After a valve failure: the review that improves the programme

1. Why one failure deserves a review

A failure is unwelcome information, but it is information. A structured review extracts it before the record goes cold. The goal is not blame. The goal is to learn whether one valve failed, or a whole class is at risk.

There is always pressure to fix the valve and move on. The repair matters, but the review protects every similar valve.

Every review should answer three questions. What did this valve actually experience? What evidence do we still hold? What should change for similar valves? Without this discipline, the same failure repeats elsewhere.

Treat the review as normal work inside valve integrity management, not a reaction to disaster. Sites that review routinely catch drift early. Sites that review only after major events learn slowly.

2. Collect the evidence before it disappears

Evidence decays quickly after a failure. Records vanish, memories smooth over, and the valve itself may be removed. Collect within days, not weeks.

1. Operating history: cycles, torque trend, date of the last full stroke.
2. Maintenance history: injections, products used, doses and the pressure response seen.
3. Service conditions: fluid, duty, and whether the gas is dry or lubricated.
4. Screening observations: pressure rise across the closed valve, sound, visible leak.

Keep screening in its place. A pressure rise indicates seat leakage; it does not prove it. The confirming step is a seat-leak test under agreed conditions. Examples are pressure decay across the closed valve, or double block and bleed monitoring. Our passing valve screening check describes the quick screen. Rising torque over years tells its own story in valve torque rising year on year.

3. Causes stay an open list

Resist closing the cause list too early. The first plausible cause is often only the most visible one. Common causes include seat or ball scoring from debris. Lubricant can wash away and leave the seal dry. In dry gas, some organic sealants harden and seize valves. Long static periods do the same.

Failure statistics support the open-list habit. A risk-based valve integrity study of upstream and midstream sites, published by Valve World, reported a clear spread. Passing valves were 61 per cent of failures. Seized valves were 28 per cent. External leakage was 6 per cent, and other modes 5 per cent. Distributions vary by population, so treat these numbers as prompts, not answers.

Then test each candidate cause against the evidence. A seized valve with clean seats points one way. A passing valve with rising torque points another way. The evidence decides, not the first impression. Ask also what the records never captured. Missing records are themselves a finding.

4. Actions at three levels

Findings should drive action at three levels, not one.

1. This valve. Decide between treatment, assessment and replacement. In suitable cases, online treatment with cleaner, lubricant or sealant is the least intrusive option. Our clean, lubricate and seal service covers this work.
2. The population. Check sister valves of the same class for the same signs. A valve condition survey finds candidates before they fail.
3. The programme. Adjust intervals, products or records, so the cause loses its grip.

Injection treatment has gates, and they come first. The valve must have injection fittings in good condition. Its design and service must suit injection. Site rules and the permit-to-work must allow the job. Where OEM documents exist, consult them for limits and warranty terms.

Be honest about endpoints. A valve with a damaged closure surface may keep passing after treatment. Then the right answer is a proper assessment, or replacement, not another injection.

5. Feed the review back into the programme

The review closes only when the programme changes. Write the finding into the valve record: cause, evidence and action taken. Next year, that entry becomes comparison data.

1. Hold the review within days, while evidence is still fresh.
2. File screening results and confirmation tests separately, so hints are never read as proof.
3. Record what was ruled out as well. Excluded causes save the next reviewer time.
4. Revisit the failed valve's class at the next survey.

Involve the people who operate the valves daily. Their account of the last hard stroke is evidence too. Repeated failures in one class usually mean an ageing population, not bad luck. That question is covered in ageing valve populations: strategy for life extension and replacement.