Risk-based inspection fails more often in implementation than in concept. The concept is simple: spend inspection effort where risk is, not where the calendar says. The implementation is a sequence, and skipping a step early shows up as an indefensible interval later. Here is the sequence as it actually runs, and the record each step must leave.
Step 1: Define scope and gather data
Decide which equipment is in the study and assemble what is known about it: design data, materials of construction, process conditions, thickness history, inspection reports, repair history. This step takes longer than any other, and it should. Every downstream number inherits the quality of this data. Document what was missing as carefully as what was found, because gaps force conservatism and the auditor will want to see that they did.
Step 2: Identify damage mechanisms
For each equipment item or corrosion loop, a multi-discipline team works out which damage mechanisms are credible: which forms of thinning, cracking, embrittlement, or mechanical damage the service can actually produce. API 571 is the reference library here. The output is a documented mechanism list per item, with the susceptibility reasoning, not just the conclusion.
Step 3: Assess probability of failure
Probability follows from two things: how aggressive the expected damage is, and how much the inspection history actually tells you about the equipment's present state. An item with a high thinning rate and excellent recent inspection data can score better than a mild service with no data at all, because uncertainty is itself a driver of probability.
Step 4: Assess consequence of failure
Consequence is evaluated independently of probability: what is released, how much, what it does to people, environment, and production. This is where process engineering and safety input matter most. Keep the consequence basis documented per item, because consequence rarely changes and probability changes constantly; separating them keeps reassessment cheap.
Step 5: Rank risk and set targets
Combine probability and consequence into a risk ranking, place items against the organization's risk criteria, and identify which items exceed target. The ranking method, matrix or quantitative, matters less than its consistency. The record must show the same method applied the same way across the study.
Step 6: Build the inspection plan
For each item above target, the plan specifies what inspection activity, with what technique and coverage, by what date, will reduce uncertainty enough to bring risk below target. This is the step where RBI earns its keep: effective inspection of the right locations replaces frequent inspection of everything. It is also where record quality feeds back, because an inspection that is not well documented may not earn credit as effective.
Step 7: Reassess and evergreen
RBI is not a project, it is a loop. Findings that contradict the model, process changes, incidents, and the simple passage of time all trigger reassessment. A study that was excellent five years ago and never touched since is, for audit purposes, expired. Build the triggers into procedure and keep the reassessment log with the study.
The thread through all seven
Every step produces a decision, and every decision needs to be reconstructable: inputs, method, people, date. Programs rarely get challenged on their risk matrix. They get challenged on whether a specific interval for a specific exchanger can be walked back to evidence. If the walk fails, the interval reverts to the time-based default and the RBI investment is written off.
intelliSPEC holds the inspection history and effectiveness evidence that steps 3, 6, and 7 keep coming back to, assembled as the work happens rather than at study time. Bring your current RBI data pack to a demo and see the difference.
