How to detect stress corrosion cracking

Start with the physics, because it explains everything that follows. A stress corrosion crack is a planar flaw: it has length and depth but almost no width. A crack 100 mm long and 5 mm deep might be a few microns wide. The volume of metal it has removed is negligible.

Now consider what a thickness measurement does. It times an echo from the far wall and reports how much steel is between. A crack removes no measurable thickness, so the reading comes back at full wall. A vessel can be extensively cracked and every UT reading on it will be nominal.

Everything below follows from that. SCC has to be found by a method that responds to a discontinuity, not to a loss of metal.

Which methods actually find it

Wet fluorescent magnetic particle — the standard for ferritic steel

Magnetise the steel, apply fluorescent magnetic particles, look under UV. Particles gather at the leakage field around a crack and the indication glows. On carbon and low-alloy steel this is the most sensitive practical method for surface-breaking cracking, and it is what the standards call for after any weld repair in wet H₂S service.

Two limits. It works only on ferromagnetic material — useless on austenitic stainless, nickel alloys and copper alloys. And it needs bare, clean metal, which means blasting off coating and scale. On a large vessel the surface preparation is most of the cost and most of the schedule.

Liquid penetrant — the answer for stainless

Chloride SCC in austenitic stainless is one of the most common cracking mechanisms in the industry, and MT cannot touch it. Penetrant is the standard alternative: dye drawn into the crack by capillary action, then pulled back out onto a developer.

It only sees what is open at the surface and clean. Grinding a surface before inspection smears metal across the crack mouth and can close it completely — a genuinely common way cracking is missed, and the reason the sequence of surface preparation matters.

Eddy current array — fast, and it tolerates coating

An array probe covers a swathe rather than a line and works on stainless where MT cannot. It tolerates thin coating, which saves the blasting bill on large areas. Depth sizing is limited, so it is a finding tool rather than a sizing tool.

ACFM — through coating, and it sizes

Alternating current field measurement induces a uniform current in the surface and measures the disturbance around a crack, giving both length and depth. It works through coating, underwater and at elevated temperature. Less sensitive than a properly prepared WFMT inspection, but on a coated structure where blasting is impractical it finds cracks nothing else would.

Phased array ultrasonics — for depth, and for cracks that are not at the surface

Surface methods find cracks. They do not tell you how deep they go, and depth is what decides whether a component can stay in service. PAUT images the cross-section and gives through-wall height — which is the number API 579 Part 9 requires.

It is also the method for cracking that starts on the far side. Sulphide stress cracking initiating at the process surface of a vessel is invisible to any external surface method until it breaks through.

Acoustic emission — is it still growing?

AE listens for the stress waves released as a crack actually extends during a controlled pressure change. It answers a question nothing else does: is this damage active?

That question matters because API 579 treats a dormant flaw and a growing flaw completely differently. A dormant crack may be acceptable as it stands; a growing one needs a crack growth analysis and a re-inspection interval. AE only hears what moves during the test, so silence is not proof of a sound vessel.

Matching method to mechanism

Look where it will be, not everywhere

SCC is not random. It needs three things at once: a susceptible material, a specific environment, and tensile stress. The third one is what tells you where to point the inspection.

Residual stress from welding is usually the largest tensile stress in a fabricated component, and it is highest at welds that were not post-weld heat treated. That is why almost every environmental cracking mechanism in the table above concentrates at welds, and why the inspection plan is a list of welds rather than an area.

When you find one

Every crack-like flaw goes to API 579 Part 9, whatever produced it — chloride, fatigue, or a welder. The assessment needs length, through-wall height, position, the material's toughness and the applied stress. Surface methods give you the first; ultrasonics gives you the second.

And Part 9 asks one further question that inspection alone cannot answer: is the mechanism still active? If it is, the flaw will grow, and accepting it as it stands is not valid. That is a process question about whether the environment that caused it is still present — and it is usually the hardest part of the assessment.