Variation

Variation is the natural spread in measured values a manufacturing process produces across multiple parts, even when the process is running normally and nothing is malfunctioning. Understanding a process's normal variation, as opposed to a genuine problem, is central to process capability and to correctly interpreting an outlier.

Why variation, not just the average, determines reliable tolerance capability

A process's variation, not just its average output, determines whether it can reliably hold a given tolerance, which is why process capability assessments focus on variation and centering together, not just whether the average part measures on target.

Where variation is assessed

Not on the drawing; variation is a property of how a process actually performs, assessed through measurement data rather than specified by the part's requirements.

Common mistakes

Assuming a process with a good average measurement has no variation problem; a process can average right on target while still having enough spread to regularly produce out-of-tolerance parts.

Is variation always a problem?

No, some level of variation is normal and expected in any process; the question is whether it's small enough relative to the tolerance being held.

How is variation different from an outlier?

Variation describes the normal, expected spread of a process; an outlier is a specific value that falls notably outside that normal spread.

Does reducing variation always require a different process?

Not necessarily; tightening control of existing process parameters can sometimes reduce variation without changing the underlying process itself.