Weld defects and inspection
The ISO 6520 catalogue of imperfections: causes, repair, and the test methods that find them.
Every weld has imperfections. What matters is which kind and how large, because ISO 5817 treats them very differently: pores have a permitted percentage, cracks have none at all.
The section is arranged by defect rather than by process, because the cure follows the defect. Porosity is a shielding and cleanliness problem. Undercut is current, arc length and angle. Lack of fusion is heat and where the arc is pointed. Cracks split into three separate mechanisms that share only the name — and telling them apart starts with when the crack appeared, not what it looks like.
Two things are worth knowing before reading any of it. First, appearance and acceptance are different questions: a rippled bead within the limits passes, and a neat bead with a crack under it does not. Second, some of the worst defects are invisible from the surface, which is why the choice between radiography and ultrasonics is not a formality.
- Undercut Why the toe of the weld gets eaten away, what ISO 5817 allows at each quality level, and which change in technique fixes it.
- Lack of fusion and penetration The difference between the two defects, why they are the most dangerous ones, how inspection finds them and how to avoid them.
- Weld cracks How hot and cold cracks differ, why hydrogen and restraint cause them, and what preheat and post-weld heat treatment prevent.
- Weld defect catalogue The main groups of weld imperfections with causes and remedies: cracks, cavities, inclusions, lack of fusion and shape defects.
- Porosity in welds Why a weld comes out porous: gas flow, draught, damp electrodes, rust and paint — and how to tell the causes apart on the bench.
Frequently asked questions
Which defects are never acceptable?
Cracks and lack of fusion, at every quality level of ISO 5817. Everything else has a limit that depends on the level and the thickness; these two do not.
The weld looks bad. Does that mean it is defective?
Not by itself. Acceptance is measured against the quality level given on the drawing. An uneven bead inside the limits passes, and a tidy-looking bead with lack of fusion underneath does not — which is the whole reason non-destructive testing exists.
Why do some defects only appear a day later?
That is hydrogen-induced cold cracking. Hydrogen needs time to diffuse to the point of highest stress, so the crack can take hours and up to two days to form. It is why critical joints are inspected after a delay rather than straight off the bench.