periodic condition, for instance, daily, weekly, or monthly. Periodic conditions can be created with the Periodic Condition tool in the Identify tools section in Workbench.

Other conditions with repeating capsules can be selected, for instance to reflect shifts, batches, or product grades.

It will be necessary to specify a maximum capsule duration, if the condition does not already have it.

Statistics timestamp location

The Statistics timestamp location specifies where the timestamp is placed for each calculated statistic value.

Start - Timestamp at the beginning of each capsule
End - Timestamp at the end of each capsule (default)
Duration - The result is displayed across the full time span of each capsule

Statistics interpolation method

The Statistics interpolation method specifies how the calculated statistic values are interpolated between timestamps.

Discrete - Bar chart display with discrete bars (default)
Linear - Line chart with linear interpolation between points
Step - Step interpolation, values hold until the next timestamp

Figure 13 shows the Trend view where the following can be observed:

Lane 1: The periodic Daily condition with the start of each day indicated by a point capsule

Lane 2: The Production signal that is being monitored

Lanes 4 to 7: The operationalized statistics for each day plotted as bars with a timestamp at the end of each day.