Risk-based inspection (RBI) explained to the people who have to apply it

Risk-based inspection means deciding where, when and how to inspect according to the risk carried by each item of equipment, rather than applying a uniform interval inherited from custom. The approach has been formalised for a long time in the oil and chemical world; it is far less common in food and beverage processing and in pharmaceutical plants, even though the same equipment raises exactly the same questions there.

Where the approach comes from and what it is for

The reference methodological framework is published by the American Petroleum Institute: API 580 describes the RBI approach and its principles, and API 581 offers a quantitative version of it. These documents were written for refining and petrochemicals, but the logic applies to any population of equipment subject to progressive degradation: pressure vessels, process piping, heat exchangers, storage tanks.

What the approach replaces is a very widespread practice: inspecting everything at the same frequency. That practice has one merit, simplicity, and two serious flaws. It spends effort on equipment that does not need it, and it leaves under-monitored the items whose degradation is accelerating. RBI does not ask you to spend more: it asks you to spend in the right place.

One point needs to be clear from the outset. RBI does not release you from any regulatory obligation. The checks imposed by in-service inspection regulations and by the site's quality system remain a floor. What the approach organises is the inspection effort that falls to your own judgement. On most sites, that effort is the larger part of the total workload, which is precisely why it deserves to be steered rather than left to habit.

The two factors, and how they are established

The consequence of failure

This is the more stable factor. It does not depend on the condition of the equipment but on what would happen if the contents escaped, if the structure gave way, if the line stopped.

Four dimensions deserve to be assessed separately:

• The safety of people: proximity of workstations, nature of the fluid, operating pressure and temperature.
• The environment: the volume that could be released, the containment capacity available, the sensitivity of the surroundings.
• Production: the length of any outage, the effect downstream, the loss of product in mid-process.
• Compliance: the equipment's role in the audit file, and the regulatory deadlines attached to it.

The rule that prevents the worst mistakes: you keep the highest of the four consequences, never an average. Averaging quietly discounts the one scenario that would actually hurt you. The practical construction of this assessment is set out in building a criticality matrix.

The probability of failure

This is where RBI parts company with a plain criticality analysis, and this is where most attempts stall.

Probability is not estimated by feel. It is built from three elements: the damage mechanism at work, the rate at which it acts, and the remaining margin before the acceptable limit is reached.

Identifying the mechanism is a prerequisite that is often skipped. General corrosion, localised pitting, under-deposit attack, fatigue, erosion: these phenomena do not evolve in the same way and are not detected with the same techniques. Inspecting in the right place assumes you know what you are looking for. Choose the wrong technique for the mechanism and you can pass an inspection while the real defect grows untouched.

The rate is established from successive readings taken at the same condition monitoring locations. This is the most valuable data and the most frequently lost, because it requires you to bring together inspection campaigns separated by several years. See tracking equipment degradation.

What you need in hand before you start

What the approach produces

The output of an RBI analysis is not a ranking: it is a differentiated inspection plan that answers three questions for each item of equipment.

Table scrolls horizontally on small screens.

That last line is the most important and the most often glossed over. An item without history is not a low-probability item: it is an item whose probability is unknown. Treating the two the same way is the mistake that costs the most in these programmes.

The mistakes that make an RBI unusable

• Rating probability without data.With no mechanism identified and no rate measured, the rating reflects the experience of the people in the room and the memory of the last incident. The approach then loses its only advantage over ordinary criticality.
• Displaying risk as a single number.The product of the two factors hides opposite situations: the two axes have to stay legible separately.
• Treating the absence of data as a low probability,the most dangerous confusion in the whole exercise.
• Forgetting the damage mechanism.Running ultrasonic thickness readings over a zone that is liable to crack detects nothing and produces a documented false assurance.
• Making RBI a one-off study.A plan keyed to frozen probabilities goes stale at the pace of the inspections that never feed back into it.
• Believing RBI lightens obligations.It organises what falls to your judgement; the regulatory floor is not up for negotiation.

Transposing the approach to food and pharmaceutical plants

RBI comes from industries where the consequences are counted in explosions and major pollution. Transposed to a food and beverage or pharmaceutical plant, the consequence axis changes in nature without losing any importance.

Product safety becomes a consequence of the first order: a loss of tightness, a lubricant migration, a foreign body arising from mechanical degradation all carry consequences that far exceed the cost of the repair. This dimension appears in no standard RBI grid: it has to be added explicitly, and it has to be weighted against the safety and environmental axes rather than bolted on as an afterthought.

Compliance with the quality system also weighs more heavily here than elsewhere. An item whose inspection history is incomplete becomes a weak point at audit time, regardless of its actual condition. The consequence is then not technical but documentary, and it is no less real for that.

Finally, seasonality places a hard constraint on the available slots. On a site whose activity peaks at a fixed period of the year, the intervention window is narrow and known long in advance. An inspection plan that ignores this calendar produces recommendations that cannot be applied. That is why anticipation matters more here than the fineness of the risk calculation: a slightly rougher assessment delivered in time beats a precise one that lands after the window has closed.

The condition that decides everything: keeping it up to date

An RBI analysis is worth exactly what its probability axis is worth. And that axis changes at every inspection campaign: new readings sharpen the rate, a finding shifts the judgement, a repair resets a counter to zero.

If that update is manual, it will not happen. Not through negligence, but because it means reopening dozens of reports, pulling out the values, matching them against the previous campaigns and recomputing. That is work nobody funds twice.

This is exactly the point that automated reading of the reports changes. Take the documents in as they arrive, extract findings, measurements and deadlines, and recompute the trends per condition monitoring location: the approach stops being a study and becomes a permanent state. The non-negotiable condition stays the same: every value on display must be traceable back to the document and the page it came from, otherwise nothing holds up in an audit.

For the maintenance side of the same logic, see risk-based maintenance and our Maintenance Intelligence page. On the equipment families concerned first and foremost: pressure equipment and piping.