Process Control: CTQs, SPC, and Capability

SEF Reference Guide G10 | Manufacturing OpEx Maturity Assessment

Supports criteria: QS-5 Process Control

Glossary terms covered: CTQ, Control Chart, SPC, Control Plan, Process Capability

How to use this guide. This is SEF's standard method and the numbers we start from. Every operation is different, so treat the targets, timings, weightings, and templates here as a starting point and adjust them to your products, customers, and risk. Write your version down before you measure against it. Do not apply a tool just to apply a tool. Every method here carries a cost in time, attention, and discipline, and the benefit has to justify that cost in your operation. If it does not, leave it out and say why. A lower score on a criterion you have deliberately chosen not to pursue is an honest score, not a failure. One thing should not change: the maturity level definitions, because they are the scoring basis for the assessment.

What it is

Process control means measuring the few characteristics that matter most to the customer, measuring them inside the process where they are created, and reacting to statistical signals before a single defect is made. It replaces “make it, then check it” with “watch the process, and act when it changes.”

The chain has four links. CTQs (Critical to Quality characteristics) translate what the customer needs into measurable specifications. A control plan says how each CTQ is measured, how often, by whom, and what to do when something goes wrong. Statistical Process Control (SPC) uses control charts to tell normal variation from real change. Process capability (Cp, Cpk, Pp, Ppk) shows whether a stable process can meet the specification with margin to spare.

Walter Shewhart proposed the control chart in 1924, as an engineer at Western Electric, and developed it further at Bell Telephone Laboratories. Western Electric’s Statistical Quality Control Handbook (1956) set out the run rules still used on the floor today.

Why it matters

Quality: SPC catches drift while parts are still in tolerance, so defects are prevented rather than sorted out.
Cost: Less scrap, rework, and inspection. Stable, capable processes need less checking.
Delivery: Fewer surprises at final test mean fewer late orders.
Customer approval: Many customers require capability studies and control plans before approving a part, and capability numbers in ongoing audits.
Better decisions: Control charts stop people from adjusting a stable process in response to normal noise, which only adds variation.

Key terms

Term	Meaning
CTQ	Critical to Quality. A measurable characteristic of the product or process that must be controlled to meet a customer requirement.
Control Chart	A time-ordered chart of a process measure with a center line and upper and lower control limits calculated from the process’s own natural variation (not the specification limits). Points outside the limits, or non-random patterns, signal that something has changed.
SPC	Statistical Process Control. Using control charts and statistics to monitor a process in real time and react to signals before defects are produced.
Control Plan	A document listing each characteristic to be controlled, how it is measured, how often, and the reaction plan to follow when the process goes out of control.
Process Capability	How well a stable process meets its specification limits. Cp and Pp compare the spread of the process to the width of the specification. Cpk and Ppk also account for how well the process is centered. Cp/Cpk use short-term (within-subgroup) variation; Pp/Ppk use long-term (overall) variation.
Common cause variation	The natural, random variation built into a stable process. Reducing it needs a change to the process.
Special cause variation	Variation from a specific, identifiable source (a worn tool, a new material lot). It shows up as a signal on the chart.
Control limits	Usually set at 3 standard deviations either side of the center line, calculated from process data.
Specification limits (LSL, USL)	The limits the customer or design allows. Never draw these on a control chart in place of control limits.
Rational subgroup	A small group of parts (often 3 to 5) made close together in time, so variation within the subgroup is only common cause.
Reaction plan	The written steps an operator follows when a chart signals: stop, contain, adjust, escalate.
In control (stable)	A process showing only common cause variation. Capability numbers mean something only for a stable process.

How to do it

Define CTQs. Start from customer requirements, drawings, and complaints. Use a CTQ tree: customer need, then the drivers of that need, then measurable characteristics with specification limits. Keep the list short. Most parts have a handful of true CTQs.
Find where each CTQ is created. Map the process and mark the step that sets each characteristic. That is where to measure (see G07 for measurement system checks).
Check the gauge first. Run MSA before charting. A poor measurement system produces false signals.
Choose the chart.

Data type	Situation	Chart
Variables (measured)	Subgroups of 2 to 9	X-bar and R
Variables	Subgroups of 10 or more	X-bar and S
Variables	One reading at a time (slow processes, batch chemistry)	Individuals and moving range (I-MR)
Attributes	Fraction defective, varying sample size	p
Attributes	Number defective, constant sample size	np
Attributes	Defects per unit, constant area of opportunity	c
Attributes	Defects per unit, varying area of opportunity	u

Collect baseline data. Take 20 to 25 subgroups from normal production. Calculate the center line and control limits.
Apply run rules. The four Western Electric rules are a common default: one point beyond 3 sigma; two of three consecutive points beyond 2 sigma on the same side; four of five consecutive points beyond 1 sigma on the same side; eight consecutive points on the same side of the center line. Use the rules your customer requires.
Write the reaction plan. For each chart: who stops the process, how suspect product is contained, who investigates, and when to escalate. Signals without reactions are just decoration.
Build the control plan. One line per characteristic, product and process, with method, sample size, frequency, control method, and reaction plan. Link it to the PFMEA (see G08).
Measure capability once the process is stable.
- Cp = (USL minus LSL) / (6 x within-subgroup sigma)
- Cpk = the smaller of (USL minus mean) / (3 x sigma) and (mean minus LSL) / (3 x sigma)
- Pp and Ppk use the same formulas with overall (long-term) sigma.
- Common guidance: Cpk of 1.33 or more is capable for ongoing production. For initial studies on new processes, the AIAG PPAP manual treats an index above 1.67 as acceptable, 1.33 to 1.67 as possibly acceptable, and below 1.33 as not meeting criteria. Agree targets with your customer.
Improve capability. Use chart signals and capability gaps to choose improvement projects. Center the process first (quick win), then reduce common cause variation (process change). Recalculate limits after a proven improvement.

What each maturity level looks like

QS-5 Process Control

Level	Assessment wording	What it looks like on the floor	How to move up
1	Little or no information is available on process performance.	Quality is judged by final inspection and scrap bins. Nobody can say how a process performed last week.	List customer requirements and complaints for the main products.
2	Customer requirements are understood and translated into Critical to Quality characteristics (CTQs).	Each main product has a CTQ list with specification limits, linked to drawings and customer needs.	Mark where each CTQ is created and start control charts at those points.
3	As Level 2, plus control charts track CTQs at key points in the process.	Charts are posted or on screen at key stations and updated by operators during the shift.	Write reaction plans for every chart and train operators to follow them.
4	As Level 3, plus SPC signals trigger corrective action through defined reaction plans.	When a point goes out of control, the operator stops, contains, and logs the action. Signal logs show the response.	Extend SPC to all CTQs, run capability studies, and set Cpk targets.
5	As Level 4, plus SPC is widespread and is the basis for continually improving process capability (Cpk).	Cpk is tracked by CTQ over time. Improvement projects are chosen from the lowest Cpk values.	Sustain: raise capability targets, reduce inspection where Cpk is high, and share methods across lines.

Evidence to look for

Document

CTQ lists linked to customer requirements.
Control plans with reaction plans, linked to PFMEA and process flow.
Control charts with calculated limits (not specification limits).
Capability studies with dates and sample sizes.
Signal and reaction logs.

Observation

Are charts at the stations where CTQs are created, and up to date this shift?
Do control limits differ from specification limits?
Ask an operator to show the last out-of-control point and what they did.

Statement (questions to ask and who to ask)

Operators: “What does this chart tell you? What do you do when a point falls outside the lines?”
Quality engineer: “Which CTQs are least capable? What project is working on them?”
Production manager: “When did SPC last stop the line, and what happened next?”

Management report

Number of CTQs under SPC versus total CTQs.
Cpk by CTQ, with trend.
Out-of-control signals and response times.
Scrap and rework for charted versus uncharted characteristics.

Common pitfalls

Specification limits on control charts. The chart then shows only whether parts are good, not whether the process changed.
Charts nobody reacts to. Operators plot points to satisfy an audit, and the process drifts anyway.
Over-adjusting. Operators tweak a stable process after every reading, which increases variation (sometimes called tampering).
Capability on an unstable process. Cpk from an out-of-control process predicts nothing.
Too many charts. Charting every dimension buries the few that matter. Focus on CTQs.
Never recalculating limits. After a real improvement, the old limits hide new signals.

Template

Control Plan Header and Columns

Field	Entry
Part number / name / revision
Phase (prototype / pre-launch / production)
Control plan number / date / revision
Core team
Customer approval (if required)

Process step	Machine / tool	Characteristic (product / process)	CTQ or special?	Specification / tolerance	Measurement method	Sample size	Frequency	Control method (SPC chart, poka-yoke, check sheet)	Reaction plan	Owner

Capability Study Record

Field	Entry
CTQ and specification (LSL / USL)
Gauge and MSA result
Subgroups / subgroup size
Stable? (control chart attached)
Mean / within sigma / overall sigma
Cp / Cpk / Pp / Ppk
Target and gap
Action and owner

First 30 days

Week 1: Choose one product family. List customer requirements and complaints, and build its CTQ list.

Week 2: Map where each CTQ is created. Confirm the gauges with MSA. Start collecting subgroup data on the top two CTQs.

Week 3: Calculate control limits, post the charts, write reaction plans, and train the operators on that line.

Week 4: Review signals and reactions with the team. Run the first capability study and draft the control plan for the product family.