Manufacturing & OperationsGlossary

What Is SPC (Statistical Process Control)?

Also known as: statistical process control, control charting

Definition

SPC (Statistical Process Control) is the use of control charts and statistical rules to monitor a process in real time, distinguishing normal common-cause variation from special-cause variation that signals a genuine change requiring action.

SPC (Statistical Process Control) Explained

The core insight is that every process varies, and reacting to ordinary variation makes output worse. SPC separates common-cause variation, which is inherent to the process, from special-cause variation, which comes from an identifiable event such as a tool change, a new material lot, or an operator substitution. Control limits are calculated from the process's own measured variation, typically at plus and minus three standard deviations of the plotted statistic, and are never set equal to engineering specification limits.

Chart selection follows the data. Variable data uses X-bar and R charts for subgroups up to about eight, X-bar and S for larger subgroups, and individuals with moving range where sampling is one at a time, as in low-volume aerospace machining. Attribute data uses p and np charts for proportion or count defective and c and u charts for defects per unit. Choosing the wrong chart produces limits that are mathematically valid but operationally meaningless.

Signals are defined by rules, most commonly the Western Electric or Nelson sets: a point beyond a control limit, nine consecutive points on one side of centerline, six points steadily increasing or decreasing, fourteen alternating, and several zone-based patterns. Enabling every rule at once on every chart raises the false alarm rate substantially and trains operators to ignore alarms, so mature programs enable a small, deliberate subset.

The most damaging implementation error is confusing control limits with specification limits. Control limits describe what the process does; specification limits describe what the customer requires. A process can be perfectly in control and entirely out of specification, or wildly out of control while every part still passes. Placing spec limits on a control chart, a default in some software, causes operators to adjust a stable process every time a point drifts, which is textbook tampering and demonstrably increases variation.

Why It Matters

  • Detects process shifts while parts are still conforming, converting quality control from post-hoc inspection into prevention.
  • Stops tampering, the very common and costly habit of adjusting a stable process in response to ordinary variation.
  • Provides the in-control evidence required before Cp and Cpk capability indices have any statistical meaning.
  • Supplies the objective process evidence customer quality systems in aerospace, defense, and automotive increasingly require at audit.

In Practice

A shop charts a shaft diameter with control limits at plus and minus 0.0006 inch against a specification of plus or minus 0.002 inch. An operator sees a point at plus 0.0005, well inside spec but near the upper control limit, and adjusts the offset down. The next natural low point triggers an adjustment up. Within a shift the chart shows a sawtooth and measured standard deviation has nearly doubled. The chart was working correctly; the response rule was not, because the operator was taught to react to position rather than to signals.

Frequently Asked Questions

Are control limits the same as specification limits?

No, and treating them as equal is the most common SPC error. Control limits are calculated from the process's own variation and describe what it naturally does. Specification limits come from engineering and describe what the customer will accept. A process can be in control and out of specification, or out of control while every part still conforms.

Can SPC work in low-volume production?

Yes, with adjustments. Individuals and moving range charts handle one-piece-at-a-time measurement, and short-run SPC techniques normalize deviation from target so multiple part numbers can share one chart. The practical constraint is that control limits need roughly 20 to 25 data points to stabilize, so limits from very short runs should be treated as provisional and recalculated as data accumulates.

Working with SPC (Statistical Process Control) in a live environment? Our engineers do this every day - and our AI agents automate most of it.