A shutdown valve is partially stroked every quarter, the positioner reports a pass, and the SIL verification now shows a comfortable margin with a longer full-stroke interval. The question is how much of that margin has been earned. Partial stroke test proof test coverage describes only what the partial stroke test (PST) can reveal. It says nothing about how much the full proof test that follows will catch.
Over-crediting usually happens when those two numbers are blended into one. This article separates them and shows how each enters the PFDavg calculation.
Two coverage figures, two different jobs
A PST moves the valve through a small part of its travel. Practitioner sources describe typical travel of 10 to 20 percent, and the aim is to show that the valve is not stuck and that the actuator and solenoid respond. A recent practitioner guide on IEC 61511 proof test intervals puts PST coverage at roughly 60–80% of the valve's dangerous failure modes. The same guide describes a full proof test as reaching 95 to 99 percent of dangerous failures.
Those are two separate model inputs:
- PST coverage is the fraction of dangerous undetected failures that the PST can reveal, tested at the PST interval.
- Full proof test effectiveness is the fraction of the remaining failures that the full test reveals, tested at the full-test interval.
Neither figure is a property of the valve alone. Each depends on what the procedure actually does, which is why the ranges above are starting points, not inputs.
Partial stroke test proof test coverage versus full proof test effectiveness
The standard formulation splits the dangerous undetected failure rate. Emerson's guidance gives the form PFDavg = λDU × DC × test interval (PST) / 2 + λDU × (1−DC) × test interval (FST) / 2. The first term is the part of the failure population the PST sees. The second is the part only the full stroke test sees.
Note the assumption built into that second term. A control-valve article states that in most cases the full stroke test is assumed to provide full coverage, so the residual PFDavg is treated as eliminated. That is a modelling convenience. IEC 61511 defines a proof test as a test to detect dangerous hidden failures so that a repair can restore the system to an "as new" condition or as close as practical. The words "as close as practical" matter here.
There is also a classification point to settle in your own model. The Emerson article notes that diagnostic coverage does not include any faults detected by proof tests. An academic treatment instead describes a proof test as "partial" if it is imperfect, detecting only some failures and leaving others latent, and lists partial stroke testing as an example. Both framings can give workable numbers. State which one you use, and apply it consistently.
What a PST can and cannot see
A PST is good at the failures that develop while a valve sits still. The distinction is the same for any PST: it proves movement, not shutoff.
- Seat leakage. A valve can move 10% perfectly well while being blind to seat leakage, so any tight-shutoff requirement stays with the full test.
- Fouling at the seat. Emerson notes that if a valve is normally open and the fluid coagulates or polymerises at the seat, a PST will not detect it.
- Service. The same source says that if the fluid is flammable or toxic and tight shutoff is essential, the PST diagnostic coverage factor will be quite low.
- Instrumentation level. ISA-TR96.05.01 examples, as summarised by one practitioner, range from 15% for a visual-only PST up to higher values where actuator or external condition-monitoring instrumentation is used. How the PST is measured therefore matters as much as how often it runs.
Emerson also warns that a 70% factor is common among consultants and engineering companies, but has to be evaluated in light of the application. Treat 70% as a question to answer, not a default to inherit.
What the arithmetic does to the interval
Emerson's worked example compares two cases. With a PST diagnostic coverage of 70%, four tests a year and a three-year full test interval, the result is 0.09λd + 0.45λd = 0.54λd. A perfect annual full test gives 0.5λd, so the longer interval looks almost free.
Now relax the perfect-test assumption. A practitioner formula for imperfect proof tests adds a term in which the fraction the test misses persists for the equipment useful life. As an illustration only, assume the three-year full test reveals 95% of what the PST missed, and that the valve stays in service for 15 years. The 30% of failures the PST cannot see then contributes about 0.30 × (0.95 × 1.5 + 0.05 × 7.5) = 0.54λd. Adding the PST term of 0.09λd gives about 0.63λd. That is roughly 26% above the annual-test baseline, not the 8% the perfect-test version suggests. These figures are my own simplified arithmetic, not a vendor result. Your FMEDA, valve life and test procedure will move them.
How to avoid over-crediting
IEC 61511 requires the frequency of proof tests to be determined through PFDavg or PFH calculation in accordance with 11.9. Edition 2 also expects the calculation to account for the coverage of any periodic proof tests and the associated procedure. In practice, work through the following:
- Trace the PST coverage figure. Tie it to the valve FMEDA or an equivalent engineering assessment, as the iFluids guide recommends: a documented coverage factor traceable to FMEDA data.
- Match the figure to the service. Reduce it for tight-shutoff duty, fouling services and valves that stay open for long periods.
- Give the full proof test its own effectiveness. If the procedure cannot exercise the full closing path, seat tightness or the solenoid, record the shortfall and carry the residual into PFDavg.
- Keep the PST and the full test procedures separate. The PST must not interfere with the SIF or prevent its proper operation.
- Check the assumptions against results. Failures that persist show that assumed coverage was not achieved. The iFluids documentation guide states that a systematic failure to achieve assumed test coverage invalidates the PFD average calculation.
What reviewers look for
The 2026 iFluids guide lists a common finding: PST results credited toward the full proof test interval without a documented coverage factor. It also lists records where as-found data is missing or reduced to a bare "pass". The latter blocks any trend analysis, so you cannot show whether the PST and full test coverage you assumed is what you are seeing.
PST also has a cost. One technical review notes that frequent stroking can degrade the valve seat, especially in soft-seated valves or abrasive slurries. A PST regime that improves stiction detection while damaging shutoff has not improved the SIF.
Key takeaways
- PST coverage and full proof test effectiveness are separate inputs. Document and justify each one.
- A PST does not replace the full proof test. It only changes how much of the failure population waits for the full test, as the interval between full tests can be extended, but the full test is still needed.
- Generic figures such as 70% need justification for the specific valve, service and test method.
- Assuming a perfect full proof test flatters the longer interval. Test the sensitivity with a realistic effectiveness figure.
- Review as-found records so that credited coverage is checked against what the field shows.
Next steps
Choosing coverage figures that stand up to challenge takes both calculation and field knowledge. The FS Engineer SIS course covers how to interpret IEC 61511 requirements and the evidence behind SIL verification. The FS Technician SIS course follows a SIF through testing and restoration, where the proof test procedures are actually carried out. Dates for both are on the course calendar.
