Running a FAT at a Korean shop: the ITP, the witness points, the sign-off

You have flown to a shop outside Busan to witness a factory acceptance test. The equipment runs, the readings look reasonable, and after two hours someone produces a report for signature. What you sign at that moment determines what you can still raise afterwards — and by then it is too late to change what was tested.

The ITP decides the FAT, not the day itself

A factory acceptance test is the execution of a plan agreed earlier. That plan is the inspection and test plan, and it is where the content of the day was fixed.

It lists each operation, what is verified, against which document, by which method, and who attends. If a check is not in the ITP, it does not happen — not out of reluctance, but because the manufacturing sequence was built around the plan and the fixture or instrument may not be available.

So the leverage sits weeks before the visit. Once you are standing in the shop, the ITP is the agenda, and requesting something outside it means either a delay or a concession.

Three attendance levels, and the difference matters

ITP entries carry a designation for each activity, and the terms are used consistently enough to rely on.

The distinction that costs money is hold versus witness. Too many hold points and the shop waits for your travel schedule; too few and the operation you cared about happens while you are in the air. A realistic plan has a small number of hold points at genuinely irreversible steps and witness points elsewhere.

Who attends is a second decision with cost consequences. Sending your own engineer gives you someone who knows the application; appointing a third-party inspection agency gives you a presence without the travel, and their reports are structured for exactly this purpose. The two are not equivalent — an agency verifies against the documents you supply and will not raise a concern that is not written down, while your own engineer notices things nobody specified. For routine items an agency is usually the better economics; for the first unit of a new design it rarely is.

Put the notice period in writing

Every hold and witness point needs a notification period, and it needs to be written into the ITP rather than assumed.

Two weeks is common for international travel and is worth insisting on. Equally important is the rule for what happens if you cannot attend: does the manufacturer proceed with documented evidence, or wait? Without that clause, a missed date stops production entirely while messages cross time zones.

Agree the mechanism too — who notifies whom, by which channel, and what counts as confirmation. A notification that reaches a mailbox nobody monitors during a Korean holiday period is technically compliant and practically useless.

What a mechanical run test does not prove

Most FATs demonstrate that the equipment operates: it runs, moves, holds pressure, reaches speed. That is necessary and it is not the same as demonstrating that it performs.

Performance depends on the process medium and the real duty, and the shop usually cannot reproduce either. A machine tested with water instead of the process fluid, or with a substitute feedstock, has been shown to function rather than to meet its rated output.

The question to settle in advance is which characteristics can be verified at the works and which can only be verified on site. Then say so in the acceptance terms: what the FAT concludes, what remains open until commissioning, and what payment is tied to each. Leaving that implicit produces a dispute at exactly the moment when leverage has passed.

Timing within the manufacturing sequence deserves thought as well. A test placed only at final assembly detects problems when correcting them is most expensive; one placed after welding or heat treatment catches the same problem while the part is still accessible. Moving a witness point earlier costs nothing at planning stage and can save a complete remake.

Checks worth adding that are usually absent

Beyond the functional tests, several items are cheap to verify at the works and expensive to discover on site.

Confirm that documentation matches the equipment — serial numbers, material certificates, revision levels of drawings against what was built. Verify interfaces: connection dimensions, electrical supply and frequency, control signal types. Check markings: nameplate content, warning labels, language requirements. Confirm the loose items — special tools, spare parts, keys — physically exist rather than appearing on a list.

That last one is worth the time. Items promised as "shipping separately" have a way of not shipping, and once the container is sealed the conversation becomes much harder.

Photographs before it is crated

The FAT is the last time you see the equipment before it is packed. It is also the only opportunity to record its condition.

Take images of the complete unit from several angles, the nameplate, the interfaces, and anything with surface finish requirements. Photograph the packing itself if you are present for it — the internal securing, the desiccant, the barrier.

If a transport damage claim arises later, these images establish the condition at dispatch. Without them, the discussion between shipper, carrier and insurer has no fixed starting point and tends to settle against whoever has the least evidence.

Remote witnessing, and its limits

Where travel is not possible, remote attendance by video is increasingly accepted and genuinely useful for some activities. It works for observing an instrument reading, watching a sequence run, and confirming that a step was performed.

It does not replace physical presence for anything requiring judgement about condition — surface finish, fit, noise, the feel of a mechanism — nor for verifying that the item in front of the camera is the item in your order.

If remote witnessing is used, agree in advance what is being shown, that the camera operator follows your direction, and that the recording is retained. And treat it as what it is: adequate for many entries on the ITP and inadequate for the few that justified travelling in the first place.

The report is a record, not a formality

The document signed at the end determines what you accepted. Read it before signing, however late in the day it is produced.

It should state what was tested, under which conditions, with which instruments and their calibration status, what results were obtained, and which items remain outstanding. A report saying only that the test was performed satisfactorily records nothing that can be relied on afterwards.

Where something is unresolved, list it as an open point with an owner and a date rather than signing with a verbal assurance. Signing subject to a punch list is normal; signing a clean report and raising the issue by email the following week is not, and it will be answered by pointing at your signature.

The line on the form that is usually empty

Look at ITPs that produced disputes and the same field is blank: the acceptance criterion. The activity is named, the method is named, the attendance is named — and the column headed "acceptance criteria" says "as per specification" or nothing at all.

That reference is not usable on the shop floor. The specification may contain several applicable values, or state a requirement in terms the test cannot directly measure. Faced with a blank, the inspector applies the manufacturer's standard, which is how a test is passed against a criterion you never agreed.

Filling it costs nothing at ITP stage: a number, a tolerance, and the document clause it comes from. Two further fields are worth the same attention — the instrument accuracy required, and what constitutes a repeat test if a result falls outside. If you have an ITP in front of you, we can look at which rows need a criterion before it is signed off — talk to us about sourcing. What the documents around it have to contain is covered under mill certificates.