How does SPC integrate with PDCA for tight tolerances?
SPC integrates with PDCA for tight tolerances by using charted measurement data to plan controls, detect assigned-cause signals, verify causes, and standardize proven corrections. PDCA supplies the workflow: Plan the critical features and reaction rules, Do the machining with controlled data collection, Check control-chart signals against inspection results, then Act on verified evidence. SPC does not prove part conformance by itself; statistical stability and drawing acceptance are separate questions. Buyers should ask for the control plan, charted characteristics, sampling logic, measurement method, and reaction owner before approving production.
PLAN: Define CTQ Features, Control Limits, and Reaction Rules
The Plan phase defines which features deserve SPC and how the data will be used. Engineers identify CTQ dimensions, datum relationships, tool-wear risks, temperature sensitivity, fixture conditions, and material behavior. Control limits must be derived from process data collected under a defined measurement method and sampling plan; specification limits remain drawing or customer requirements. Capability indices such as Cp and Cpk become decision evidence only after the process is shown stable and the measurement system is suitable for the characteristic. The approved control plan should state how these controls apply to workflows such as CNC machining, CNC boring, and CNC grinding for the named size or geometric characteristic. The plan should also define warning criteria, stop rules, offset-adjustment authority, suspect-part containment, and escalation when a signal appears. Material selection influences process behavior. For example, stainless steel SUS304 and aluminum 7075 have different thermal and cutting responses, so a control plan should not transfer between them without requalification. Nickel-based alloys such as Inconel 718 or cobalt alloys like Stellite 6B may justify different tool-life, coolant, or temperature controls when pilot-lot evidence shows dimensional drift.
DO: Collect Process Data Without Losing Control
During the Do phase, operators collect SPC data while following the approved route. Data may come from in-machine probing, air gauges, bore gauges, CMM checks, surface testers, or operator measurements. The measurement device, feature, datum alignment, sample timing, and data resolution must match the control plan. Operators record process metrics while executing tasks such as multi-axis machining or precision machining. As variation develops, control charts can expose signals associated with tool wear, thermal drift, fixture movement, or material-lot changes. Detection timing depends on sampling and measurement; a chart cannot guarantee that every nonconforming part is prevented. Adjustments should follow a written reaction plan rather than personal judgment. For aerospace, medical, or safety-related components, SPC can monitor a selected drawing characteristic only when the measurement method, datum setup, and chart logic match that characteristic; final GD&T conformance still requires the specified acceptance method. Surface consistency can also be monitored at this stage. Techniques like as machined finishing or electropolishing can be checked against a defined texture or finish requirement when the method and sampling frequency are stated. Buyers should ask who may adjust offsets, when production stops, and how suspect parts are identified and segregated.
CHECK: Separate Stability, Capability, and Acceptance
SPC belongs in the Check phase because chart signals must be interpreted before action is taken. X-bar and R charts, individual-moving range charts, histograms, control limits, and capability results answer different questions; chart choice depends on data type, subgroup logic, and collection frequency. A process can be statistically stable yet unable to meet the drawing tolerance, or a current lot can pass inspection while the chart shows an assigned-cause signal. Engineers should compare the chart with the control plan and specification without treating a control limit as a tolerance limit. Suspected signals require measurement review, tool-life data, fixture checks, material-lot evidence, and cause-and-effect analysis before an offset is changed. This matters for materials such as titanium (Ti-6Al-4V) and copper (C110), where thermal movement, work hardening, fixture response, or residual stress can shift a measured feature. The Check phase must also assess measurement-system error, because a false signal can trigger an unnecessary adjustment and move a conforming process toward nonconformance.
ACT: Verify and Standardize the Correction
The Act phase converts verified SPC findings into controlled process updates. A corrective action may change a tool-life limit, coolant condition, fixture detail, offset rule, measurement frequency, or a permitted step such as heat treatment . The change should be tested on the affected characteristic before it becomes the new standard. Revised control plans, work instructions, inspection programs, and traveler notes should identify the change, evidence, approver, and effective date. Teams serving aerospace and aviation, medical device manufacturing, and automotive manufacturing may use PDCA-SPC integration within their quality systems, but ISO 9001, AS9100, ISO 13485, and customer-specific obligations come from the applicable scope and contract. The buyer value is verifiable evidence: the supplier can show why containment began, how the cause was confirmed, whether capability changed, and which controlled document prevents an obsolete method from returning. For RFQs, request the CTQ feature, proposed chart type, measurement tool, subgroup or sample logic, capability criterion, and reaction plan for each tight-tolerance characteristic.
For buyer review, SPC should not be treated as a chart pasted into a quality report. A useful PDCA-SPC package names the feature, datum reference, measuring device, sample size, control chart, capability criterion, and reaction owner. The supplier should explain what happens when a point falls outside a control limit, when the agreed run rule identifies a sustained shift or trend, or when capability falls below the customer-approved criterion. A practical RFQ can request first-article data, pilot-lot capability review, production sampling frequency, and the format of nonconforming-part containment records. This evidence lets the buyer compare suppliers by process-control discipline rather than claimed machine accuracy, especially when several quotations offer the same tolerance band.
