Statistical Process Control

Statistical Process Control (SPC) is a diagnostic tool that allows you to determine “special” versus “common” causes of variation to monitor the output quality of a process. Common causes of variation affect every process, and are why nothing is ever exactly the same. Special causes of variation occur when something happens that is not usually part of the process. SPC allows you to identify when these special causes occur so that you can eliminate them and bring predictability, or “control” to a process without overreacting to normal variability. SPC by itself does not change a process - you must have a reaction plan in place that tells you what to do when a special cause occurs.

By using a control chart you can graphically monitor a process variable and continuously perform a statistical test to determine if the mean has shifted, an indication that the process has changed. Control charts also identify an increase or decrease in process variability. The upper and lower limits of expected variability are calculated from historical data.

Anatomy of a Control Chart

This shows the basic composition of a control chart. The red dashed lines indicate the range of expected variability 99.73% of the time. If a single point goes outside of these limits, you are probably safe in reacting as if your process has changed in an unusual way. There are other ways to determine if a process change has occurred, called “pattern rules.” Here are the pattern rules ROI uses:

• Runs — eight or more points in a row on one side of the centerline
• Trends — six or more points in a row decreasing or increasing
• Cyclical Patterns — a repeating pattern
• Alternating Values — 14 or more points in a row that alternate direction, e.g., up/down/up etc.
• Excessive Variability — eight or more consecutive points outside the center third
• Lack of Variability — 14 or more points in the center third (near the centerline).

A common misconception about control charts is that they control a process. In reality, they are a powerful diagnostic tool to see if the process has shifted, and although they can be used to change the process when a special cause occurs, they are not the most efficient way to control a process in real time. Engineering Process Control (EPC) when used to its full capability will give you lower variability around a target. To see a slideshow presentation on other common problems people have with SPC, click on Why Doesn't SPC Work?

If you want to learn more about SPC, you can purchase Quality With Confidence, written by Alliance member Mike Petrovich and LWI founder Dr. Jeffrey Luftig. The ROI Alliance has the premier SPC training in the world available to be taught at your location(s).