Shipbuilding and marine fabrication
Weld monitoring for shipbuilding
High-volume welding under classification-society rules. The codes that apply, the four monitoring pain points that cost yards most, and which Therness signal addresses each.
ISO 6520-1 group 3 · Solid inclusions
Slag inclusion (ISO 6520-1 ref. 301): causes and detection
ISO 6520-1 reference 301 is trapped slag left behind by a flux-shielded process. Cause matrix, why radiography leads the detection routes, and the standards that govern acceptance.
- ISO 6520-1 reference
- 301
- Group
- 3 — Solid inclusions
- Also called
- Slag inclusion, Slag entrapment, Trapped slag
Slag inclusion is the imperfection every flux-shielded process produces if the process is not managed: SMAW, SAW and flux-cored wire all generate a slag layer that is supposed to separate from the pool and float clear before the metal solidifies. Reference 301 is what is left when some of it does not.
The cause matrix is almost entirely about giving slag time and room to escape. A pool that freezes too fast, from high travel speed or low heat input, traps whatever slag has not yet separated. A joint that is too narrow gives it nowhere to rise to. An electrode angle aimed at the wrong point in the pool drives it toward the sidewall instead of behind the arc. Missed interpass cleaning is the one non-thermal cause on the list, and the easiest to control with a hold point rather than a parameter.
That is also why radiography, not ultrasonic testing, leads detection: slag is less dense than the surrounding weld metal and takes an irregular shape, giving it reliable film contrast. In-process evidence — travel speed, current per pass, a pool camera view of slag separation — earns its place earlier, flagging a run worth stopping before the cap pass buries the result.
Reference 401, lack of fusion, shares several root causes; reference 3041, tungsten inclusion, is the other common solid inclusion but belongs to GTAW, a different process family entirely.
Typical causes
| Cause | Mechanism | Control |
|---|---|---|
| Slag not removed between passes | Chipping, grinding or wire brushing that misses a pocket leaves solidified flux residue for the next pass to bury rather than displace. | Treat interpass cleaning as a hold point and inspect the joint, not just the clock, before striking the next arc. |
| Travel speed too high for the process | In flux-shielded processes the pool needs time for slag to separate and float ahead of solidification; a fast-moving pool freezes around it instead. | Measure travel speed against the qualified range rather than judging bead appearance alone. |
| Electrode angle or weave pattern wrong for the joint | The wrong angle drives slag ahead of or beside the arc instead of behind it, trapping it against the sidewall or root. | Qualify the technique for the joint geometry and verify it visually in the cell. |
| Joint preparation too narrow or fit-up too tight | A restricted groove gives molten slag nowhere to rise to the surface before the pool closes over it. | Treat repeated slag findings in one joint design as a design or access problem, not an operator problem. |
| Current too low for the electrode or wire | A cold arc produces a sluggish, viscous pool that cannot separate slag cleanly from molten metal. | Record current per pass against the qualified range instead of trusting the panel setting. |
Detection: which route finds it
The column that matters most is the last one. An inspection family marked as not finding this imperfection will not find it however carefully it is applied, and planning around that is cheaper than discovering it during a final examination.
| Family | Method | Capability | What it sees |
|---|---|---|---|
| Visual | Visual testing to ISO 17637 | Limited | Only a slag pocket that breaks the surface is visible. Buried slag under a clean-looking cap is the normal case, not the exception. |
| Thermal | In-process infrared thermography | Supporting signal | A trapped slag pocket disturbs the local cooling pattern of the bead, which flags the run for inspection rather than confirming the inclusion. |
| Camera | High-speed weld pool imaging | Supporting signal | Shows slag failing to separate from the pool, or an electrode angle driving it into the sidewall, while the run is still being made. |
| NDT | Radiographic testing to ISO 17636 | Primary route | Slag's irregular, often elongated shape and lower density than the surrounding metal give it strong, well-established radiographic contrast. |
Where acceptance is defined
Therness pages never reproduce a standard's acceptance values: those tables are copyrighted and belong in the copy you hold. What follows is the pointer — which document governs, and which part of it to open.
| Standard | Clause topic | Note |
|---|---|---|
| ISO 5817 | Quality levels for inclusions, indexed by ISO 6520-1 reference number | The form the slag takes changes which quality-level row applies, so identify it before grading. |
| AWS D1.1 | Visual and radiographic acceptance for structural welds, with separate criteria for statically and cyclically loaded connections | Which criteria apply depends on the loading, so establish that before inspecting. |
| API 1104 | Acceptance for pipeline girth welds, assessed by radiographic interpretation | Pipeline acceptance is its own regime; do not carry a structural criterion into it. |
See the condition, not the consequence
Most weld imperfections are produced by conditions that exist while the arc is running. Therness records those conditions on every weld, so a suspect run is known on the shift that made it rather than in a batch report.
