Risk-based inspection for CUI: how the ranking is built

Risk-based inspection, RBI, is a way of deciding what to inspect, how, and when, by estimating risk instead of counting years. It has a framework document, API Recommended Practice 580, and a methodology document, API Recommended Practice 581, that turns the framework into numbers. Corrosion under insulation is one of the damage mechanisms API RP 581 models explicitly, which makes it a good case for understanding what an RBI ranking is actually made of.

What risk-based inspection is

Under time-based inspection, every item gets inspected on a fixed interval, at most the maximum the code allows. Under RBI, each item gets a risk value, the product of the probability of failure and the consequence of failure, and the inspection plan concentrates effort where risk is highest and relaxes it where risk is low. API RP 580 sets out what a credible RBI program must contain: a defined scope, data collection, damage mechanism identification, probability and consequence assessment, risk ranking, an inspection plan, mitigation, and reassessment. API RP 581 provides one accepted way to calculate the numbers, and codes such as API 570 and API 510 allow RBI-derived intervals to replace the default ones when a program meeting API RP 580 is in place.

The probability side: the CUI damage factor

API RP 581 expresses the likelihood of each damage mechanism as a damage factor that scales a generic failure frequency. The CUI damage factor for insulated carbon and low-alloy steel is built from a small set of attributes, all of which a plant already records somewhere:

• Operating temperature, which sets a base corrosion rate. The rate is highest in the warm, wet band and falls off at both ends.
• Driver: the external environment. Marine and cooling-tower plume locations get higher rates than an arid inland site.
• Insulation type, through an adjustment for how well the material holds water.
• Complexity: the number of branches, penetrations and supports per length, since each is a water entry point.
• Insulation condition, from the visual inspection.
• Coating: whether a coating exists under the insulation and how old it is.
• Time in service since the last effective inspection, because expected wall loss accumulates.
• Inspection effectiveness: how much confidence past inspections give that the loss is actually known.

The rate, the time and the inspection history combine into an estimate of how much of the corrosion allowance has probably been used and how uncertain that estimate is. The damage factor rises steeply as the estimated loss approaches the allowance. For austenitic stainless steel, a separate factor treats external chloride stress corrosion cracking under insulation, driven by temperature, chloride exposure and time.

Inspection effectiveness

API RP 581 grades each past inspection from A (highly effective) to E (ineffective) for the mechanism in question. For CUI, removing insulation and measuring at the right locations is rated far more effective than a visual walk-down, and a through-insulation screen sits in between depending on coverage. The grade and the number of inspections at that grade reduce the uncertainty term in the damage factor. This is the mechanism by which a good inspection earns a longer interval, and a poor one does not.

The consequence side

Consequence in API RP 581 covers area affected, injury potential, environmental cost and business interruption, driven by the fluid, its phase and inventory, the release size, detection and isolation, and the equipment's role in the process. For CUI ranking within one unit, a simpler view is often enough: a high-temperature hydrocarbon line and a cooling-water return may have the same CUI likelihood, but the first one ranks higher because a leak is worse. The workbench treats this as an optional weighting, so likelihood and risk can be viewed separately.

What the ranking produces

The output of an RBI assessment is a risk-ranked list and an inspection plan per item: which methods, what coverage, and by when, chosen so that the inspection brings risk back under a target before the next assessment. For CUI that plan typically says things like "PEC screen the supports and low points on circuit 14 before the turnaround, open windows at any reading below 80 %, and re-seal."

Prioritizing without a full RBI program

A full API RP 581 implementation is a serious project: data quality work, fluid modelling, software, and a process to keep it current. Many plants do not have one, and for CUI in particular the likelihood side can be approximated well enough to order the circuits without computing consequences to the dollar. The semi-quantitative approach:

1. List the insulated circuits and their attributes: material, operating temperature and cycling, insulation and jacketing type, coating age, environment, supports and penetrations, last inspection date and method, and previous findings.
2. Score each attribute on a simple scale drawn from API RP 583 susceptibility guidance and the API RP 581 factor structure.
3. Weight and sum the scores into a likelihood reading, and optionally weight by a coarse consequence class.
4. Backtest. Hide the findings from the last campaign, rank the circuits blind, and count how many known findings sit near the top of the list compared with the order the plant would have used anyway.
5. Recalibrate the weights on the plant's own findings, then use the ranking to scope the next campaign.

Step four is the one most spreadsheets skip, and it is the only evidence that the ranking is better than time-since-last-inspection. It is also cheap: the data already exists.

Pitfalls

• Garbage in. The ranking is only as good as the temperature and coating-age fields. Where a field is unknown, treat it as unfavourable rather than average.
• Ranking the inspected. Circuits with the most history look best understood and drift down the list; circuits nobody has opened look fine because there is no bad news. Inspection effectiveness is how RBI counters this, so record the method, not just the date.
• Static weights. Published guidance is a starting point. A plant's findings should move the weights, and the backtest is how to know whether they should.
• Confusing screening with measurement. An RBI plan can call for a screen; only a measurement can update the remaining-life estimate. See CUI inspection methods.

References

• API RP 580, Risk-Based Inspection: the framework and minimum program elements.
• API RP 581, Risk-Based Inspection Methodology: damage factors including CUI and external chloride SCC, inspection effectiveness tables, consequence models.
• API RP 583, Corrosion Under Insulation and Fireproofing: susceptibility guidance used to score the attributes. Our summary.
• API 570 and API 510 for how RBI-derived intervals are permitted.