Relief valve inspection records in Korea: what an HQ audit flags and why

The internal audit lands on a Thursday. Someone from group HSE picks three pressure relief valves off the equipment register, asks for the last inspection record on each, and by the afternoon all three are on the finding list. The records are in Korean, signed and stamped, and the valves are on the line lead-sealed. Nothing about any of them is wrong. They still do not answer the questions the corporate mechanical integrity procedure asks.

The instinct is to order fresh tests and get records that look familiar. That is the expensive way to close a finding. Pulling a valve that is legally inside its interval means a shutdown window, a broken seal, a re-test and a re-seal — and if the site runs a four-year interval for a reason, the off-cycle test now sits on the file as the site conceding the interval was never valid. This is a documentation gap, not an integrity gap, and it closes with documents.

Three rules, three intervals, and the valve does not decide which

The interval does not come from the valve, the manufacturer's recommendation or your corporate standard. It comes from which statute the equipment sits under, and there are three tracks.

The interval basis in Article 261 changed with a 2024 amendment — the direct-contact case moved from one year to two, and the rupture-disc case from two years to three. A site procedure or due-date field set before that is worth checking against the current edition. If the certificate carries no rule reference, the contractor's scope document will, and that one line settles which track you are arguing inside.

The four-year interval belongs to the site, not to the valve

This is the finding that comes back most often. The corporate standard says two years, the record says four, and the auditor reads it as a lapsed test.

The four-year interval is written into the rule, but it attaches to the workplace rather than the equipment — the process safety report regime plus an excellent implementation-status rating. Nothing on the valve says that, and nothing on the certificate says it either. The evidence is the evaluation result, and it lives with the site safety department.

Two consequences catch people. The rating is not permanent, so when the grade changes the interval changes with it, and every due date computed from a fixed interval field in the maintenance system is then wrong in the direction that matters. And a site newly inside the process safety report scope stays on the two- or three-year interval whatever a sister plant in the group runs — a difference auditors comparing two Korean sites log as an inconsistency.

The two- versus three-year split inside Article 261 turns on the physical arrangement, not on service severity. Fit a rupture disc upstream to deal with fouling or seat leakage and the interval basis moves with it — a change made by process engineering that rarely reaches the relief device register.

What the record proves, and what it was never asked to prove

The rule is specific about how a test is completed. The device is verified as operating properly at its set pressure using a gauge calibrated by a national calibration institution designated by the Minister of Trade, Industry and Energy, then lead-sealed. The certificate carries exactly that: a set pressure, a verdict, a calibrated instrument behind it, a seal. It is a compliance statement, and complete as one.

A procedure written around API 576 asks different questions. What did the valve do before anyone touched it — the as-found pop pressure. Did it hold at the seat. What condition were the internals in. Those are trending questions, and a compliance certificate is not built to answer them.

Both sets of data usually exist. The as-found figure and the bench observations go onto the test shop's own bench sheet, a separate document from the certificate that reaches your file. If nobody named it in the scope, it stays with the contractor. The question is not why the data is missing but whether anyone asked for it.

The gauge works the same way. Calibration accreditation runs through KOLAS, operated by the Korean Agency for Technology and Standards, and the traceability chain an auditor wants is not printed on the inspection record. Ask for the calibration certificate covering the gauge that was used — its number and calibration date — not for a statement that the shop is accredited.

What gets flagged, and what it means on a Korean site

What the auditor flagsWhat it usually means hereWhere to confirm itIf you read it wrong
Interval shown as four yearsThe workplace holds an excellent implementation-status rating; the rule allows four years on that basisThe evaluation result held by the site safety departmentAn off-cycle test and a broken seal to prove what the rule already permitted
Three years on one valve, two on a similar oneA rupture disc is fitted upstream of the firstThe relief arrangement on the P&ID, not the valve tagA non-conformance, and a valve re-tested before it was due
One pressure figure, no as-found valueThe certificate states the verified set pressure; as-found is on the shop's bench sheetThe inspection contractor's job file for that campaignAs-found gets logged equal to set, and trending runs on numbers nobody measured
Gauge traceability not shownThe rule requires a nationally designated calibration institution; its certificate is a separate documentCertificate number and calibration date for the gauge usedThe chain is reported as broken and a supplier audit opens that finds nothing
A rule article cited instead of API 510 or 576Korean records cite the domestic rule that required the testThe article number on the certificate or the contractor's scopeWritten up as no recognised code basis
Set pressure in kg/cm²G where the procedure expects bargOlder Korean nameplates and records use kg/cm²G; one kg/cm² is 0.980665 barThe nameplate and the vessel design document togetherA two percent conversion drift is logged as a deviation, and a correct valve is re-adjusted

Put the data in the scope before the turnaround, not after the audit

The bench sheet, a copy of the calibration certificate and the seal numbers cost nothing when named in the purchase order. After the campaign closes they are a favour, and a year later they may not exist. None of it requires anyone near a valve in service.

One line in the inspection and test plan does the work: as-found pop pressure before any adjustment, seat tightness result, calibration certificate number and date, seal number against each tag. Nothing there changes what the contractor does at the bench. It changes what leaves the bench in writing.

The seal number deserves its own field. When the next auditor asks whether the valve on the line is the valve on the certificate, that number answers it — captured on paper at handover to operations, not by anyone reading a tag on a hot line.

A valve that does not lift within tolerance as found is the item the corporate procedure wants to see, and the one that decides between overhaul and replacement. Disassembly, lapping, spring replacement and re-test are shop work, and fitting them inside the same shutdown window rather than the next is the difference between the valve back on its nozzle at start-up and a certified spare having to be found at short notice. The binding constraint is usually bench capacity during turnaround season, so overhaul work is booked long before the valve comes off.

The Korean original is the record; the English version is a map

The certificate exists to satisfy a Korean inspection, so the Korean original stays in the plant file and is not replaced by a translation. What headquarters needs is a mapping, not translated prose — a one-page English cover naming each field on the Korean record, the corporate procedure field it satisfies, and the field it does not. Built once, reused every campaign, it turns an hour of argument into a page an auditor reads and moves past.

Two terms are worth getting right on that cover, because flattening them makes an interval justification stop making sense. 안전밸브 and 압력방출장치 sit under different articles with different intervals, and both arrive in English as safety valve or relief valve depending on who typed it. Under the high-pressure gas track the statutory action is 조정, adjustment; under the occupational safety rules it is 검사, inspection. An English reader calls all of it testing — fine in conversation, a problem in a document that has to show which rule was satisfied.

The same problem arrives earlier with new equipment. Relief devices shipped with a package carry the manufacturer's set pressure certificate. Whether that certificate stands as the first record on the site register, or an inspection has to be repeated in Korea before the clock starts, follows from which track the equipment sits under — and it is settled far more cheaply while the package is still being scoped than after commissioning. Which documents travel with the package and which are produced in Korea belongs in scoping the delivery.

The lines that come back blank

When a plant asks us to arrange relief device inspection or overhaul in Korea, three things are almost never in the first message, and nothing else can be planned without them.

  1. Which rule the equipment sits under. Article 261 chemical facility, Article 116 boiler, or the high-pressure gas track. Valve lists arrive with tag numbers and set pressures and no statutory basis, and the interval, the scope and who may perform it hang off that one attribute.
  2. Whether the site holds an excellent implementation-status rating, and the date on it. Without it the four-year interval has no basis. With it, a good share of the findings never open.
  3. Whether as-found data is required. Not preferred, not appreciated — required, written into the scope, or it does not get recorded.

Send the valve list with tag numbers, the set pressures exactly as stamped rather than converted to barg, and the rule the equipment falls under, and we will come back with the scope and what the record will contain — get in touch.