Understanding Process Variation
William Edwards Deming (1900-1993) was an important contributor to statistical process control and its use in manufacturing. According to the American Society for Quality (ASQ), his 14 key points on quality management are a core part of modern quality management programs.
Understanding process variation is an integral aspect of using Statistical Process Control (SPC) to improve your manufacturing processes. Dr. Deming’s first principle states, “The central problem in lack of quality is the failure of management to understand variation.” Only after management understands variation can a manufacturer succeed in implementing Dr. Deming’s second principle: “It is management’s responsibility to know whether the problems are in the system or in the behavior of the people.”
Types of Process Variation
There are two types of process variation:
- Common cause variation is inherent to the system. This variation can be changed only by improving the equipment or changing the work procedures; the operator has little influence over it.
- Assignable cause variation comes from sources outside of the system. This variation can occur because of operator error, use of improper tooling, equipment malfunction, raw material problems, or any other abnormal disruptive inputs.
The goal of SPC is to understand the difference between these two types of process variation—and to react only to assignable cause variation. Processes that show primarily common cause variation are, by definition, in control
and running as well as possible.
Control versus capability
Note that keeping a process in control doesn’t mean that the product is acceptable; the system must also be capable
of making acceptable products. Control
are different concepts.
SPC uses statistical tools—such as control charts—to identify process variations. Special cause variations—those outside the standard or expected variation—are identified and their causes need to be eliminated or at least understood.
Example of special cause variation
Suppose you drive to work each day. Your path has inherent or common variations, such as traffic lights. But suppose there is a railroad crossing that causes you to be 30 minutes late for work. That day’s commute would be special variation,
and the railroad crossing would be the assignable cause.
As a result of understanding and reducing or eliminating assignable cause variations (perhaps there is a route with no railroad crossings), processes can be kept in control and continually improved. Adjusting an in-control process when there is no identified need is called tampering
and only increases the variation of the system.