Evidence grades: witnessed test, operator report and inference

1. Why evidence comes in grades

Every valve record answers two questions: what was seen, and how much it deserves trust. The claim that a valve is tight can describe a witnessed test or a rough guess. Grading evidence keeps those two claims apart, and it costs nothing.

Some signals can only screen a valve, and honest language reflects that. A pressure rise across a closed valve suggests a seat leak. Bubbles at a drain, hissing sound, or heavy operation suggest the same.

1. Screening evidence: torque feel, sound, bubbles and pressure behaviour. These signals indicate a problem, and they never confirm one.
2. Confirming evidence: a seat-leak test under agreed conditions, such as pressure decay across the closed valve, double block and bleed monitoring, or a witnessed leak test.
3. Factory references: API 598 covers seat leakage testing, and ISO 5208 defines leakage classes. Both support factory acceptance, and neither is a field test result.

2. Grade one: the witnessed test

The top grade is a test that a named witness watched and recorded. The method is stated, the conditions are written down, and the readings sit beside the valve identity. Anyone can re-read the record later and know exactly what was proven.

A strong record says what was measured before treatment and after it. Seat tightness under an agreed test is one sound option.

Opening and closing torque is another, and it tracks operating effort directly. Measuring torque improvement after treatment shows what such a record looks like.

Be honest about the limits of this grade as well. A witnessed test proves the valve met a stated condition on a stated date. It cannot promise how long the result will hold in service.

It also cannot speak for any other valve on the site. The recorded conditions carry weight too, so read them first. A test far below service pressure proves far less than a test at service conditions.

3. Grade two: the operator report

Most site knowledge arrives as an operator report, and that is understandable. The valve became noticeably easier to turn after the treatment. The leak slowed after injection, and the pressure response felt normal.

This grade is common because it is cheap, fast and wide ranging. Treat it as real but unverified, and grade it accordingly. The operator saw something true, yet the report carries no baseline reading.

It rarely states the conditions, and memory smooths away detail over time. You can strengthen this grade a great deal with simple logging habits.

1. Log at the valve, at the time, never from memory at shift end.
2. Record the date, the person, the action taken, and what changed.
3. Note the conditions, such as line pressure and any valve cycling.
4. Keep the wording plain, because 'easier' and 'slower leak' beat 'fixed'.

A contemporaneous log is much stronger evidence than a confident recollection. The gap between those two records is exactly what grading is for.

4. Grade three: inference from symptoms

Inference reasons from an observed symptom back to a likely cause. High operating torque suggests a dry or contaminated seal area. A visible leak suggests seat damage or stem seal failure.

Inference is often correct in practice, and it is never proof. Use this grade for triage, and not for closing a case. Sorting valves into likely causes directs scarce time and skills well.

Declaring a valve fixed, because a symptom faded, is where inference fails quietly. The next evidence step is small, and it is usually cheap. Turn the inference into a written screening record with conditions first.

Then agree a confirming test for every valve that truly matters. Some cases end honestly without any injection treatment at all. If a critical valve stays ambiguous after proper testing, the answer is a full assessment, or replacement.

5. How to grade vendor evidence

The same grading applies to evidence that others show you. A case study, a survey report and a site demonstration all carry a grade. The vendor may not name that grade, so you must assign it yourself.

1. Who witnessed the result, and were they independent of the work?
2. What was measured before treatment, and by which stated method?
3. Under what conditions was the after result actually taken?
4. What failed, what was excluded, and what is quietly not mentioned?

Our own rule is simple, and it binds our own records first. Our field teams, working across the Middle East, Asia and Africa, leave records a third party can check. A valve condition survey should show readings and conditions, not adjectives.

For more depth, read how to read a valve maintenance case study critically. Then apply the same questions to reading a valve survey report from any provider. These habits also shape how to choose an online valve maintenance provider, because records reveal competence.