Failure mechanism
Weld failure
A weld rarely fails because the steel was too weak. It fails because the joint that was made is not the joint that was analysed: the throat is thinner, the effective length is shorter, the fusion is intermittent, or there is a crack in it that opened the night it cooled. Capacity is a product of geometry, so a millimetre of leg lost, or a bead that never reached the root, can remove a large part of the strength of the connection while leaving a weld that looks finished.
How this failure works
A fillet weld carries its load across the throat, so its capacity is throat thickness multiplied by effective length and by the design strength of the weld. The throat is the leg size multiplied by the cosine of the angle, which for a nominal 45 degree fillet is roughly seven tenths of the leg — the leg is the dimension that can be measured on a finished weld, and the throat is the one that carries the load, which is why a weld that looks adequate from the toe can be undersized. Effective length is not the full run either: where the weld is not wrapped or returned at the ends, the code deducts an allowance at each end for the crater, and a weld made intermittent where the drawing showed it continuous reduces length by a different arithmetic again.
The failure modes are distinct and most of them are invisible from the outside. Lack of fusion at the root or at the side wall happens when the arc does not melt into the parent metal, leaving a sharp notch that looks like a weld from the toe. Lack of penetration in a joint detailed as complete joint penetration leaves an unfused ligament exactly where the design assumed full thickness. Toe cracking follows the fusion line where the section changes, and is where fatigue cracks begin. Lamellar tearing opens the parent plate in the through-thickness direction behind a heavily restrained tee. None of these is found by measuring leg size, and all of them are found late.
Hydrogen induced cold cracking needs three conditions at the same time: a hard microstructure, hydrogen in the weld, and restraint. The hard zone forms in the heat affected zone when the weld cools quickly — a thick section, a high carbon equivalent steel, a low heat input, a cold ambient. Hydrogen arrives with moisture: unbaked or damp electrodes, damp flux, rust, oil, or condensation on the plate. Restraint comes from a joint that cannot contract as it cools and from the surrounding structure holding it. Because all three are needed, breaking any one of them works, and the practical set is preheat, dry low-hydrogen consumables issued through a controlled oven, heat input inside the qualified range, and on thick or highly restrained work a post-heat or a delay before inspection or loading.
A weld is a qualified operation rather than a skill applied generously by eye. The weld procedure states the process, the consumable, the current range, the pass sequence, the preheat and the interpass temperature, and it is supported by a procedure qualification record made on the steel and thickness range actually in use. The welder is qualified separately for the process and the position, which matters because a welder whose work is sound in the flat position will not necessarily produce the same joint overhead or vertical. Distortion is the other output of the process: weld metal shrinks as it cools, an unbalanced sequence pulls a member out of plumb and out of square, and a member that has to be forced into position carries an inbuilt force that no connection design accounts for.
Illustration
Conceptual illustration: a section through a fillet weld showing the leg, the effective throat and the fusion faces, with lack of fusion at the root drawn as a void.
Typical causes
- A fillet smaller in leg than the drawing called for, or a continuous weld made intermittent, with no check before the joint was loaded.
- Lack of fusion at the root or the side wall from low current, fast travel, a poor electrode angle, or restricted access to reach the root.
- A joint detailed as full penetration made without back gouging, a backing bar or access to the second side, leaving an unfused ligament.
- Hydrogen cracking where thick or high carbon equivalent steel was welded with damp consumables, no preheat and high restraint.
- Fit-up gaps wider than the procedure allows, bridged with weld metal so the visible bead is not the joint the design analysed.
- The wrong consumable, or one taken from an oven that was not working, with no control over its condition at the moment of use.
- Distortion from an unbalanced welding sequence, leaving members out of square that are then forced into position on site.
Warning signs
- Undercut along the toe, a ropey convex bead, or a leg visibly smaller than the same detail elsewhere on the member.
- A thin dark line along the toe or across the bead face, particularly where paint has flaked off it.
- Rust bleeding from an open root at a tee joint or from a lap where the weld did not reach the faying surface.
- Heavy slag, spatter and arc strikes left on the parent plate; an arc strike is a hardened notch in the steel.
- A plate visibly bowed along a weld line, a bracket out of square, or a member that will not fit without pulling up with a spanner.
- Paint applied over uncleaned weld, or slag inclusions visible once the bead was chipped.
Inspection methods
- 01Measure leg size, throat and length with a weld gauge, and check them against the weld symbol rather than against the neighbouring weld.
- 02Look for undercut, crater, spatter, arc strikes and open root, and confirm that continuous welds are continuous and that end returns exist.
- 03Check fit-up and distortion of the fabricated item before it leaves for site, since both are difficult to correct afterwards.
- 04Require dye penetrant examination of toes and root passes wherever the design calls for it, by a qualified testing agency.
- 05Require ultrasonic examination of complete joint penetration welds in tension members, to the extent the design specifies.
- 06Review the weld procedure, the procedure qualification record and the welder qualification, and check consumable storage and baking records against the procedure.
Preventive measures
- Show every weld on the drawing with a symbol stating type, size, length, and whether penetration is complete.
- Qualify the procedure on the steel and thickness range in use, and follow it at the work face rather than treating it as paperwork.
- Control preheat and interpass temperature on thick sections and on steel with a higher carbon equivalent.
- Keep low-hydrogen consumables in a controlled oven, issue them against a record, and discard what has been out too long.
- Provide fit-up tolerance and welder access in the design, so the root can be reached instead of a gap being bridged.
- Plan a welding sequence that balances shrinkage, and check plumb and square after welding rather than before.
- Weld on clean, dry surfaces free of rust, oil and moisture, and never over paint or a damp plate.
Relevant standards
The codes and specifications a design is normally judged against. Naming them is not a claim that AKEN certifies, inspects or approves against them — the design authority for a project does that.
- IS 816 — code of practice for welded construction, including the weld types and their application.
- IS 800 — general steel construction: weld design, throat thickness and effective length.
- IS 2062 — structural steel, including the carbon equivalent that governs preheat and hydrogen control.
- IS 4000 — welded and high strength bolted joints, including workmanship and qualification.
- Eurocode 3 (EN 1993-1-8) — the European basis for weld design and weld detailing.
Related AKEN services
Scope note
This article explains a failure mechanism for engineering education. It is general guidance, not a design check, an inspection report or a certification for any particular structure. Assessments of an actual structure must be carried out by the responsible design authority against the project’s own specifications.