Assessing Probability of Failure
Two questions, two answers
Probability of failure in RBI is built from two linked judgments:
- Susceptibility and rate – can the identified damage mechanism act on this component, and how fast does it degrade the material? This comes from material, environment, and operating data, and from measured corrosion or crack growth rates versus remaining wall or remaining life.
- Likelihood of failure given the damage – if damage exists, what is the chance the component fails before it is found? This is driven mainly by inspection effectiveness, i.e., the probability that the damage is detected and acted on in time.
For thinning, a typical calculation compares the degradation rate to the remaining thickness to estimate remaining life, then expresses POF over the assessment interval. For cracking mechanisms, the mechanism's susceptibility and the effectiveness of the detection method dominate.
Failure modes
POF is assigned per damage mechanism and per failure mode (small leak, large leak, rupture). A single component can carry several mechanisms, each with its own POF; the governing one usually drives the plan.
Traps
- Reporting a corrosion rate as a probability of failure.
- Assuming that adding inspections reduces the degradation rate - it does not; it shortens the window in which damage goes undetected.
- Forgetting that generic failure frequencies must be adapted to the specific equipment and its inspection history.
- Ignoring time: POF generally rises with time, so the assessment interval and remaining life must be consistent.
Rule of thumb
Ask "how fast can this mechanism cause a failure, and would our current inspection method find it in time?" Fast mechanisms plus weak detection equal high POF regardless of nominal corrosion allowance.