10 mm Fillet, 180 A and a Cold Root
A subcontractor is welding 10 mm thick S275 gusset plates to a 150 × 90 channel with 15 mm leg-length fillet welds, one pass per side, using manual GMAW with 0.8 mm solid wire, 80% Ar/20% CO2 at 18 L/min, 180 A, 20 V, 6 m/min wire feed and 250 mm/min travel. The first three brackets passed a visual inspection, but after grinding a toe and running a dye-penetrant check the inspector found a straight line of lack-of-sidewall fusion along one toe and undercut measured at 0.8 mm. The welder says the set is 'maxed out' and he has welded this joint for years with the same settings. Yard temperature is 3 °C and the plates have condensation on them. The specification requires a 6 mm minimum throat, full fusion to the root, 100% visual examination and spot ultrasonic testing.
1. Show by calculation whether the stated single-pass procedure can deliver a 15 mm leg and full root fusion, and explain the observed lack of sidewall fusion and undercut.
Read the worked answer
The joint cannot be made in one pass with this procedure.
- Leg length 15 mm gives a throat of 15 × cos 45° ≈ 10.6 mm and a fillet cross-section of ½ × 15 × 15 = 112.5 mm².
- Metal required per metre = 112.5 mm² × 1000 mm = 112,500 mm³; at 7.85 g/cm³ that is about 0.88 kg/m.
- At 250 mm/min the travel speed is 15 m/h, so the deposition rate needed is about 13 kg/h.
- A 0.8 mm wire at 180 A realistically deposits 1.0–1.5 kg/h. Even at 2 kg/h the procedure can lay down roughly one-seventh of the metal the drawing calls for.
The consequence is that the welder is not making a 15 mm leg fillet at all: he is running a small bead tucked into the corner, so the specified throat does not exist. A fillet weld gauge would have shown this immediately - the visual inspection that 'passed' the parts had no dimensional content.
Lack of sidewall fusion. A 0.8 mm wire at 180 A is at very high current density and runs in short-circuit transfer at 20 V. That combination gives a stiff, shallow-penetrating arc, unstable wire feed and a fast travel to avoid burn-through, so the arc runs on the pool rather than on the sidewall. The 10 mm channel is a heavy heat sink and the plates are at 3 °C with condensation on them, so the weld freezes before it has wet into the parent metal. The result is cold lap along the toe, which opens up into a straight-line indication under dye penetrant.
Undercut. 0.8 mm measured against a typical acceptance limit for this class of work (about 0.5 mm) fails. It comes from excess current density and travel speed, with insufficient filler volume to fill the gouge the arc cuts at the upper toe.
2. Write the revised procedure variables you would specify and the qualification documents required before production restarts.
Read the worked answer
Joint preparation. Clean both faces to bright metal for 25 mm each side; remove mill scale, rust and any primer. Fit-up gap 0–1 mm, no tack welds in the root where fusion is critical. Dry the plates and warm them to at least 20 °C to remove condensation - no code preheat is needed for 10 mm S275, but moisture is a defect source.
Consumable and gas. 1.2 mm solid ER70S-6 (G3Si1), which is the standard step up for this thickness; shielding 80% Ar/20% CO2 at 15–18 L/min with a gas lens and a flow check at the torch.
Parameters. 260–290 A, 26–30 V, travel 300–400 mm/min, giving heat input in the region of 1.0–1.5 kJ/mm. Three stringer runs per side rather than one pass, with interpass temperature limited to 250 °C. Work angle 45°, travel angle 5–15° push, slight weave limited to three wire diameters, and the arc held on the joint root for the first run so the sidewalls are melted, not just wetted.
Sequence. Balance the welding - alternate sides, backstep the runs - to control angular distortion of the channel.
Documentation before restart:
- A welding procedure specification for this joint, supported by a procedure qualification record to ISO 15614-1 (or ASME IX/AWS D1.1 where the contract names it).
- Welder qualification to ISO 9606-1 for fillet welds in this position and thickness.
- Welding coordination to ISO 14731, with the WPS issued to the welder and the machine settings locked so the next operator cannot simply wind the set back to 180 A.
- A documented first-off inspection before the rest of the batch runs.
3. Define the examination you would require on the next three brackets and explain why the original visual inspection was not adequate.
Read the worked answer
Examination on the next three brackets:
- Dimensional. Fillet weld gauge on both legs at four positions per weld, recording leg and throat - this is the check that was missing and it would have exposed the procedure failure on the first part.
- Visual to ISO 5817. After cleaning, with undercut measured on a profile gauge and compared with the acceptance class; 0.8 mm fails for this class.
- Surface examination. Dye penetrant or magnetic particle testing on the ground toes to detect surface-breaking lack of fusion; dye penetrant is suitable for the toe but will not find the root.
- Volumetric spot check. Angle-beam ultrasonic testing from the plate, since lack of sidewall fusion is a planar defect and radiography is poorly oriented to detect it - the beam may run parallel to the defect plane and show nothing.
- Destructive proof. Cut one production bracket through a weld, polish and macro-etch it, and measure throat and root fusion. This is the only direct proof that the new procedure does what the WPS says.
Only when all five are satisfactory should the remaining brackets be released, with periodic sampling thereafter.
Why visual inspection failed. Lack of sidewall fusion is a planar, internal defect. It has no surface opening - it is a cold lap inside the joint - and it sits under the weld cap, hidden from the eye. The toe only yielded an indication after grinding, which is not part of a normal visual examination. Visual inspection can find undercut, spatter, profile and size defects, but it cannot certify fusion, which is why the specification called for spot ultrasonic testing in addition and why the weld size should have been gauged. The deeper lesson is that 'it looks the same as always' is not evidence of a sound procedure.