Skip to content

Root Cause Analysis

When a process shows signs of unexpected variation or a shift in behavior, it is essential to determine what changed and why. GS provides tools such as Control Charts and Pareto Charts to help perform root cause analysis.

The most effective way to detect these behavioral shifts is to use Real-Time Failures and Alerts during data entry. Real-Time Failures in GS are a set of common signals of process changes, such as a data record being out of control, or a long trend of data records increasing in value. Detecting these indicators during data entry results in quicker feedback and can prevent the process from moving in an undesirable direction as early as possible. Or, if a desirable change has occurred, you might use these signals to determine the change and incorporate it into the process.

Nevertheless, sometimes these real-time alerts are not sufficient to prevent variation and defects. In this guide, you'll learn how to use Dashboard visualizations to investigate process changes and identify likely causes.

Prerequisites

This guide assumes:

  • You have sufficient data to perform analysis (at least 25-30 data records).
  • These data records have associated Traceability values.
  • You have an understanding of the difference between common causes and special causes, and an understanding of how these causes should be addressed differently.

If you do not have these prerequisites, first complete the following guides:

SPC Root Cause Analysis

This section of the guide assumes you are concerned about a particular Characteristic which you would like to investigate further. This concern may have been raised in a variety of ways, such as from customer complaints or a design of experiments.

SPC variation analysis is usually performed with Control Charts. When viewing Control Charts, watch for signs of variation, like:

  • Points beyond control limits.
  • A sustained shift in the process centerline.
  • Cyclical or trending behavior.

Using Control Charts

Begin by drawing a Control Chart of the concerning Characteristic. In this example, we are investigating a bolt's Diameter.

An image showing a control chart of the diameter of a bolt

Use the Control Chart's Group By property to change how data is grouped within the chart. Grouping points lets you analyze data by distinct Traceability values. GS calculates control limits and means for each group of data in the chart. For example, the following chart is grouped by Shift. There is slight variation between the mean and spread (variation) of the different Shifts, but they are largely similar.

An image showing a control chart of the diameter of a bolt grouped by Shift

After trying a few more Traceability fields in the Group By property, eventually Cavity is used. There is quite obviously a problem with Cavity 4; both the mean and the spread of Cavity 4 are worse than the other groups.

An image showing a control chart of the diameter of a bolt grouped by cavity

To further investigate why this cavity is an issue, you might:

  • Speak with production or maintenance teams to verify the issue.
  • Hover on each point in the chart to investigate if any related notes have been entered.
  • Expand the Retrieval's date range to discover if this is a recurring pattern.
  • Filter the Retrieval by Cavity 4 and re-perform the above analysis (changing the chart's Group By).

DMS Root Cause Analysis

This section of the guide assumes you are concerned about the presence of Defects in your processes. Defect causes can be analyzed using a Pareto Chart and drill down.

Using Pareto Charts

Begin by drawing a Pareto Chart of the Parts or Processes you are concerned about. GS automatically groups the chart by Defect, showing the Defect with the most occurrences at the top of the Chart. The chart's Group By property controls how the bars are grouped.

An image showing a Pareto chart of the Final Inspection process grouped by Defect

To investigate causes of the most common Defect, Dent, click on its bar. You will see a small overlay appear with other grouping options in a dropdown. When you select one of these options, the Pareto Chart will display the Dent Defects with the new grouping option. For example, the following chart has been filtered by Dent and grouped by Operator.

An image showing a Pareto chart of the Final Inspection process grouped by Operator

We can see that C. Shultz has produced almost all of the Defects. If we want to further investigate which machine this operator was using when the Defects were produced, click on that bar and then select Machine. It is obvious that C. Shultz needs help when working on HX 2000.

An image showing a Pareto chart of the Final Inspection process grouped by Operator then by Machine

Next Steps - Insights

While the steps presented in this guide as useful for manual analysis, GS can use automated Insights to perform most of it for you.

See Also