October 21, 2026 · Indumor
People assume a good-looking weld is a sound weld. It is the other way around: the defects that matter most are the ones you cannot see. A smooth bead with evenly stacked ripples can hide lack of fusion from end to end; an ugly one covered in spatter can be perfectly sound inside. Looks say something about the welder, but they are not enough to accept a joint.
These are the questions we get asked most about weld defects, answered from the shop floor.
No. A discontinuity is any interruption in the structure of a weld: a pore, an inclusion, an undercut. It becomes a defect when it exceeds what the governing code accepts — AWS D1.1 for structural steel, ASME for pressure vessels and piping, API 1104 for pipelines. That is why the same pore can be acceptable in a structure and a reason for rejection in a pressure vessel.
Five show up again and again: porosity, lack of fusion, incomplete penetration, undercut and cracks. Add slag inclusions in flux-based processes, and distortion, which is a defect of the part rather than the joint, but accounts for its fair share of complaints.
Because gas gets trapped in the solidifying metal. The usual causes are moisture, oil, paint or rust in the joint; badly adjusted shielding gas; drafts that blow the gas away, typical of MIG welding outdoors; and electrodes that have picked up moisture. Low-hydrogen electrodes such as E7018 are kept in ovens for exactly that reason. Porosity often shows on the surface, but it can also be buried inside.
It is when the filler metal does not fuse with the base metal or with the previous pass: it sits there, stuck on, but not joined. It comes from low heat input, too high a travel speed or the wrong torch angle, and it is a well-known risk of short-circuit MIG on thick sections. It is a concern because it is planar, like a crack, and almost never visible from outside. That is why AWS D1.1 does not treat short-circuit MIG (GMAW-S) as prequalified: to use it on structures, the procedure has to be qualified by testing.
The weld does not reach the root of the joint and leaves part of the back side unfused. It comes from a tight root opening, an insufficient bevel, too little current or an electrode too large for the bottom of the groove. In a complete-joint-penetration weld, which has to carry the full load of the member, it is an automatic reject.
A groove left in the base metal along the toe of the weld, where the arc melted it and the filler did not fill it back. Too much current or voltage, too high a travel speed or a poor angle cause it. It reduces the section and concentrates stress right at the toe of the weld, which is why codes limit how much is acceptable by thickness and, above all, by whether the structure sees fatigue loading, like a crane runway girder or a vibrating screen.
No. Structural welding codes do not accept cracks; AWS D1.1, for one, rejects them regardless of size. There are hot cracks, which form as the metal solidifies, and cold cracks, caused by hydrogen, which can appear hours after welding. That is why, for some high-strength steels, final inspection is done no sooner than 48 hours after the weld is finished. Preheat, low-hydrogen consumables and controlled cooling are the prevention.
A crack is not fixed by covering it with another bead: it stays underneath and keeps growing. It is removed completely, checked to make sure nothing is left, and only then rewelded.
Pieces of slag trapped between passes in processes that use flux: stick, flux-cored, submerged arc. They are avoided by cleaning every pass with a chipping hammer and wire brush before the next one, and by keeping a bead profile that leaves no pockets where slag can hide.
First, by eye. Visual inspection is the cheapest method, and it stops many problems before they ever reach a test, as long as someone who knows what to look for does it on a clean weld. Then come non-destructive tests: dye penetrant and magnetic particle for what reaches the surface; ultrasonic and radiographic testing for what lies inside. For decades, film radiography was the standard way to see inside a weld; today phased-array ultrasonics does much of that work, without clearing the area for radiation and with results on the spot. We explain how they are used in NDT: radiography and dye penetrant.
It is removed completely by grinding or air-arc gouging, checked with a test to confirm nothing is left, rewelded with the same procedure and inspected again. Some specifications limit how many times the same spot can be repaired, because every repair reheats the material.
Written, qualified welding procedures, welders qualified to those procedures, visual inspection by a trained inspector and the percentage of testing the project specification calls for. All of it on paper: it is part of the quality dossier. If your project calls for it, at Indumor we build it from the first cut, along with the welding itself.
A good-looking weld is not the same as a sound weld. The difference lies in the procedure, the welder and whoever inspects it. If you have a job with quality requirements, send us the specification and we will tell you how we would approach it.
Send us the specification and we will tell you how we would approach it, with the procedures and testing it calls for.
© Indumor 2010–2026 — Industrial steel fabrication, Córdoba, Argentina
