Тwo acronyms sit next to each other on every reliability roadmap, and almost nobody can say why both are there. Risk-based inspection and reliability-centred maintenance are routinely described as alternatives. They are not. They answer different questions, and a plant that treats them as a choice will get the wrong answer twice.
What each one actually asks
RBI asks: where should I look, and how often? It is inspection-oriented. It ranks equipment by the product of failure likelihood and consequence, then allocates inspection effort accordingly. Its natural home is static equipment — vessels, piping, exchangers — where the dominant degradation is corrosion, erosion or cracking, and where the only meaningful intervention before failure is to look.
RCM asks: what must this asset do, and what stops it? It is function-oriented. It begins with what the asset is required to deliver in its operating context, enumerates the ways that function can be lost, and selects a maintenance task for each failure mode — or explicitly decides to run to failure. Its natural home is rotating and instrumented equipment, where interventions exist beyond inspection.
Where the confusion comes from
Both produce a ranked list. Both use criticality. Both end in a maintenance plan. To a steering committee reviewing slides, they look like competing vendors of the same product.
The difference surfaces the moment a failure mode has no inspectable precursor. A bearing entering spalling gives a vibration signature, not a wall-thickness reading. An instrument drifting out of calibration announces nothing to a thickness probe. RBI has no answer for these; it was never asked the question. Meanwhile, RCM applied to a corroding pipeline will dutifully conclude that the correct task is inspect on condition — which is where RBI begins, and where it does the harder quantitative work of setting the interval.
How they compose
- Run RCM first, at the system level. Establish functions, failure modes and consequences. This tells you which assets matter and why.
- Where the selected task is inspection, hand the asset to RBI. RBI sets the interval and the technique with more rigour than an RCM worksheet ever will.
- Where the selected task is condition monitoring, hand it to CBM. Same logic, different instrument.
- Feed inspection findings back into the RCM failure-mode data. Without this loop, both analyses drift from the plant they describe.
The plant that gets this right has one criticality assessment, not two. It has one asset register. And it has one conversation about consequence — not a mechanical-integrity version and a maintenance version that quietly disagree about which pump matters.
What AI changes
Both worksheets now generate in minutes. The failure-mode library assembles itself from the CMMS. The likelihood side of the RBI matrix improves as corrosion models ingest more inspection history.
What does not generate is the consequence side. Consequence is a statement about the business: about the delivery commitment, the regulatory exposure, the safety case, the reputational cost of a release. No model holds that. Someone has to.
And someone has to notice when the two analyses disagree — when RBI says a vessel is low risk and RCM says the function it serves is critical. That disagreement is information. It is also, reliably, the thing an automated pipeline discards.