Examples of Common Nonrandom Patterns Occurring in Shewhart-type Charts

By Dave Harrold, CONTROL ENGINEERING March 1, 1999

E valuating chart patterns is part of Six Sigma training. The following are some common nonrandom patterns that may occur on Shewhart-type control charts. These nonrandom patterns may be used as a basis for control chart interpretation and/or establishing test runs to determine out-of-control conditions.

Source: Control Engineering with data from Motorola

Mixture patterns: This pattern is indicated when the plotted points tend to fall near the control limits, with relatively few points near the center line.

A mixture pattern is caused by two or more overlapping distruibutions generating the process output. The severity of this pattern depends on the extent to which the distributions overlap.

Sometimes mixtures result from ‘over-control’ where operators respond to random variations, rather than assignable causes.

Source: Control Engineering with data from Motorola

Cyclic patterns: In X charts, such a pattern may indicate a systematic change such as temperature, operator fatigue, operator or machine roatation, fluctuations in voltage, pressure, or other factors.

In R charts this pattern may indicate maintenance schedules or tool wear.

Source: Control Engineering with data from Motorola

Shifting patterns: Shifts may result from introduction of new operators, methods, raw materials, machines, etc. They also could be from a change in inspection methods, standards, or from a process improvement.

Source: Control Engineering with data from Motorola

Trending patterns: Trends are usually due to a gradual plugging, wearing out, or deterioration of a tool or some other critical process component.

Trends can also result from seasonal influences.

Source: Control Engineering with data from Motorola

Stratification patterns: When plotted points have a tendencey to cluster around the center line it may indicate improperly calculated control limts. It can also mean the process is getting better. In either case, the control limits should be re-evaluated.