← Back to Blog

How Do You Inspect a Lot of Stainless Steel Ball Valves Before It Ships?

What does a pre-shipment inspection of a ball valve lot decide?

It decides whether this lot ships. It does not decide whether the supplier is any good. Qualifying the supplier is the work of the audit in What Should You Check When You Audit a Ball Valve Factory? and of the approval of the first sample in How Do You Approve a Stainless Steel Ball Valve Sample? A lot inspection asks a narrower question: does what was built match the order and the approved sample.

Which checks belong on every valve, and which can be sampled?

Anything that decides whether a valve holds pressure or seals belongs on every valve, tested by the factory under a record the buyer can read. Anything that can be measured or looked at can be sampled. Valve World’s inspection guide lists the functional tests for a ball valve as a high-pressure hydrostatic shell test, hydrostatic and pneumatic seat tests, a torque test, a seat cavity relief test and double block and bleed testing, and says the extent of each, full batch or sampling, is set by the valve standard and any supplementary requirements. The order has to settle it. A working split:

How is a lot defined, and why does it matter?

A lot is the group of valves a sampling plan treats as one population. ISO 2859-1 is intended primarily for a continuing series of lots, so each lot should come from one process. For valves, the practical boundary is the heat number and the production order. Mixing a good heat with a bad one hides the bad heat in the average, and a sample drawn across both will not show it.

A first order from a new supplier is not a continuing series. ISO 2859-1 can be used on isolated lots, but its users are advised to consult the operating characteristic curves to find a plan that gives the protection they want. ISO 2859-2 provides plans indexed by limiting quality for that case.

How does AQL sampling pick a sample size?

Lot size and inspection level give a code letter, the code letter gives the sample size, and the AQL with that sample size gives the number of defective valves at which the lot is accepted or rejected. General Level II is the usual default. Level I lowers the sample size and Level III raises it, and the special levels S-1 to S-4 are described as reserved for destructive or high-cost tests.

Two cases from inspection-industry descriptions of Z1.4. A lot of 4,000 at General Level II takes code letter L and a sample of 200, and at AQL 2.5 the lot is accepted with 10 or fewer defective valves and rejected at 11. A lot of 501 to 1,200 takes a sample of 80.

What does an AQL plan actually let through?

The 200-valve plan above accepts a lot that is 5 per cent defective, which is 200 bad valves in 4,000, about 58 times in 100. AQL is the quality level at which a plan accepts the lot most of the time. It is not a ceiling. ISO 2859-1 aims to push suppliers to keep their process average at or better than the AQL, and the standard itself notes that lots at the AQL can still be accepted fairly often and that the designation does not imply a desirable quality level.

Does accepting only a clean sample solve it?

No. Even if a lot is accepted only when the sample contains no defective valve, a lot that is 1 per cent defective is accepted 43 times in 100. Take 80 valves from a lot of 1,000. With 5 bad valves in the lot, 0.5 per cent, it passes about 66 times in 100. With 10 bad, 43. With 20 bad, 19. With 30 bad, 8. A plan that accepts on zero is not an ISO 2859-1 plan, and it is used here only to show the limit of any sample.

Sampling can show that a process is bad. It cannot show that an untested valve is good. For a defect that means a leaking valve, the only check that covers the whole lot is the factory testing every valve. The buyer’s sample then does a different job: it checks that the factory’s records describe what the valves actually do.

What AQL should be set for each kind of defect?

Set it for each defect class, and write into the order what counts as each class. Inspection firms commonly set separate AQLs for critical, major and minor defects, with critical at 0. QIMA describes 2.5 as the most common level for general consumer goods and 4.0 as a level where minor cosmetic defects are acceptable. The standard does not tell you which valve defect belongs in which class. A starting classification to adapt:

How do you run the inspection, in order?

1. Define the lot. Group the valves by production order and heat number, and keep a lot to one process. Count the cartons and check the quantity against the order.
2. Check the documents first. A certificate for every heat in the lot, and a test record for the lot with pressure, duration, medium and the gauge used. Stop if a heat has no certificate.
3. Draw the sample at random. Take valves from across the cartons and pallets, not from the top layer, in the number the sampling plan gives.
4. Match the marking to the paper. Heat number and grade on each sampled body against the certificates, and against the approved sample.
5. Measure and look. Dimensions, end connections, finish, cleanliness and packing, against the drawing revision on the order.
6. Check the alloy. Handheld XRF on the sampled valves for the alloying elements. It does not read carbon, so a low-carbon grade needs the certificate or an OES or LIBS reading.
7. Repeat the leak test on a few. Witness the factory test or repeat it on a small number of valves, to check that the factory records match what the valves do.
8. Count defects by class and decide. Compare the count of critical, major and minor defects with the accept and reject numbers. On rejection, apply the disposition written in the order.

What belongs in the order?

A lot inspection can only reject against what the order wrote down. The items that settle it:

• How a lot is defined, by production order and heat number
• The sampling standard and inspection level, for example ISO 2859-1 at General Level II
• The AQL for each defect class, and which defects fall in each class
• The checks the factory performs on every valve, and the records it hands over
• The approved reference sample that lots are compared with
• What happens to a rejected lot: sorting, rework or replacement, and how it is re-inspected
• Whether inspection tightens after a rejected lot

ISO 2859-1 includes switching rules for this: a switch to tightened inspection, or a halt to sampling, if quality slips, and reduced inspection at the discretion of the responsible authority if it stays consistently good.

What can a lot inspection not show?

Whatever is inside the casting. Porosity and heat treatment leave no mark on a finished valve, which is why porosity and the furnace record sit with the audit. How the seat behaves after a year in service. Whether the supplier’s certification covers the site that shipped the lot, which What Does a Valve Manufacturer’s Certification Actually Prove? sets out. And, as the two curves show, whether the lot you accepted contained defective valves. What the factory’s pressure test proves is the better evidence for that.

Frequently Asked Questions

What is a pre-shipment inspection for ball valves?

It is a check of a finished lot against the order before it ships. It covers quantity, documents, marking, dimensions, appearance and packing on a sample, and the factory’s test records. It decides whether the lot ships and does not qualify the supplier.

What sample size does AQL give for 1,000 valves?

For a lot of 501 to 1,200 at General Level II, the sample is 80 valves. The accept and reject numbers then depend on the AQL chosen and on the edition of the standard, so read them from its table.

What AQL should a buyer set for ball valves?

The standard does not choose it. Inspection firms commonly set the AQL separately for critical, major and minor defects, with critical at 0. QIMA describes 2.5 as the most common level for general consumer goods, which is a reason to consider a stricter level for pressure-bearing parts.

Does a lot that passes AQL contain no defective valves?

No. A plan with a 200-valve sample and acceptance of up to 10 accepts a lot that is 5 per cent defective about 58 times in 100. ISO 2859-1 itself says the AQL designation does not imply a desirable quality level.

Should the buyer repeat the factory’s pressure test?

Where the order requires the factory to test every valve, the buyer reads the records and witnesses or repeats the test on a few valves, to check the records against the valves. Re-testing a sample cannot replace testing every valve.

What is the difference between AQL and limiting quality?

AQL plans in ISO 2859-1 are indexed for a continuing series of lots. ISO 2859-2 gives plans indexed by limiting quality for isolated lots, such as a first order from a new supplier.

How is a lot defined for ball valves?

By production order and heat number, from one process. A sampling plan treats the lot as one population, so mixing a good heat with a bad one hides the bad heat in the average.

Published by LINS Valve Industrial Co., Ltd., Taichung, Taiwan. Last Updated: 2026-10-20