Corrosion is one word covering two different engineering problems. Metal that thins evenly behaves like a consumable: it has a rate, the rate can be measured, and a designer can add thickness to absorb it. Metal attacked at a single point does not. It can perforate a wall that is otherwise nearly intact, on a timescale nobody budgeted for. Separating the two is usually the first substantive question in a claim, because they lead to different documents and different responsible parties.

Uniform attack is an allowance question

General corrosion is the case the design codes were built around. Where an environment removes metal broadly at a roughly steady rate, the response is a corrosion allowance — wall thickness beyond what pressure and load require — paired with an inspection interval short enough to confirm the assumption holds.

That makes uniform loss a documentary problem as much as a metallurgical one. The allowance, the assumed rate and the inspection frequency were written down somewhere. Where a wall thinned predictably and still failed, the questions are whether those assumptions were reasonable, whether the service stayed inside them, and whether anyone measured.

Localised attack has no useful average

Pitting differs in kind. A passive film breaks down at a discrete site, commonly under chloride, and the pit becomes its own micro-environment: oxygen-starved, acidified by hydrolysis of dissolved metal ions, and more aggressive as it deepens. The chemistry driving it is generated by the attack itself, so it accelerates rather than tapering off.

Initiation is also stochastic. Identical components in the same service pit at very different times, and a rate in millimetres per year says nothing useful about a mechanism whose character is that penetration concentrates where the average never looks.

Crevice corrosion is a geometry problem

Crevice attack produces similar occluded-cell chemistry but is governed by geometry rather than surface condition. A gasket face, a lap joint, a thread, a tube-to-tubesheet gap or a deposit on a tank floor all create a gap tight enough to restrict oxygen replenishment while still holding electrolyte.

So a material can perform for years on open surfaces and fail where it is bolted, sealed or fouled. Attention then shifts to joint design, gasket selection and deposit control — decisions usually made by parties other than whoever specified the alloy.

How general loss is measured

Broad thinning is quantified by thickness. Ultrasonic surveys on a defined grid, compared against nominal and against prior surveys, give remaining wall and an implied rate, and it is repeat measurement at the same locations that makes the rate meaningful.

Where a coupon or recovered component exists, mass loss over a known exposure gives an independent figure. That depends on removing corrosion product without removing sound metal, and ASTM G1 sets out the cleaning procedures for it, including the repeated-cleaning approach that separates product removal from attack on the base metal.

How pitting is measured and reported

Pits are described by density, size and depth. ASTM G46 provides the standard guide and rating charts for doing that consistently, along with the pitting factor — deepest penetration divided by average penetration — which is the number expressing how badly an average understates the worst case.

Depth is measured by microscope focus, by depth gauge, or by sectioning through the pit and examining it metallographically, which also shows whether attack followed grain boundaries or a weld-affected region. Undercut and subsurface pits present far less at the surface than they have taken from the wall.

Susceptibility testing on the alloy itself

Where the question is whether the material suited the service, standardised tests characterise the alloy rather than the part. ASTM G48 uses ferric chloride solutions to rank the pitting and crevice resistance of stainless steels and related alloys, with distinct methods for each.

For austenitic stainless steels the adjacent question is sensitisation — chromium carbide precipitation at grain boundaries after welding or slow cooling, leaving neighbouring metal depleted and open to intergranular attack. ASTM A262 sets out the practices for detecting that susceptibility, from oxalic acid etch screening through the acid immersion tests.

Why an accelerated exposure number rarely settles anything

Salt spray testing to ASTM B117 appears in specifications constantly and is widely over-read. It is a comparative, quality-control test run in a continuously wet, constant-condition chamber, and the standard is explicit that its results are not a prediction of service life.

Real exposure is cyclic — wetting and drying, temperature swings, varying chloride loading — and that cycling governs both the rate of attack and how protective the products formed turn out to be. Cabinet hours can show one coating outperformed another under one artificial condition, not how long either lasts on a structure.

Where the average overstates the margin

Fitness-for-service assessment exists because of this distinction. Inspection codes such as API 570 for piping, and the assessment procedures in API 579-1/ASME FFS-1, treat general and local metal loss under separate rules, because a component with generous average remaining wall can still be one pit from a through-wall leak.

A remaining-life calculation built on averaged readings therefore answers a question nobody asked. What matters is the deepest verified penetration and whether the survey was dense enough to have found it.

What the report needs to separate

A useful corrosion report distinguishes measurement from inference: which readings were taken where, how the specimen was cleaned, how pits were sized, and what any implied rate rests on. It should also name the mechanisms it excluded and why.

That matters most at the boundaries. Stress-corrosion cracking can occur at or beneath a corroded surface; it is a separate mechanism with separate evidence, handled on its own terms rather than folded into a wall-loss discussion.

This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.