A control plan can list many checks and still fail to control production. Vague characteristics, frequencies such as “periodic,” and reaction columns that say only “inform supervisor” do not tell operators what to measure, what to stop, or which product to contain.
Build the control plan from process flow, risk analysis, drawing and customer requirements. For every controlled characteristic, define the product or process feature, method, sample and frequency, record, specification, reaction, containment boundary, authority, and restart evidence. Keep it synchronized with the actual mold, machine, material, gauges, and production phase.
Use this guide when you need to write the control plan itself — one row per characteristic, with a method, a frequency, and a reaction. This guide defines the rows and how they are completed; it does not design the traceability records the rows point to. For how resin lots, cavities, and records are traced from receipt to shipment, see Material Traceability for Automotive Molded Parts: Resin Lots, Cavities, and Records. It does not define what a PPAP submission contains either: for the submission scope and who is responsible for each element, see PPAP for Automotive Molded Parts: Submission Scope and Supplier-Customer Responsibilities. And it does not compare the approval documents themselves: for which document to request — FAI, PPAP, or a T1 sample report — and what each one proves, see FAI vs. PPAP vs. a T1 Sample Report: What Should an Automotive Buyer Request?. Use this guide to write the control plan, 436 to trace the material, 482 to scope the submission, and 288 to choose the document.
Link each control to a requirement and failure mechanism
Begin with the released drawing, specifications, special-characteristic designations, process flow, PFMEA, lessons learned, customer-specific requirements, and demonstrated process data. Trace every row to why the control exists. If a dimension affects connector fit, name that function; if drying protects a hydrolysis-sensitive resin, link the material requirement and defect risk.
Do not copy every drawing dimension into one document without prioritization. The control plan is a manufacturing control strategy, while complete layout inspection may be managed elsewhere. Include product characteristics that require ongoing prevention or detection and process parameters that maintain the approved state. Customer-designated special characteristics require the customer’s prescribed symbols and controls.
For multi-cavity molds, decide whether results are recorded by cavity. Pooling samples may hide a cavity-specific drift. Include mold configuration, active cavities, interchangeable inserts, and first-off checks after maintenance. For insert molding, control insert identity, orientation, presence, and position in addition to final dimensions.
AIAG’s Control Plan 1st Edition provides current guidance on developing and using control plans, links to APQP, and safe-launch considerations (AIAG Control Plan). Use the licensed manual and applicable customer requirements for mandated format and details; a blog checklist is not a substitute.
Map controls across receiving, material preparation, setup, startup, steady production, inspection, rework if permitted, packaging, and shipment. A strong final check cannot compensate for unverified material or an uncontrolled startup.
Specify characteristic, method, and evidence so another shift can execute it
Each row should be unambiguous enough for a trained person to perform without asking the author what was intended.
| Field | Useful definition | Weak wording to avoid |
|---|---|---|
| Characteristic | Feature ID, cavity relevance and requirement source | “Critical dimensions” |
| Specification | Numerical or categorical acceptance with revision | “Per drawing” when several revisions circulate |
| Method | Gauge/test ID, fixture, condition and work instruction | “Check visually” |
| Sample/frequency | Quantity, cavities, time/lot/startup trigger | “Random” or “periodic” |
| Record | System/form and data retained | Check mark with no traceability |
| Reaction | Stop, contain, verify, escalate and restart steps | “Adjust if needed” |
One filled row — the connector datum from the four-cavity housing (illustrative)
| Illustrative example — hypothetical teaching values, not an AutoMoldingPro project and not a real part, drawing, or customer requirement. The example is the four-cavity housing this article already describes in "Choose frequencies from risk and process behavior, then define triggers": one connector datum, one appearance zone, and a material-drying requirement. Every value below is assumed for the purpose of showing how a complete row is written; substitute the actual characteristic, specification, gauge, frequency, and reaction from the released drawing and the customer requirements before using this row on a real part. | Characteristic | Specification | Method | Sample / frequency | Record | Reaction |
|---|---|---|---|---|---|---|
| Connector datum diameter, cavities 1–4 (product characteristic; customer-designated special characteristic) | 12.00 ± 0.05 mm, drawing rev. C (illustrative) | CMM, fixture F-20-01, work instruction WI-20-03 | 5 pieces from every active cavity at startup; then 5 pieces, rotating one cavity every 2 hours, during steady production | Cavity-tagged results on the SPC chart, form F-20, retained per the customer retention requirement | Stop press; hold parts since the last conforming check; notify the shift lead and the quality engineer; verify the gauge; first-off from every active cavity conforming before restart |
Read the row as the article’s field definitions require. The characteristic names the feature, the cavities it applies to, and the requirement source. The specification is numerical with a revision, not "per drawing". The method names the gauge, the fixture, and the work instruction, so another shift can execute it without asking the author. The sample and frequency name a quantity, the cavities, and the triggers — startup and steady production — with no "random" or "periodic". The record links the result to the cavity, so a drifting cavity cannot hide in a pooled sample. The reaction names the stop, the containment boundary (parts since the last conforming check), the roles to notify, the gauge verification, and the restart evidence — not "adjust if needed".
Blank template — copy one row per characteristic
| Copy this table into your control plan and fill one row per controlled characteristic, from the released drawing, the PFMEA, and the customer requirements. Leave every value cell blank until it is filled from the actual part data — no specification, gauge, frequency, or reaction in this template is pre-filled, and none should be copied from the illustrative row above. | Characteristic | Specification | Method | Sample / frequency | Record | Reaction |
|---|---|---|---|---|---|---|
Fill the header of the plan itself before the rows: part number and revision, production phase (prototype, pre-launch, or production), mold and cavity configuration, and the approval authority. A row is not complete until all six fields are filled — a characteristic with no reaction, or a frequency that says only "periodic", is an open row, not a control.
Measurement conditions matter for plastics. Define time after molding, temperature, moisture conditioning, fixture state, and datum simulation where these can affect results. Confirm the measurement system is suitable for tolerance and use. A high-resolution display does not prove measurement capability.
For process parameters, distinguish setup targets, validated windows, alarms, and monitored signals. Machine setpoints, actual values, and cavity response are not interchangeable. State whether a deviation requires immediate stop or engineering review. Password-protect recipes and record revisions where feasible.
Visual checks need boundary samples, lighting, viewing distance, zones, and defect definitions when appearance matters. Attribute gauges need calibration or controlled verification and version identity. Functional gauges should represent assembly without masking excessive force.
Records must link to part, revision, batch, cavity, material lot, mold, process revision, inspector, date/time, result, and disposition as appropriate. How those links are built — resin-lot identity, cavity records, and the chain from receipt to shipment — is the work of Material Traceability for Automotive Molded Parts: Resin Lots, Cavities, and Records; this guide says what each control-plan row must record, it does not design the traceability system. This allows trends and containment, not just audit presentation.
When the method in a row is a camera or sensor rather than a gauge, the same six fields still apply: the program or algorithm revision, the challenge pieces, the confidence threshold, the reject mechanism, the bypass control, and the reaction to an equipment fault belong in the Method and Reaction fields — they are specified in the automated-inspection paragraph in "Maintain the control plan as a living production baseline". An automated result is not a control until its row is filled the same way a manual check is.
Choose frequencies from risk and process behavior, then define triggers
No universal “one per shift” rule fits every molded characteristic. Frequency should reflect failure severity, occurrence mechanism, detection opportunity, process stability, cycle and cavity count, automated monitoring, customer requirement, and containment risk.
Use event-based checks at setup, material change, mold change, maintenance, cavity activation, recipe change, restart after interruption, and first production after approval. Use time-, quantity-, or lot-based sampling during steady production. Continuous monitoring can control some process variables, but sensors need limits, calibration, data integrity, and reaction rules.
Establish frequencies conservatively during launch or safe launch, then adjust only through controlled evidence and customer authorization where required. A stable capability study can support a change, but it does not remove controls for assignable events such as wrong resin or damaged tooling.
Illustrative example—hypothetical, not an AutoMoldingPro project. A four-cavity housing has one connector datum, one appearance zone, and a material-drying requirement. Startup checks cover all cavities; steady production rotates cavity samples at a defined quantity; the dryer dew point and temperature are monitored continuously with alarms; a full check is triggered after a stoppage exceeding the approved limit. The example illustrates different control mechanisms, not universal frequencies.
Review actual data periodically. If failures occur between samples, increase prevention, monitoring, or sampling and investigate the mechanism. If a control never detects relevant variation, confirm whether the process is genuinely stable or the method is insensitive. Frequency optimization should improve risk control, not simply reduce inspection labor.
Write reaction plans that protect product and restore a known state
A reaction begins with the signal: out-of-spec result, trend rule, alarm, missing material record, wrong insert, broken cavity feature, or failed gauge verification. It must tell production what happens immediately.
Define stop or hold authority; isolate material since the last known conforming check; identify affected machines, cavities, lots, containers, and shipments; notify named roles; verify the measurement; investigate cause; correct the process under authorization; inspect or disposition contained product; and document restart evidence. If customer notification or deviation is required, identify who decides and communicates.
Do not permit operators to adjust around a dimensional failure without limits. Uncontrolled parameter changes can conceal tool, material, cooling, or measurement problems and invalidate the approved process. Define allowed adjustments, escalation thresholds, and who can revise the recipe.
Restart criteria might include corrected equipment, verified material, successful first-off from every active cavity, acceptable functional test, restored process signature, gauge verification, and quality release. Link temporary actions to an expiry and permanent corrective action.
Test reaction plans with scenarios. Ask a shift to respond to a failed cavity-3 dimension, a dryer alarm, an unreadable material lot, and a broken insert sensor. Check whether records identify the containment boundary. Correct ambiguous instructions before production pressure exposes them.
Maintain the control plan as a living production baseline
Review the plan when the drawing, material, supplier, mold, cavity configuration, machine, process, gauge, packaging, location, software, or customer requirement changes. Also review after complaints, scrap trends, maintenance discoveries, audit findings, and corrective actions.
Control document revision and effective date. Remove obsolete copies from the floor, train affected personnel, update work instructions and inspection programs, and preserve approval. Confirm the PFMEA and process flow remain aligned; inconsistent documents create conflicting reaction rules.
At an RFQ or supplier review, request the proposed control-plan architecture: characteristics, cavity strategy, methods, frequencies, records, traceability, reaction and launch controls. Provide drawing and special characteristics, volumes, customer format, submission level, material condition, appearance standards, testing, and retention requirements.
For a production feasibility discussion, contact AutoMoldingPro with the released part data, annual demand, material, critical functions, validation scope, and customer-specific requirements. The output should identify proposed controls and evidence gaps, while final approval stays with the authorized customer and program team.
Audit the plan against an actual order from material receipt to shipment. Observe whether the operator can find the current recipe, distinguish every active cavity, access the specified gauge, recognize a boundary sample, enter actual data, and execute the reaction without informal help. Compare labels and records with physical stock. Then trace one failure through hold, investigation, disposition, corrective action, and restart. This floor exercise often exposes missing permissions, obsolete work instructions, unrealistic frequencies, or containment boundaries that a desk review cannot see.
Automated inspection is not a separate control — it is a Method entry, and it must fill the same six fields as any manual check.
A camera or sensor does not close a row merely because it reports a result. Define the inspected feature, algorithm or program revision, challenge pieces, confidence threshold, reject mechanism, bypass control, data retention, and reaction to equipment fault. Verify false accept and false reject risk using representative defects. If an automated system is unavailable, the approved contingency must provide equivalent protection or stop production; an undocumented manual workaround is not a control.
Conclusion
A usable control plan connects each characteristic to a risk, defines an executable method and frequency, and gives a containment-and-restart reaction. Build it from controlled requirements, prove it on the floor, review its data, and revise it whenever the approved production state changes.