A line should not treat every non-passing reading the same way. A part may have a setup issue, a measurement alarm, a correctable geometry condition, a no-progress response, a suspected material issue or a confirmed nonconformance. Without distinct routes, repeated correction can conceal a measurement problem, damage the part, mix lots or create an untraceable quality decision.
This article is a quality-routing framework. It does not claim that StraighteningTech currently supplies automatic sorting, specific sensors, rejection thresholds, traceability software or authorized rework rules. Those functions require verified local scope and quality approval.
*Engineering concept illustration. It shows a candidate production cell, not an installed automatic sorting system or a defined defect-detection capability.*
Define the Disposition States
At minimum, distinguish PASS, NOK, rework-eligible, stop/hold and engineering/quality review. The definitions must reference the controlled drawing, measurement method, process condition and authority to release. A process alarm is not automatically a reject; it may be an escalation signal that requires verification.
| State | Required definition |
|---|---|
| PASS | Meets the agreed released-state requirement with required records |
| NOK | Does not meet the defined requirement or has a confirmed nonconformance |
| Rework-eligible | Meets documented conditions for another approved attempt |
| Stop / hold | Requires setup, process-signal, quality or engineering review before further processing |
| Engineering review | Needs a controlled decision outside the normal automatic route |
Keep Setup Issues Separate From Part Nonconformance
Before routing a part as NOK, consider whether the signal could result from datum seating, fixture variation, sensor condition, part identity, wrong recipe or an unapproved support state. The procedure should define the allowed clean/reseat/repeat checks and prevent unlimited retries.
*Engineering concept illustration. It depicts a measurement decision point, not proof of an automatic classification or a specific alarm rule.*
See fixture repeatability and datum seating and crack detection during automatic straightening for the separation between a signal and its verified cause.
Set a Rework Limit Before the First Correction
Rework should be an approved route with a defined maximum number of attempts, permitted contact/support conditions, remeasurement requirement, surface/feature checks and authority to continue or stop. The limit is not a universal number; it depends on material, process stage, feature risk, customer requirements and validated response. When a part does not respond as expected, the appropriate action may be hold or reject rather than additional force.
Why Unlimited Retries Corrupt the Line
Allowing a line to re-correct a part until the gauge finally passes feels productive and quietly destroys two things. The first is the part itself: every plastic bending cycle consumes a share of the section’s low-cycle fatigue life, and the reverse-loading behavior described for reverse loading in straightening means later cycles move the material more per unit force than earlier ones did. A part that reaches tolerance on its sixth attempt is not equivalent to one that reached it on the first, even though both read PASS.
The second casualty is the line’s own information. When retries are unlimited, the gauge sees only eventual outcomes: parts that converge, slowly, under repeated force. A drifting correction setup, a worn anvil, a bad lot of stock and a genuinely struggling part family all produce the same visible symptom — more attempts per part — and none of them generates a stop condition. The retry count is one of the cheapest and most informative process signals a straightening line produces, and it only exists as data if the line is required to stop and route the part at a defined limit instead of looping.
Thinking of the Rework Limit as a Budget
A more useful mental model than “three strikes” is a budget of deformation that each part may spend. The budget is set by the material’s condition and the consequence of failure in service, and it is spent by every correction stroke — forward or reverse. A hardened safety-critical shaft and a soft structural bar may carry very different budgets, which is why no universal rework number is defensible and why the limit belongs to the part family’s validated process, reviewed by quality. The same budget concept explains a rule that otherwise looks arbitrary: a part that has consumed its budget and still fails is rejected even though one more attempt might conceivably pass, because the part that would emerge is no longer the part the drawing specifies in any meaningful sense.
Budgets also need edges. Two conditions that stop rework regardless of remaining attempts are no-progress — corrections that stop changing the geometry, indicating the process is no longer engaging the error — and overshoot oscillation, where each attempt moves the part from one side of tolerance to the other. Both are stop signals, not tuning opportunities, and both feed the integrity-review concerns that repeated plastic cycling raises.
Maintain Isolation and Traceability
Each disposition must preserve part/lot identity, drawing/recipe version, measurement condition, correction attempts, result, operator/automation event, time, review decision and physical location. Isolation must prevent a held or NOK part from returning to the PASS flow without explicit authorization.
*Engineering concept illustration. It emphasizes released-state verification and recorded disposition, not a promise of a DMC/MES interface or automatic sorter.*
Designing Segregation That Cannot Be Defeated
Disposition states only work if the physical line enforces them. The failure mode is always the same: a held or NOK part re-enters the good flow during a busy shift, through an operator judgment call, a shared conveyor or an unlabeled tote. Effective segregation design follows mistake-proofing logic rather than instruction signs. Good and rejected parts take different physical paths with no shared section where a part can be redirected by hand. Hold locations lock or require an access step that itself creates a record. Labels and bins are part-specific, not generic scrap tags, so a held-for-review part cannot decay into anonymous scrap or anonymous stock. Where volume justifies it, diverters and gates make the routing decision the machine’s, not the operator’s — the automation boundary discussed in manual versus automatic straightening applies to disposition as much as to correction.
The record side has the same anti-defeat requirement. A disposition that lives only in an operator’s memory, or in a spreadsheet anyone may edit, does not survive an audit or a containment event. The minimum durable record is the tuple — part identity, lot, drawing and recipe version, measurement condition, attempt count, result, decision, decision-maker, time — retained long enough to answer the customer question that eventually comes: which parts from that week, from that machine, from that shift.
What the Disposition Data Buys Later
A line that routes honestly accumulates, as a by-product, the dataset that every improvement program wishes it had. Attempt-count trends flag tooling wear and setup drift while they are still cheap to fix. Lot-to-lot shifts in incoming geometry point upstream to suppliers with evidence instead of complaints. The ratio of setup-issue holds to true part nonconformances measures how much of the line’s apparent quality problem is actually its measurement and fixturing — information directly relevant to planning a gage R&R study. And when a customer raises a field concern, traceability converts a week of panic into an afternoon of records. None of that value survives a culture of retry-until-pass; all of it is the natural dividend of the routing discipline described here.
Build a Verification and Escalation Route
- identify the part, lot, drawing and recipe;
- verify the measurement/datum setup within the approved checks;
- capture the reading, process state and any alert;
- apply only the allowed rework route when eligibility is confirmed;
- release and remeasure under the agreed condition;
- route PASS, NOK, hold or engineering review;
- retain traceability and physical segregation;
- use recurring patterns to trigger process or quality review.
For the needed evidence architecture, see straightening sample test and acceptance and Gage R&R for straightening lines.
No-Progress and Correction-Limit Gates
The exact phrase straightening correction limit no progress gate is medically polluted in SERP, so the canonical no-progress logic lives on the NOK sorting and rework-limit page. Define maximum correction cycles, force/stroke ceilings and mandatory stop/sort actions before production release.
FAQ
Does NOK always mean scrap?
No. A NOK state can require verification, controlled rework, engineering review or final rejection depending on the approved quality route.
Can a line retry until the reading passes?
No. Rework needs a defined maximum-attempt limit and quality authority. Unlimited retries can mask a process or measurement problem.
Does this article mean the line has automatic sorting?
No. It is a routing framework. Actual sorting, traceability and interface functions require separate verification.
Why does the attempt count matter more than the final reading?
Because the final reading describes where the part ended, while the attempt count describes what it cost to get there. Two parts reading identical PASS values are not equivalent if one took a single correction and the other took a full budget of repeated cycles. The attempt count is also the line’s earliest drift alarm — a setup, tooling or material change usually shows up as more attempts per part long before it shows up as rejects.
Should rework limits differ between part families?
Yes. The limit is a deformation budget set by material condition, feature risk and service consequence, so a hardened safety-critical shaft and a soft structural bar justify different numbers. The limit belongs to each family’s validated process with quality approval, not to the line as a single setting.
What is the minimum physical segregation a small line needs?
Separate paths or lockable, labeled, part-specific containers for PASS, hold and NOK — arranged so a held part cannot return to the good flow without a recorded, authorized step. The design principle is that doing the wrong thing must require effort, not just discipline. Escalate to machine-controlled diverters when volume or staffing makes human routing unreliable.
Sorting decisions are also capacity decisions – the manual vs automatic comparison and the ROI case for automatic straightening both price the NOK stream a manual line tends to hide.