Final Inspection & Sampling (AQL)¶
Source attribution
This article is an editorially original synthesis. [Industry Knowledge] = common industry practice; [Framework Reference] = AQL sampling framework introduction; [Editorial Opinion] = researcher's original judgment. Model specifications are subject to data.hycarfilter. Backlinks: data home.
Finished-product inspection is the third gate, with one core question: which items must be 100% full-inspected, and which can be sampled by AQL? Getting the split wrong means either missing fatal defects or inflating cost.
I. Items that must be 100% full-inspected (safety-critical)¶
[Industry Knowledge] Once the following items fail, it is a leak / oil-starvation / filtration-failure-grade accident, so every unit must be inspected:
| Item | Category | Method |
|---|---|---|
| Leak-tightness | Spin-on oil filter, fuel filter | Inline pressurize-and-hold / differential-pressure leak detection; reject on excessive pressure drop |
| Thread go/no-go | Spin-on oil filter / fuel filter | Go/no-go gauge 100% pass inspection |
| Seal ring assembly | Spin-on filter | Visual / sensor confirmation of proper seating, no flipping |
| Outer dimensions | Air / cabin filter (hard-dimension parts) | Gauge 100% pass inspection (scrap if it won't fit the housing) |
[Industry Knowledge] The "100% leak-tightness test" of spin-on oil filters is an iron rule of the industry — because leakage cannot be seen visually and the consequence is fatal, it cannot be covered by sampling. This is also a point repeatedly emphasized in the source site's "quality oil filter vs. inferior oil filter" comparison (100% leak-tightness).
II. Items sampled by AQL (non-destructive, acceptable trace defects)¶
[Framework Reference] The AQL (Acceptable Quality Limit) framework (ISO 2859 / GB/T 2828) determines sample size n and accept/reject numbers Ac/Re according to lot size → inspection level → AQL value. Typical applications:
| Item | AQL tier (reference) | Description |
|---|---|---|
| Appearance defects (scratches / stains / printing) | AQL 1.0–2.5 | Minor defects |
| Dimensions (non-critical) | AQL 1.0 | Functional dimensions |
| Packaging / labels | AQL 1.5–4.0 | Packing, label correctness |
[Editorial Opinion] AQL is not an excuse to "allow a certain proportion of defective goods into the customer's hands," but the statistical language of sampling acceptance: AQL 1.0 means that under that sampling plan, the whole lot's "acceptable quality limit" is 1%. When buyer and supplier agree on acceptance, they must simultaneously agree on the AQL value, inspection level, and defect grading (critical / major / minor); otherwise "sampling" is an empty word.
III. Sampling of destructive tests on finished products¶
[Industry Knowledge] Destructive items cannot be full-inspected; they are sampled by frequency / batch:
- Static burst: first article + periodic sampling;
- Filtration efficiency / dust holding capacity: sampled by batch and sent to the laboratory (long cycle, see Laboratory Equipment);
- Bypass valve opening pressure: batch sampling.
IV. The combination logic of full inspection + sampling¶
[Industry Knowledge] The mature scheme is a three-layer stack of "safety-critical items 100% full-inspected + appearance/minor items AQL-sampled + destructive items batch-sampled." All three are indispensable: full inspection without sampling → cost out of control and appearance defects slip through; sampling without full inspection → fatal leaks are missed.
Editorial view¶
[Editorial Opinion] During a factory audit, asking "of your finished-product inspection, which items are full-inspected, which are sampled, what is the AQL, and how are defects graded" is a hundred times more useful than asking "do you have quality control." A factory that can state the full-inspection/sampling boundary and the AQL value is one that truly runs quality control as a system.
Continue reading: Laboratory Equipment