Hourly Stamped Part Inspection: Cycle Time, Sampling Logic, and Reaction Rules

An hourly stamped-part inspection plan should define much more than “check one part every hour.” It should state which part is selected, when the inspection window opens and closes, which characteristics are evaluated, how long each task may take, what happens when a scheduled check is missed, and which material must be contained when a result signals risk.

Build the plan around three connected controls. First, set a cycle-time budget that includes part retrieval, identification, loading, measurement, analysis, review, reporting, data transfer, and recovery from normal interruptions. Second, combine a time-based sample with event-triggered checks after setup, coil or material change, tool maintenance, process adjustment, restart, or a quality signal. Third, write reaction rules that distinguish a warning trend from an out-of-specification result and identify the owner, containment boundary, confirmation method, and release condition for each event.

Full-field 3D scanning is useful because one controlled acquisition can evaluate trim lines, holes, formed surfaces, springback, and local profiles. An automated measurement solution can also repeat the route, analysis, and reporting logic with less operator variation. The inspection frequency still has to match process risk and the plant control plan. Automation makes a defined plan repeatable; it does not decide the plan.

An automated optical inspection cell measures a stamped industrial component in a repeatable fixture.

Define What the Hourly Check Must Control

Start with the failure modes and process changes the check is expected to detect. Stamping variation can arise from die wear, material properties, lubrication, press settings, feeding, part ejection, temperature, coil transitions, and fixture or handling effects. These sources do not move every characteristic in the same way. A trim edge may drift gradually as a tool wears, while a hole position may change after a die adjustment and a panel surface may respond to a material or restraint change.

Create a short list of characteristics that are sensitive to the process and important to fit, function, assembly, appearance, or safety. Link each item to the released drawing, CAD revision, control plan, and applicable GD&T requirement. Typical stamped-part checks include hole and slot locations, flange angles, trim boundaries, surface profile, local sections, formed depth, springback, and selected gap or flush conditions in a checking fixture.

Do not fill an hourly routine with every characteristic from a full layout report unless the process and available capacity require it. A layered approach is usually clearer: a fast hourly program monitors a controlled set of process-sensitive features, while a longer periodic or event-driven program evaluates broader coverage. Both programs should use approved alignments and evaluation rules from the same 3D inspection workflow.

Build a Complete Cycle Time Budget

The inspection cycle begins when the sample leaves the process and ends when the result is available to the person who can act. Scanner acquisition time is only one element. A credible budget includes sample retrieval, safe transport, cleaning or stabilization, identity verification, loading, fixture confirmation, scanning, processing, alignment, feature evaluation, report generation, result review, transfer to SPC or QMS, unloading, and return or disposition.

Measure elapsed time and labor touch time separately. An automated cell may process data while an operator performs another task, so labor demand can be lower than elapsed inspection time. Conversely, a fast scan can still produce a slow response if the part waits in a queue or a reviewer must approve every report. The production control plan needs the elapsed time to result because that value determines how much additional material may be produced before a problem is recognized.

Use a time study on representative parts and normal shifts. Record the median, upper-tail performance such as the 95th percentile, and credible recovery time for events such as a failed identity check, incomplete coverage, or data-transfer retry. An 11-minute average does not establish that a routine will reliably fit a 15-minute window; upper-tail cycle times and recovery events determine the required margin. The plan needs capacity for ordinary variation, planned verification, and the actions required when a result is questionable.

For an hourly check, define a sampling window instead of relying on an exact clock minute. A plant might require the selected part to be produced within a stated interval and the verified result to be available before a defined deadline. The specific window should reflect production rate, lot size, travel distance, inspection capacity, customer requirements, and containment risk. Record what counts as on time and how a late result is escalated.

Match the Measurement Program to Stamped Parts

Flexible sheet metal requires controlled support. Define the fixture, locator identity, clamp sequence, restraint condition, and allowed handling. If the drawing requires a free-state evaluation, the inspection method must not hide deformation through excessive clamping. If a functional checking fixture is required, use the approved locators and forces and verify their condition.

The measurement program should establish the specified datum reference frame before calculating conformance results. A best-fit alignment may help diagnose distributed springback, but it should not replace the required datum alignment for product acceptance. Define the surfaces and edge regions included in feature extraction, filters, fitting methods, mesh settings, exclusions, and decision rules. Trim edges and thin sheet boundaries need particular care because poor incidence, mixed pixels, burrs, and incomplete edge coverage can change the constructed result.

The current KSCAN-E intelligent wireless 3D scanner can support broad surface acquisition and detailed feature capture where a portable workflow is appropriate. For repeated hourly routines, lock the validated mode, resolution, target strategy, coverage requirement, alignment, and report template. A saved program should stop or flag the run when required datums or features are incomplete rather than returning a pass from partial data.

Digital path preparation and part verification help standardize a recurring automated inspection routine.

Set Sampling Logic from Risk and Process State

One part per hour is a schedule, not proof of process capability and not a universal sample size. Select the frequency and subgroup structure from the control plan, process history, characteristic severity, production rate, expected failure mechanism, detection delay, and measurement capability. Customer-specific requirements take precedence where they apply.

Use a time-triggered baseline to provide regular process visibility, then add event-triggered samples. Common triggers include first-off production after setup, new coil or material lot, die change or maintenance, press parameter adjustment, unplanned stop, quality alarm, sensor or feed fault, and restart after containment. An event-triggered check should not automatically replace the next scheduled sample unless the control plan explicitly allows that substitution.

Keep the sample representative. If a die produces several parts per stroke or uses multiple cavities or stations, rotate or stratify samples so that each relevant source is covered. If the process runs several coils, presses, dies, or part families, preserve those identities in the data. Combining unlike sources into one chart can hide a shift or create signals caused by known differences rather than an unstable process.

Define a missed-window rule before production. The operator or system should record the missed check, reason, time, and affected production interval. The plan should specify whether production may continue, whether the next sample must be expedited, whether extra parts must be checked, and who may authorize a deviation. Do not enter a later measurement as if it occurred in the original hour. Accurate time and lot identity matter more than a cosmetically complete chart.

Use Control Charts Without Confusing Limits and Tolerances

Specification limits come from product requirements. Control limits are calculated from process behavior under defined conditions. A result can remain within specification while a control chart signals an unusual shift, and a statistically stable process can still produce parts outside specification if its spread or centering is inadequate.

Choose charts that fit the data structure. Individual and moving-range charts may suit one result per interval. Subgroup charts require a rational subgroup whose observations share a meaningful short-term condition. Keep sample size consistent where the selected chart requires it. For a stamped part with many correlated characteristics, select a manageable group of sensitive measures or approved summary statistics rather than treating every mesh point as an independent observation.

Establish trial control limits from a representative baseline. Investigate special-cause signals and exclude observations from limit estimation only when an assignable cause and the corrective action are documented; retain the original records. Do not copy tolerance limits into the control chart or calculate new limits from a short unstable launch run. Review trends, runs, shifts, and dispersion signals under the plant's approved SPC rules. The NIST engineering statistics guidance explains that control charts display a characteristic over time against a center line and control limits, while rational subgroup selection determines which variation the chart can detect.

A robot mounted optical scanner captures full field geometry from a repeatable part position.

Prove the Measurement System Before Using the Trend

A reaction rule is useful only when the measurement result is trustworthy. Validate the complete installed process, including scanner, fixture, target layout if used, software configuration, alignment, part loading, operator influence, environment, and report logic. Repeat runs on stable parts, include removal and reloading, and compare critical results with an accepted reference method under equivalent datum and evaluation definitions.

Perform the measurement system analysis required by the organization and customer. The study should reflect the intended production method, part range, operators or automated routines, and the variation that can enter during hourly use. Confirm that rounding and report resolution do not mask small changes. Monitor a check standard or controlled artifact to separate measurement drift from actual production movement.

Calibration supports traceability, while interim verification shows that the system remains fit between calibration events. The SCANOLOGY calibration center describes calibration capability for optical 3D measurement systems. The plant still needs a local verification schedule, acceptance limits, records, and a defined response to a failed check.

Write Reaction Rules as Executable Instructions

Reaction rules should identify the trigger, immediate action, containment boundary, verification step, owner, notification path, and release condition. Separate four conditions: a scheduled pass, a warning trend, a measurement-system warning, and a confirmed nonconforming result. Each requires a different response.

A warning trend usually calls for review and an increased check without automatically declaring the product nonconforming. A measurement-system warning calls for control of the measurement process before product decisions continue. A confirmed out-of-specification result calls for product containment and production response under the approved control plan.

The initial containment boundary often begins after the last known acceptable result, but that boundary should not be assumed blindly. Consider the last verified process event, coil change, tool adjustment, restart, suspect time, and traceability available. Quality and production owners should define the exact rule before launch. Preserve suspect material by time, lot, container, or serial identity so it can be evaluated without mixing it with known good output.

Do not adjust the press from one questionable measurement without following the confirmation rule. First verify part identity, CAD and program revision, fixture loading, datum coverage, system status, and report completeness. Repeat measurement only when the procedure permits it, and retain both results. If a confirming sample is required, state whether it comes from the same stroke, current production, the last known good interval, or the contained population.

A modular automated inspection layout can combine repeatable part positioning with controlled optical measurement.

Choose a System Architecture That Fits the Hourly Routine

For mixed medium-to-large stamped components, the AM-CELL C Series optical automated 3D measurement system is a relevant starting point. It combines an optical 3D measurement system, robot or cobot, positioner, digitization software, and central control platform. Its modular layouts and multi-positioner options support repeated inspection of sheet metal, castings, welded assemblies, and other industrial parts. Vision-based checks can verify part type, placement, and clamping before measurement, while standardized routines support automated analysis and reporting.

For an hourly application, qualify the installed cell against the actual part portfolio, fixture, surface condition, program, report, and factory interfaces. The cycle-time study should include queueing and response time, not only robot motion. Validate safe loading, part identification, route recovery, collision controls, calibration or verification behavior, and the treatment of incomplete scans.

DefinSight supports acquisition and data processing. In the AM-CELL C workflow, SCANOLOGY also describes CAD comparison, GD&T analysis, reporting, and production-system data exchange. For production use, control the program and template revisions, user permissions, characteristic mapping, units, rounding, pass-fail logic, and transfer acknowledgments. A failed MES or QMS transfer should create a visible exception rather than silently completing the cycle.

SCANOLOGY is an industrial 3D scanning brand of Scantech (Hangzhou) Co., Ltd. System selection within that line should follow the part envelope, required characteristics, throughput, automation level, working environment, traceability needs, and verified measurement capability.

Preserve Data Needed for Containment and Improvement

Every result should identify the part number, part or container identity, press, die, cavity or station where relevant, coil or material lot, timestamp, operator or cell, fixture, CAD revision, inspection program revision, software version, equipment identity, verification status, characteristic results, status, and disposition. Record process events such as setup, adjustment, maintenance, and restart so a signal can be related to a plausible change.

Retain enough measurement evidence to support the intended investigation. A compact report may be sufficient for routine review, but the retention policy should state when raw scan data, meshes, alignment diagnostics, images, or audit logs are stored. Use consistent feature names and characteristic IDs so SPC, QMS, MES, and inspection records can be reconciled.

Measure the performance of the inspection process itself. Useful operational indicators include on-time completion, result-to-reaction time, missed checks, incomplete scans, verification failures, false calls, program changes, queue length, and data-transfer exceptions. These metrics show whether the hourly control is functioning as designed.

A compact automated station links repeatable optical capture with digital inspection results.

Release Checklist

• The control plan identifies the monitored characteristics and their process or product risk.
• The inspection window defines sample selection result deadline and treatment of missed checks.
• Time-triggered and event-triggered samples cover relevant dies stations cavities coils and restarts.
• The fixture restraint datum alignment coverage extraction and decision rules are approved.
• Cycle time includes transport loading processing review transfer and normal recovery.
• Measurement capability repeat loading reference correlation and interim verification are accepted.
• Control chart type subgroup logic baseline limits and signal rules are documented.
• Warning measurement-system and nonconformance responses have distinct instructions.
• Containment begins from a defined breakpoint and preserves lot and time traceability.
• Restart requires documented correction passing confirmation samples and authorized release.
• Program CAD report and interface revisions are controlled and changes trigger review.
• On-time completion reaction time and inspection exceptions are monitored after launch.

Conclusion

Hourly stamped-part inspection works when the timing, sample, measurement, and response form one controlled process. The plan should deliver a trustworthy result early enough to limit risk, cover both scheduled production and significant events, and tell the plant exactly what to do when the data changes.

The strongest implementation begins with a measured cycle-time budget and a short list of process-sensitive characteristics. It then adds representative sampling, validated measurement capability, controlled software and fixtures, traceable records, and reaction rules that define containment and release. With those controls in place, 3D scanning can provide both fast production feedback and detailed geometric evidence without turning the hourly check into an ambiguous report.

Frequently Asked Questions

Is one part every hour enough for stamped parts?

Not by itself. The frequency and sample structure must follow risk, production rate, process history, characteristic severity, expected failure mode, measurement capability, and customer requirements. Add event-triggered checks after changes that can shift the process.

Should an event triggered sample replace the hourly sample?

Only if the approved control plan says so. Keeping both can preserve regular time coverage while verifying the specific event. The rule should prevent operators from making different substitutions on different shifts.

When does the hourly cycle time start?

Use a definition tied to the control objective. For most production controls, elapsed response time should start when the selected part leaves the process and end when a reviewed result reaches the person authorized to act. Also measure labor touch time separately for staffing.

What should happen after one failed measurement?

Verify identity, revision, fixture loading, datum coverage, measurement-system status, and report completeness. Follow the approved confirmation rule, retain all results, contain the defined production interval, and notify the named owners. Do not create an informal remeasure-until-pass practice.

Can a full surface color map replace characteristic results?

No. A color map is valuable for diagnosing springback and distributed surface behavior, but acceptance should use explicit dimensions or GD&T results under the approved datum alignment and decision rule.