Risk-based inspection — deciding what to inspect, and how often

A petrochemical plant contains tens of thousands of pressure-retaining items — vessels, exchangers, tanks and a very large amount of piping. Every one of them is degrading, and no site has ever had the money, the people or the shutdown hours to inspect them all equally.

So the question is not whether to prioritise. It is whether the prioritising is done deliberately, or by whoever shouts loudest during shutdown planning.

Risk-based inspection is the deliberate version.

Risk is two questions

• Likelihood of failure — what is attacking the item, how fast, how much wall is left, and how well that is actually known.
• Consequence of failure — what escapes, how much, whether it burns, what it reaches, and what it costs in production.

The combination sets the inspection plan. Neither on its own does.

Naming the mechanism is the whole job

The part that carries the technical weight is not the matrix. It is the damage mechanism review — going through each circuit and asking what is actually attacking it.

Get this wrong and everything downstream is confident and useless. You will inspect the right equipment, on a sensible interval, with a technique that cannot detect the thing that is going to fail it.

Filter Damage mechanisms on pressure equipment, and why each gets missed

12 rows

Mechanism	Where it occurs	How it is found	Why it gets missed
General thinning	Wherever a corrosive stream contacts the wall	Ultrasonic thickness readings at fixed monitoring locations	Rarely missed — but the readings must be at the same points each time to give a rate.
Localised thinning	Downstream of injection points, mixing tees and elbows	Scanning rather than spot readings, or profile radiography	Spot thickness readings walk straight past a pit or a groove.
Corrosion under insulation	Insulated equipment in the wet temperature band	Stripping insulation at targeted points, or profile radiography	Nothing whatsoever shows from outside intact cladding.
Sulphidation	Hot hydrocarbon streams carrying sulphur, crude and vacuum units	Thickness monitoring, with attention to low-silicon components	Rate varies sharply with the steel's silicon content, so identical-looking components thin at different rates.
Naphthenic acid corrosion	High acid crude at temperature, in high-velocity and turbulent areas	Thickness survey targeted at the turbulent locations	Attacks fastest where flow is disturbed, not uniformly along the line.
High temperature hydrogen attack	Steels in hydrogen service above the limits in API 941	Advanced ultrasonics — and by keeping within the Nelson curves	Damage is internal and the wall keeps its thickness. Ordinary UT finds nothing.
Wet H₂S cracking	Sour service — blistering, HIC, SOHIC and sulphide stress cracking	Wet fluorescent magnetic particle inspection, shear wave ultrasonics	It is cracking, not thinning. A thickness survey reports the wall as sound.
Amine cracking	Carbon steel in amine service, concentrated at welds not stress relieved	Wet fluorescent magnetic particle inspection at welds	Follows the weld line, so inspection away from welds finds nothing.
Caustic cracking	Carbon steel in caustic, worst where it is hot or steam traced	Surface crack detection at welds and bends	Local heating from tracing or a steam-out can put a line into the susceptible range.
Chloride stress corrosion cracking	Austenitic stainless with warm chlorides, often from wet insulation	Dye penetrant inspection, and by controlling insulation chlorides	The chloride source is usually outside the pipe, so process reviews overlook it.
Creep	Furnace tubes and hot headers held near their temperature limit	Diameter growth measurement, replication, targeted ultrasonics	Accumulates over years with no change until damage is well advanced.
Fatigue	Small-bore connections and anything on a vibrating line	Vibration survey, then surface crack detection at the root weld	Not a corrosion mechanism at all, so a corrosion-driven programme never looks.

Naming the mechanism is the whole job. Inspecting the right vessel with the wrong technique finds nothing and reports it as good — a thickness survey will not find cracking, and a visual check will not find hydrogen attack. Every row here has a technique that works and several that do not.

What the codes are for

Code	Covers	What it gives you
API 510	Pressure vessels in service	Inspection intervals, rating, and the rules for repair and alteration
API 570	Process piping in service	The same for piping, worked in terms of circuits and monitoring locations
API 653	Aboveground storage tanks	External and internal inspection, and rules for repair and reconstruction
API 580	Risk-based inspection, principles	What a credible RBI programme has to contain to be defensible
API 581	Risk-based inspection, methodology	The quantitative calculation behind likelihood and consequence
API 571	Damage mechanisms on fixed equipment	What each mechanism is, what it looks like, and how to inspect for it
API 941	Steels in hydrogen service at temperature	The Nelson curves — the operating limits that keep hydrogen attack away
API 574	Piping system components	Inspection practice and the expected thicknesses to measure against

These divide by equipment type rather than by problem, so a single damage mechanism can span three of them. Sour service cracking is an API 571 mechanism, found under an API 510 or API 570 programme, at an interval set by API 580 reasoning.

The division is by equipment type, which is worth knowing because a single mechanism crosses several codes. Sour service cracking is described in API 571, found under an API 510 or API 570 programme, at an interval justified by API 580 reasoning.

What an RBI study actually produces

Not a risk number. A plan, and for each item it states four things:

1. What the credible damage mechanisms are.
2. Where to look — the locations those mechanisms attack, which is rarely “anywhere on the line”.
3. Which technique will detect them, and what coverage is needed.
4. When, and what evidence would change that date.

Point four is the one that separates a working programme from a filed report. An interval is a prediction, and predictions get revisited.

Where it connects

RBI is the planning layer above everything in this section. It draws on the materials chosen in pipe materials and plate materials, targets the hidden damage described in insulation and coatings, and uses the examination methods from welding and NDT.

The equivalent reasoning applied to structures rather than pressure equipment is in structural inspection.

Thickness data, and the two corrosion rates

An inspection plan produces thickness readings at fixed condition monitoring locations — the same spots, measured the same way, so that successive surveys are comparable. A reading from a randomly chosen point tells you the wall thickness there. A reading from a CML tells you what is happening over time, which is the useful thing.

From those readings, two rates are calculated, and they answer different questions.

• The long-term rate uses the original or earliest thickness against the most recent. It is the settled, averaged behaviour over the life of the line.
• The short-term rate uses the previous survey against the most recent. It is what has happened lately.

Remaining life then follows directly: the metal available above the minimum required thickness, divided by the governing corrosion rate. That figure sets the next inspection interval, which is typically half the remaining life subject to a code maximum.

When measured thickness falls below what the calculation requires, the line is not automatically condemned. A fitness-for-service assessment, following API 579, evaluates whether the equipment is acceptable for continued operation in its present condition — possibly at a reduced pressure, possibly with a shortened interval, possibly with a repair. It is an engineering evaluation, and its recommendations carry the authority of whoever performed it rather than of the inspector who found the reading.

What to take away

• Risk is likelihood and consequence. Corrosion rate alone never told you where to look.
• RBI redistributes inspection effort. A study that reduces it everywhere has been misused.
• The damage mechanism review carries the technical weight, not the matrix.
• Match the technique to the mechanism. Thickness surveys are blind to every form of cracking.
• The output is a plan — what, where, which technique, when — not a score.
• An RBI study that is never revisited slowly becomes fiction.