ISO 2859-1 vs ANSI/ASQ Z1.4: how AQL sampling works on a valve inspection

6-step field procedure: how an inspector actually runs AQL on site

The tables are only half the job. This is the order in which we apply ISO 2859-1 / ANSI/ASQ Z1.4 at a valve factory, using the 300-piece lot from the sample report. The figures are the single-sampling, normal-inspection reading of the tables; exact cells must always be read from the buyer’s licensed edition of ISO 2859-1 or ANSI/ASQ Z1.4.

1. Confirm the lot size (N) and the readiness gate. Count the finished units in the inspection lot — one PO line, one type and size, made under the same conditions (300 valves in the sample report). Random sampling is only representative when the lot is 100% finished and typically at least 80% packed. If half the valves are still on the assembly line, we report that and agree a new date or a DUPRO instead of sampling what happens to be ready.
2. Select the inspection level. General Inspection Level II (G-II) is the default for visual and workmanship checks. Special level S-3 or S-4 is used for time-consuming gauge or instrument checks such as dimensions and coating DFT. Pressure-test witnessing and loading supervision are not taken from the AQL table: the witness quantity follows the PO or the client ITP (for example 20 pieces, or 100%), while the factory tests 100% of the lot to the product or test standard.
3. Look up the code letter and the sample size (Table I → Table II-A). Table I gives the sample-size code letter from the lot size and the inspection level; Table II-A (single sampling, normal inspection) converts the letter into a sample size. For N = 300: G-II gives code letter H → 50 pieces; S-3 gives code letter D → 8 pieces.
4. Read Acceptance (Ac) and Rejection (Re) numbers for each defect class. Each class has its own AQL column: Critical — commonly AQL 0, i.e. Ac 0 / Re 1, so any single critical defect fails the lot; Major — commonly AQL 2.5 (1.0 or 1.5 for stricter buyers); at n = 50, AQL 2.5 gives Ac 3 / Re 4; Minor — commonly AQL 4.0 (or 2.5); at n = 50, AQL 4.0 gives Ac 5 / Re 6.
5. Follow the table arrows — the most common mistake. Where a code-letter row meets an AQL column at a downward or upward arrow instead of a number, follow the arrow to the first number shown, and use the sample size of that new row for that defect class. For example, at code letter D (8 pieces) the AQL 1.0 column normally points down to the row with Ac 0 / Re 1 and a sample of 13 pieces. Different classes can therefore end up with different sample sizes on the same lot. If the adjusted sample size equals or exceeds the lot size N, inspect 100% of the lot for that class. Always read the exact cells from the buyer’s licensed ISO 2859-1 or ANSI/ASQ Z1.4 tables — we do not reproduce the copyrighted tables here.
6. Draw cartons across the whole lot, then count by unit. Samples are drawn from cartons spread across the entire lot — in the sample report, cases 2, 5, 9, 13, 17, 21 and 24 of 25 wooden cases — never from boxes the factory has pre-staged. Counting is unit-based and by worst defect: if one valve has 1 Major and 1 Minor defect, it is counted once, under Major, in the unit tally, and both defects are photographed and listed. Measured specification deviations — dimensions outside the drawing tolerance, coating DFT below the specified minimum, a missing EN 10204 3.1 MTC, or any leak on a pressure test — are logged separately and lead to PENDING or FAIL; they are never offset against the visual AQL totals.

Why a measured specification deviation is logged separately

A visual AQL plan answers a different question from a measured test. Sand inclusions and paint runs are classified and counted. A coating-thickness average that falls below the specified minimum, or a dimension outside the drawing tolerance, is a specification deviation on that sample. In our sample report, the low DFT reading on sample S3 is listed in the defect list with the note that it is not counted in the visual AQL totals.

That is why a lot can be “within AQL” for workmanship and still not be a PASS. The measured check has its own acceptance rule, taken from the coating specification or the drawing, not from the AQL table.

PASS, FAIL and PENDING

Our reports use three overall conclusions:

• PASS — every section conforms, and visual defects are within the AQL acceptance numbers.
• FAIL — any critical defect, defects at or above a rejection number, or a failed pressure test.
• PENDING — a final decision is needed from the client. Typical reasons are a missing document, a measured deviation outside the agreed checklist, or a required test that was not witnessed.

In the sample, visual results conformed, quantity, packing, marking, dimensions, operation and the witnessed pressure tests conformed, but the report is PENDING because the MTC for heat H2611-07 was not on site and one DFT sample was below the specified minimum. The client is asked to decide on shipment release.

Sources

• ISO 2859-1, Sampling procedures for inspection by attributes — Part 1: Sampling schemes indexed by acceptance quality limit (AQL) for lot-by-lot inspection — iso.org
• ANSI/ASQ Z1.4, Sampling Procedures and Tables for Inspection by Attributes — ASQ
• Our sample valve PSI report (fictitious data), Sections 1 and 3.
• Pre-shipment inspection service page.

We report facts from the samples inspected. Standards named in these guides are public standards we work to, not certifications or accreditations we hold. We do not issue product certificates or accreditation marks. Test pressures, durations, leakage rates and acceptance numbers must be read from the licensed edition named in your purchase order; we do not reproduce copyrighted tables from the standards.