Why Arc Stud Welds Look Solid but Fail Shear and Bend Tests Months Later

Why Arc Stud Welds Look Solid but Fail Shear and Bend Tests Months Later (What One Fabrication Shop’s Trial Actually Measured)

I’ve spent twelve years on fabrication floors watching studs that look perfect on the day of installation come back later with the weld collar cracked or the stud sheared clean at the base. The usual explanation from the crew is “the machine is fine, the settings were right.” Most of the time the settings were close enough, but one or two small things on the plate or in the ground path were enough to leave incomplete fusion that only shows up under real load.

Last year I spent four days at a structural steel shop that builds bridge and deck components. They were running a drawn-arc machine on 19 mm diameter shear connector studs into 12–16 mm mild-steel plate. Settings were textbook: 1 150–1 250 A, 0.8–1.0 s weld time, 3 mm lift, standard ceramic ferrules, gun held square. Visual collars looked full and bright. On the first day of bend testing (30° per the usual structural procedure) roughly one in six studs broke in the weld zone with almost no deformation. Tensile pull tests on the failed pieces averaged 38–42 kN; the same size studs that passed averaged 62–68 kN. That is well below the expected capacity for a properly fused low-carbon stud of that diameter.

The sharp problem was not the power source drifting or the studs themselves being bad. It was incomplete fusion driven by two everyday shop conditions that the settings table never mentions: residual mill scale plus light oil on the plate, and magnetic arc blow from the way the ground clamps were placed. When we cleaned a section of plate to bright metal and re-welded with identical parameters, the bend-test pass rate jumped and average pull strength rose to 64 kN. On the uncleaned plate the fracture surface showed a clear ring of lack of fusion around part of the perimeter; the cleaned plate showed full fusion and a ductile break above the collar. The difference was measurable in under an hour of testing.

Magnetic arc blow made the same numbers worse. The shop had been clamping the ground cables at one end of the long beams. On studs near the opposite end the collar was consistently thinner on one side. We moved the grounds to a more balanced position (two clamps, roughly opposite sides of the work) and the one-sided collars disappeared. The same studs that had been failing the bend test then held through 30° with no cracking. Arc blow does not always throw a dramatic splash; sometimes it just pulls the arc enough that one side of the stud never fully melts into the plate. The collar can still look mostly complete until you section it or load it.

Field conditions that keep producing this exact failure pattern:

• Plate arrives with mill scale or light protective oil and the crew only grinds a small spot instead of a full clean circle the diameter of the ferrule plus a few millimetres.
• Ground clamps sit on painted or scaled surfaces, or all the return current is forced to travel the long way around a large assembly.
• Ferrules are stored open in a humid shop and pick up moisture; the collar then shows porosity even when current and time are correct.
• Lift height drifts because the gun shaft is dry or slightly bent; a 1 mm short lift is enough to produce a cold-looking weld that still has a collar.
• Operators rely on the collar appearance alone and skip the start-of-shift bend test once the line is running hard.

None of this requires a different machine. On that job we kept the same power source, the same stud batch, and the same nominal settings. We simply required a clean metal circle under every ferrule, moved the grounds, and checked lift height with a gauge at the start of each shift. Rejection rate on the bend test dropped from about 15–18 % to under 3 %. The shear and tensile numbers on the sectioned samples moved into the range expected for a full-fusion weld.

If you are already seeing studs that pass visual inspection but fail the bend or shear test, look at the plate surface under the ferrule first, then at ground placement, then at actual lift height and ferrule dryness. Those three checks have explained the majority of the “mystery” failures I have opened up over the last twelve years. The rest is usually just a cold plunge or a ferrule that is the wrong size for the stud. The collar can lie; the fracture surface and a simple bend test do not.