| Lot size (N) | AQL | Confidence | c | Mode | Typical sample (n) | Accept / Reject rule |
|---|---|---|---|---|---|---|
| 500 | 1.00% | 95% | 0 | Normal | ~299 | Accept if 0 defects; reject at 1+ |
| 2,000 | 2.50% | 90% | 1 | Normal | ~120 | Accept if ≤1 defect; reject at 2+ |
| 10,000 | 0.65% | 95% | 0 | Tightened | ~575 | Higher sample to reduce escape probability |
| 1,200 | 4.00% | 80% | 2 | Reduced | ~60 | Lower effort when performance is stable |
- Defect rate conversion: AQL% → p = AQL / 100.
- Acceptance probability: Paccept(p) = Σk=0..c C(n,k) pk(1−p)n−k (binomial approximation).
- Detection confidence: CLachieved = 1 − Paccept(p). The tool finds the smallest n that meets your target CL, capped by “Maximum sample cap”.
- Mode adjustment: Normal uses the base n. Tightened increases sample (~25%). Reduced decreases sample (~20%). The achieved CL is recalculated after adjustment.
- Risks: Producer’s risk α = 1 − Paccept(AQL). Consumer’s risk β = Paccept(LTPD).
- Rectification indicators: AOQ(p) ≈ p·(N−n)/N · Paccept(p) and ATI(p) ≈ n·Paccept(p) + N·(1−Paccept(p)).
- Cost model: InspectionCost = (n / (rate·inspectors))·hourly + setup, EscapeCost ≈ (N−n)·p·escapeCost, ExpectedTotal = InspectionCost + EscapeCost.
- Enter your lot size, part name, and lot identifier.
- Choose severity and set AQL, confidence, and acceptance number (c).
- Set LTPD to evaluate consumer risk for bad lots.
- Pick a mode or use Auto with accept/reject history.
- Review the output above the form after submission.
- Use CSV or PDF downloads for audits and suppliers.
Inspection objectives and defect severity
Inspection plans connect product risk to sampling effort. Critical characteristics normally demand c=0 and higher confidence to reduce escapes. Major defects often use balanced settings that protect downstream assembly. Minor defects can tolerate slightly higher acceptance numbers when suppliers show consistent capability. Documenting severity, AQL, and the rationale supports repeatable decisions across shifts and sites. Use separate plans for incoming, in-process, and final inspection when exposure differs.
Selecting AQL, LTPD, and confidence targets
AQL represents the defect rate you are willing to accept as “good.” LTPD represents a poor-quality threshold you want to catch, used to estimate consumer risk. Confidence determines how strongly the sample should detect lots at the AQL rate beyond the acceptance number. Higher confidence drives larger samples, so align targets with safety, warranty exposure, and customer requirements. When contracts specify an operating characteristic curve, validate c and n against both risks.
Operational capacity and inspection cost planning
Sampling is only useful if it fits real capacity. The builder converts sample size into hours using inspection rate and inspector count. Adding setup cost captures staging, tooling, and measurement preparation. Expected escape cost provides an economic view of under-sampling by valuing defects that bypass inspection. Comparing expected total cost across modes helps justify tightened checks during instability and reduced checks during sustained performance. For destructive tests, use caps to stay feasible and log deviations as concessions.
Interpreting acceptance rules and risk metrics
The plan outputs accept if defects ≤ c and reject at c+1. Acceptance probability at AQL approximates how often a good lot passes; producer risk is the complement. Consumer risk is the acceptance probability at LTPD, indicating exposure to bad lots. AOQ and ATI assume rectification on rejects, highlighting outgoing quality and inspection workload under typical conditions. Use beta targets to decide when to quarantine lots, and record risk levels in supplier scorecards for trend-based escalation and management review.
Continuous improvement and switching logic
Inspection should evolve with supplier behavior. Auto mode uses recent rejects to suggest tightened sampling when risk rises, and consecutive accepts to suggest reduced sampling when stability is proven. Track defect patterns, measurement variation, and corrective action closure to refine settings. Periodically review AQL and LTPD with stakeholders so targets remain aligned with design changes, field feedback, and process capability.
c is the maximum defects allowed in the sample. If observed defects are ≤ c, the lot is accepted. If defects are ≥ c+1, the lot is rejected and can trigger 100% screening.
If the maximum sample cap is too low, the tool cannot reach the requested confidence. Raising the cap, lowering c, or reducing the confidence target will typically increase achieved confidence.
Use AQL to represent acceptable process performance and LTPD to represent unacceptable quality that customers will not tolerate. Set LTPD higher than AQL, then review β to confirm risk is acceptable.
Use tightened when recent rejects, process changes, or new suppliers raise uncertainty. Use reduced after sustained acceptance and stable capability. Auto mode recommends one using your recent reject count and consecutive accepts.
AOQ estimates average outgoing defect rate when rejects are rectified. ATI estimates average units inspected per lot, combining sampling and full inspection on rejects. They help compare workload and protection across plans.
Include part name, lot ID, lot size, AQL, confidence, c, sample size, decision rule, and mode. Add inspection level, time and cost estimates, and any special notes for measurement methods or containment.