Cracks are the one weld imperfection with no acceptance debate: ISO 5817 permits none of them, at any quality level. Porosity has percentage limits, undercut has depth formulas — cracks (ISO 6520-1 group 100) get a flat not permitted at B, C and D alike.
The engineering question is therefore not “how much is acceptable” but which cracking mechanism you’re fighting — because hot cracks, cold cracks, and lamellar tearing have different causes, different timing, and almost opposite prevention levers.
The crack families
| Family | ISO 6520-1 | When it forms | Where |
|---|---|---|---|
| Hot (solidification) cracking | 1011 (longitudinal), 104 (crater) | During solidification | Weld centreline, crater |
| Hot (liquation) cracking | 1012 | During welding | HAZ grain boundaries |
| Cold / hydrogen cracking (HICC) | 100 group | Minutes to 48 h after welding | HAZ or weld metal, often toe/root |
| Lamellar tearing | — (parent metal) | During cooling/restraint | Parent plate, step-like, below the weld |
Hot cracking: a solidification problem
As the pool solidifies, the last liquid to freeze is enriched in low-melting impurities — sulphur and phosphorus above all. If shrinkage strain pulls on a centreline still wetted by these films, it opens. Signature: a centreline crack down the bead, or a star-shaped crater crack (104) at an abrupt stop.
Drivers:
- Chemistry: high S/P parent metal (free-machining steels are notorious), high dilution pulling impurities into the weld
- Bead shape: deep, narrow beads (high depth/width ratio) concentrate the last liquid at the centreline — typical of high-current single passes
- Restraint: thick, stiff joints pulling transversely on the solidifying bead
- Craters: stopping the arc without filling the crater leaves a shrinkage pipe that cracks
Prevention: control dilution and bead shape (width ≥ depth as a rule of thumb), use crater-fill programs, select filler with appropriate Mn/S ratio, reduce restraint where design allows.
Cold cracking: the hydrogen triangle
Cold cracking (hydrogen-induced, HICC) needs three ingredients simultaneously:
- Diffusible hydrogen — from damp consumables, moisture, rust, oil, coatings
- Susceptible microstructure — martensite from fast cooling of hardenable steel
- Tensile stress — residual shrinkage stress plus structural restraint
Remove any one and cracking stops. That’s the entire logic of cold-crack prevention:
- Hydrogen: low-hydrogen consumables, baked and stored per supplier specs; clean, dry joints
- Microstructure: preheat and interpass temperature control slow the cooling rate so the HAZ transforms to tougher products instead of martensite. The controlling variable is the t8/5 cooling time — explained in heat input, cooling rate and t8/5
- Stress: welding sequence, avoiding over-restraint, post-heat (hydrogen release) on critical work
The treacherous property is delay: cold cracks form up to 48 hours after welding. A weld inspected immediately and passed can be cracked by morning — which is why codes require delayed NDT on susceptible steels. Full prevention playbook: hydrogen-induced cracking prevention.
Lamellar tearing
Step-like cracking in the parent plate under a weld that loads the plate in its through-thickness direction. Cause: flattened non-metallic inclusions in rolled plate, opened by through-thickness shrinkage strain in heavily restrained corner/tee joints. Prevention is mostly design and material: Z-grade plate (through-thickness tested), joint details that reduce through-thickness strain, buttering layers.
Detection
- Visual + MT/PT: surface-breaking cracks — magnetic particle testing on ferritic steel, penetrant elsewhere; timed after the delay window on hydrogen-susceptible work
- UT/PAUT: subsurface cracks and lamellar tearing — planar reflectors respond strongly (PAUT guide)
- In-process thermal monitoring: cracking itself is post-weld, but its precursors are thermal — cooling rate outside the qualified t8/5 window, interpass temperature violations, missed preheat. Monitoring every joint’s thermal cycle catches the conditions for cracking on the weld where they occur, instead of discovering them at delayed NDT. See real-time cooling-time analysis.
The compliance angle
Because cracks are never acceptable, audits focus on process discipline: preheat records, interpass measurements, consumable handling logs, delayed-inspection hold points. Under ISO 3834 these are exactly the records that must exist per joint — and automated thermal records make that evidence a by-product of production rather than a clipboard exercise.
Cracks are the most severe entry in the defect landscape — for the full map, see the welding defects guide with ISO 5817 acceptance criteria.
Frequently Asked Questions
What are the main types of weld cracks?
Three families dominate: hot cracks (solidification and liquation cracking, ISO 6520-1 code 1011/1012), which form at high temperature while the weld solidifies; cold cracks (hydrogen-induced cracking), which form hours to days after welding in hardened microstructure; and lamellar tearing, step-like cracking in the parent plate under through-thickness shrinkage strain. Crater cracks (104) are a hot-crack variant at weld stops.
Are any cracks acceptable under ISO 5817?
No. Cracks (ISO 6520-1 group 100) are not permitted at any ISO 5817 quality level — B, C or D. The only nuance concerns micro-cracks, whose acceptance can depend on the parent material and application. Any detected crack is a mandatory repair with root-cause analysis; on hydrogen-susceptible steels, inspection must wait long enough for delayed cracking to appear.
What is the difference between hot cracking and cold cracking?
Hot cracks form during solidification, typically along the weld centreline, caused by low-melting-point films (sulphur, phosphorus segregation) plus shrinkage strain — visible immediately. Cold cracks form after the weld is finished, from the combination of diffusible hydrogen, hard microstructure (martensite), and residual stress — they can appear up to 48 hours later, which is why codes mandate delayed inspection on susceptible steels.
How do you prevent hydrogen (cold) cracking?
Attack at least one corner of the triangle: hydrogen (low-hydrogen consumables, dry storage, clean joints), microstructure (preheat and interpass temperature control to slow cooling and avoid martensite — verified via t8/5 cooling time), and stress (joint design, sequence, avoiding excessive restraint). Preheat per EN 1011-2 or the steel supplier CET/CEV formula, and verify temperatures with measurement, not chalk guesswork.
