API RP 583 — Corrosion Under Insulation and Fireproofing

Scope: what API RP 583 covers, and what it deliberately does not

API RP 583, Corrosion Under Insulation and Fireproofing, was first published in 2014 and revised in 2022. It is a recommended practice — guidance, not a code — and it addresses the external surfaces of carbon steel, low-alloy steel, austenitic stainless steel and duplex stainless steel equipment and piping covered by thermal insulation, personnel-protection insulation, cold-service insulation, or cementitious and intumescent fireproofing. It deals with the mechanisms, the materials and systems that resist them, the design details that invite or prevent water ingress, and the inspection techniques available to find damage without stripping everything.

What it does not do is where most of the misuse starts. API 583 sets no inspection intervals, no minimum thickness and no acceptance criteria for the metal loss it helps you find. Intervals and required actions come from API 510 for pressure vessels, API 570 for piping and API 653 for storage tanks. Remaining-life and fitness-for-service decisions come from API 579-1/ASME FFS-1. The practice also does not address internal corrosion, buried piping, or corrosion beneath internal refractory and linings, and it treats soil-to-air interfaces only where they interact with an insulation termination.

Read it as the technical basis document. It tells you where CUI will occur, why it occurs there, what to specify to prevent it, and how to look for it economically. The decisions about when to look and what to do about the result live in the inspection codes, and a CUI programme that cites API 583 as its authority for either of those has a hole in it that any competent auditor will find on the first pass.

The susceptible temperature bands, and the four ways they get misapplied

The numbers matter, so start with them. For carbon and low-alloy steel the susceptibility range in the practice is approximately -12 °C to 175 °C (10 °F to 350 °F), and the highest corrosion rates fall in the middle of that band, where water under the jacket persists as a liquid film rather than flashing off. For austenitic stainless steels the concern is not thinning but external chloride stress corrosion cracking, with susceptibility beginning around 50 °C (120 °F) and extending to a similar upper bound. Duplex grades are more resistant but the practice does not treat them as exempt.

Misapplication one is screening a circuit list by design temperature. Design temperature is not operating temperature, and CUI responds to the metal temperature at the surface beneath the jacket. Misapplication two is excluding equipment that normally operates above the band. Lines above 175 °C still cool at supports, at dead legs, at valve bodies, at the end of runs, and every single time the unit comes down. Misapplication three is excluding equipment below the band, when cold and cryogenic service passes through it on every defrost and every warm-up. Misapplication four is excluding intermittent service entirely, when intermittent service is the worst case there is.

A circuit list built from operating temperature bands with no cycling or downtime overlay is the finding that recurs most often on CUI programmes. The correction is cheap and entirely documentary: an explicit column for intermittent, cyclic, steam-traced, out-of-service and cold-box equipment, and a recorded reason for every exclusion. An exclusion that cannot be defended in one sentence is not an exclusion, it is an omission.

Two different damage mechanisms under the same jacket

CUI of carbon steel is aqueous corrosion accelerated by chlorides and sulphates leached from the insulation and drawn in from the environment. It appears as general thinning, as localised pitting near jacket laps, penetrations and fasteners, and as severe wastage where water collects — the six o'clock position on horizontal runs, the underside of vessel heads, the top of ring supports, the base of skirts and the flats of pipe shoes. The distribution is not random and a programme that samples randomly wastes most of its budget.

CUI chloride stress corrosion cracking of austenitic stainless steel is not thinning at all. It is branched transgranular cracking with negligible wall loss, and every thickness-based screening technique in common use is blind to it. The chlorides can come from the insulation itself, from marine air, from cooling-tower drift or from fire water testing. The controls are correspondingly different: low-chloride insulation qualified to ASTM C795 with the C692 stress corrosion test and the C871 chemical analysis, aluminium foil wrap over the steel, or a qualified coating barrier. Specifying a coating alone and calling the stainless circuits protected is a common and expensive misreading.

The practical consequence is that one CUI programme must contain two inspection strategies. Pulsed eddy current and profile radiography address wall loss on carbon steel; they say nothing at all about cracking. Stainless circuits require insulation removal and surface examination — penetrant, or eddy current on the bare surface — at the locations the susceptibility ranking nominates. Building both strategies into one coherent written procedure, with a defensible technique justification for each, is standard work for ASNT Level III consulting.

Coatings, insulation and design details: the requirements that actually bite

The mitigation hierarchy runs design first, then coating, then insulation selection and weatherproofing, and inspection last. Design details matter more than anything else on the list: eliminate water traps, slope horizontal runs so water leaves rather than sits, terminate insulation with a bevelled and sealed end instead of a butt joint against bare steel, lap the jacket so the overlap faces down, avoid through-jacket fasteners at the twelve o'clock position, and use proper support boots rather than metal-to-metal contact at shoes and clamps.

Coating specification is where audits find the weakest paperwork. The companion document is AMPP/NACE SP0198, and qualification testing of coatings for CUI service is addressed by ISO 19277. The recurring error is specifying an atmospheric-exposure system for a surface that will spend its service life hot, wet and oxygen-starved beneath insulation. Immersion-grade epoxy phenolics, thin-film aluminium-pigmented systems and thermally sprayed aluminium are the usual answers depending on temperature. A standard epoxy-polyurethane read across from the structural steel specification is not, and it will fail long before the insulation is next opened.

For austenitic stainless, insulation chemistry is a purchase requirement rather than a preference. ASTM C795 compliance, verified by certificate per lot at delivery, and enforced on the installation contractor, who will otherwise fit whatever the local supplier has in stock. That is a document control problem as much as a corrosion one, and the certificates belong in the same system that holds the rest of the mechanical integrity records rather than in a warehouse file that no one opens again.

Corrosion under fireproofing — the half of the practice that gets skipped

The F in the title is not decorative. Cementitious fireproofing on vessel skirts, structural columns, pipe rack legs and vessel support lugs is a water trap with steel reinforcement mesh embedded in it. Cracking, spalling, loss of bond and corrosion of the mesh let water in and then hold it against steel that in many plants has not been looked at since construction.

Corrosion under fireproofing is systematically under-inspected for two structural reasons. The steel is usually load-bearing rather than pressure-retaining, so it falls outside the circuit lists that API 510 and API 570 programmes are built from and no one owns it. And removing fireproofing is disruptive, expensive and requires reinstatement by a specialist, so the work is deferred until there is a reason, and by then the reason is a visible failure. Skirt-to-head junctions and column base plates are the classic locations, and they fail at the worst possible moment.

The guidance in the practice is proportionate: survey fireproofing condition visually on a defined cycle, treat cracking, staining, rust bleed and drummy areas as evidence rather than as cosmetic defects, and remove selectively at the worst locations rather than attempting a blanket strip. The single most valuable action is administrative — add the fireproofed structural steel to the inspection register explicitly, with its own locations and its own frequency. Steel that is not on a list does not get inspected.

Screening before stripping: inspection technique selection

The techniques the practice covers fall into three groups. Non-intrusive screening through insulation and jacket: profile and tangential radiography, pulsed eddy current, guided wave ultrasonics, digital and real-time radiography, neutron backscatter for locating wet insulation, and infrared thermography for moisture ingress. Inspection through small removable plugs or inspection windows. And full insulation removal, followed by visual examination, ultrasonic thickness, and penetrant or magnetic particle examination on the bare surface.

Every screening method has a stated limitation that the practice is explicit about, and those limitations are exactly what disappears between the field and the integrity model. Pulsed eddy current averages over a footprint and understates isolated pitting. Profile radiography images a projected wall and is blind to metal loss out of the projection plane. Guided wave detects cross-sectional loss above a threshold and needs a bare band for the collar, with sensitivity falling off sharply past supports and bends. Presenting a pulsed eddy current survey as a thickness survey is a defensible technique used indefensibly, and an independent review of inspection reports against the technique sheet is where that gets caught before the number becomes a corrosion rate.

The economic argument in the practice is one of sequencing: screen broadly, strip narrowly, and concentrate the removal budget where the screening results and the susceptibility ranking agree with each other. Recording results against a physical location — a specific support, a specific elbow, a specific skirt quadrant — rather than against a line number is what makes the next campaign cheaper than the last one. Holding those locations on the asset geometry in a digital twin removes the ambiguity that written location descriptions never quite manage to resolve when someone tries to re-find a point five years later.

How API 583 sits alongside API 510, 570, 653, 580 and 581

API 570 requires that CUI be considered for insulated piping systems in susceptible service, and it gives guidance on the extent of insulation removal when damage is found or suspected. API 510 and API 653 carry equivalent expectations for pressure vessels and storage tanks respectively. In every case the code establishes the requirement and the interval, while API 583 supplies the technical detail that makes the resulting inspection intelligent rather than random. The two are complements, not alternatives.

On risk-based programmes, API 581 provides a CUI damage factor driven by susceptibility, coating quality and age, insulation type, complexity and the effectiveness grade assigned to previous inspections. That effectiveness grade is the single most influential input a reviewer can get wrong. A pulsed eddy current screening covering five percent of a circuit is not a highly effective inspection, and grading it as one produces an interval extension the underlying data does not support. When the grade is challenged during an audit, the technique justification is the only defence available.

Atlantis supports owner-operators as a technical resource on this work: writing the CUI procedures and technique justifications, qualifying the personnel who perform the examinations, validating contractor reports against the procedures they claim to follow, and structuring the resulting data so it survives past the campaign. The interval decisions, the risk assessment ownership and the code compliance sign-off remain with the owner's authorised inspector and inspection engineer, where they belong.

The misreadings that generate audit findings

The list is short and it repeats across sites. A CUI programme with no written susceptibility ranking, so exclusions cannot be defended when questioned. Susceptibility bands applied to design temperature instead of operating temperature. Stainless circuits screened with thickness-based methods that physically cannot detect the cracking mechanism they face. Fireproofing absent from every inspection register on the site. Coating specified by reading across from the structural steel standard. Insulation chloride certificates never collected at installation. Inspection effectiveness over-graded in the risk model. Findings recorded against line numbers rather than physical locations, so no one can re-find them at the next campaign.

A ninth is subtler and more expensive than any of them: repairing the steel and then reinstating the same insulation system, the same jacket detail and the same drain path that caused the damage in the first place. CUI repairs that do not change the water path recur on exactly the schedule they ran on the first time, and the second repair is always in a worse location than the first.

The correction to all of it is a written, auditable programme — susceptibility ranking with documented exclusions, a technique justification for each method deployed, personnel qualification records, and a data structure that stores results by location rather than by line. Where a CUI programme has grown out of an inherited spreadsheet and a contractor's report folder, moving the records into an inspection data management system with location-anchored history is usually the first thing worth fixing, because every other improvement depends on being able to compare this campaign's results with the last one.

Request a consultation