Control Chart and Trigger Rule¶
Statistical process monitor — instantiates Conformance Control and Corrective Feedback
Plots a characteristic over time against statistical control limits so a drifting process trips a predeclared trigger before its output crosses the spec.
Most conformance checks ask "is this unit good?" A Control Chart and Trigger Rule asks a different question: "is the process still behaving the way it did when it was in control?" It plots a characteristic as a time series and compares each point (or subgroup) not to the specification but to control limits computed from the process's own recent variation. Its defining idea is the separation of common-cause noise — the ordinary jitter you should leave alone — from special-cause signals — a shift, trend, or spike that means something changed and warrants action. A predeclared trigger rule fires on the special-cause pattern and calls for a process adjustment, ideally while the output is still in spec, so the loop catches a drift before it becomes a defect rather than after.
Example¶
A beverage line fills bottles to a labelled net content. Every fifteen minutes an operator weighs a small rational subgroup of ≈5 bottles and plots the subgroup average and range. The specification allows a fairly wide band, but the control chart's limits — set from the line's own stable behaviour — are tighter than the spec. One afternoon the subgroup averages drift upward for seven consecutive points. Nothing is out of spec yet; every bottle would still pass acceptance. But the run rule for "seven rising points" fires,[n1] and the operator checks the filler and finds a valve creeping open. Adjusting it pulls the average back before a single underfilled or overfilled bottle is ever made.
The value is entirely in the earliness: the chart converted a slow, still-conforming drift into an actionable signal, and the trigger rule told the operator to touch the process now — and, just as importantly, to leave it alone the rest of the day when the points were merely bouncing inside the limits.
How it works¶
- Sample in rational subgroups. Draw small subgroups at intervals chosen so that variation within a subgroup is pure common-cause noise, and variation between subgroups is where a real shift shows up.
- Set limits from the process, not the spec. Control limits come from the process's own stable variation (typically ≈3σ), so the chart measures behavioural change, not conformance to a customer tolerance.
- Fire on predeclared patterns. A point beyond a limit, a run, a trend, or a hug of the centreline each has a rule; the trigger is declared in advance so a signal is not argued about after the fact.
- Feed the process, not the part. A fired trigger routes to a process adjustment — the loop's job is to correct the producer, so the next output is centred rather than to disposition the current unit.
Tuning parameters¶
- Control-limit width — usually ≈3σ. Tighter limits catch shifts sooner but raise false alarms and tempt over-adjustment; wider limits are calmer but slower.
- Subgroup size and sampling interval — larger, more frequent subgroups detect small shifts faster at higher measurement cost; the pair sets how small a drift the chart can see.
- Trigger-rule set — how many run/trend rules are armed. More rules increase sensitivity to subtle patterns but multiply false signals and operator fatigue.
- Chart type — averages-and-range, individuals, or attribute (p/c) charts; matched to whether the characteristic is measured, counted, or one-at-a-time.
- Reaction protocol — what a trigger obliges (investigate, adjust, stop). Strong reactions contain drift fast but disrupt flow when the signal is false.
When it helps, and when it misleads¶
Its strength is leading rather than lagging: it can catch a process going wrong while the output is still conforming, and it disciplines the opposite error too — telling operators which wiggles are noise to be left alone. It is the natural home for the feedback half of the quality loop on a continuous process.
Its two failure modes are mirror images. Under-reacting, a chart nobody watches or acts on is decoration. Over-reacting, an operator who adjusts to every point amplifies variation — chasing common-cause noise as if it were signal is tampering, and it reliably makes a stable process worse.[n2] Control limits also silently go stale when the process legitimately re-centres, and a chart says nothing about whether the gauge itself is trustworthy — a wandering measurement system will paint special-cause signals that are pure metrology. The discipline is to act only on the declared rules, recompute limits when the process genuinely changes, and confirm the measurement system before believing the chart.
How it implements the components¶
measurement_and_sampling_plan— defines the rational-subgroup sampling: what to measure, subgroup size, and interval — a process-monitoring sampling plan (distinct from lot-disposition sampling).conformance_comparison_rule— supplies the statistical trigger rules that compare points to control limits and run patterns (distinct from a deterministic spec pass/fail).process_feedback_and_control_update— its whole point: a fired trigger routes a signal back to adjust the producing process.
It does not size or verify the measurement system (Measurement-System Capability Analysis), find and fix the root cause of a recurring shift (Corrective and Preventive Action Cycle), or gate release of the current output (Release Hold and Signoff).
Related¶
- Instantiates: Conformance Control and Corrective Feedback — it is the continuous-monitoring, feed-the-process half of the loop.
- Consumes: Measurement-System Capability Analysis — the chart's signals are only real if the gauge is capable.
- Sibling mechanisms: Corrective and Preventive Action Cycle · Automated Conformance Check · Risk-Stratified Acceptance Sampling Plan · Measurement-System Capability Analysis · Upstream Quality Feedback Packet
Editorial Notes¶
Form Classification¶
Form family: Monitoring, Sensing & Alerting
Rationale: Plots a characteristic over time against statistical control limits so a drifting process trips a predeclared trigger before its output crosses the spec, making its operative form an ongoing sensing arrangement that repeatedly observes state and surfaces changes or alerts.
Independent corroboration: The frozen evidence defines Control Chart and Trigger Rule as 'Plots a characteristic over time against statistical control limits so a drifting process trips a predeclared trigger before its output crosses the spec', so its operative form is Monitoring, Sensing & Alerting.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Statistical process control cohered Shewhart charts plus predeclared run and trend rules for detecting special causes before specification failure.
Related originating lineages:
- Engineering & Design — Quality engineering applies the triggered response to maintain conformance in production.
Review resolution: Both reviewers agree on statistical control limits and engineering intervention practice. The chart-plus-trigger mechanism joins those lineages and is used across multiple process domains; operations research is an application rather than a separate origin.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Run/trend tests such as the Western Electric rules ("N points beyond a limit," a run on one side of the centreline, a monotone trend) are standard predeclared patterns for detecting special-cause signals on a Shewhart chart before any single point exceeds a control limit. ↩
[n2] Adjusting a stable process in response to common-cause variation increases its variability — Deming's funnel experiment is the canonical demonstration. It is why the trigger rule matters: it authorizes action only on genuine special-cause signals. ↩