An arc struck where it shouldn't be — on the base metal outside the joint. No weld metal is deposited, but the shallow, fast-quenched crater is a hard, brittle spot that can crack and eat into fatigue life. Small mark, real consequences.
An arc strike is an arc accidentally established between the electrode or torch and the base metal outside the weld joint — on the plate, a clamp, a stiffener, anywhere the current found a path. The arc melts a shallow crater, and the crater is quenched by the cold surrounding metal almost instantly.
No weld metal is deposited, so it is easy to dismiss as cosmetic. But in carbon and low-alloy steel the quench leaves a hard, untempered martensitic spot that can crack under restraint and act as a fatigue initiation site; on stainless it destroys corrosion resistance through chromium-carbide sensitization. That is why most codes prohibit arc strikes outright and require them to be ground out.
Stray arc, arc gouge, accidental arc strike, arc burn.
Arc strikes are usually found by accident — a QC walk-through, a paint removal, a lucky angle of light. Here's what you're looking for.
Arc strikes are a contact and grounding problem. They happen when the circuit closes somewhere it shouldn't — through a habit, or through a bad ground that forces the current to find its own path.
An accidental arc melts a small pool of base metal outside the joint. The pool is tiny and the surrounding metal is cold and massive, so the heat is sucked away almost instantly — the cooling rate far exceeds anything a normal weld sees. In carbon and low-alloy steel, that quench produces untempered martensite: hard, brittle, and full of locked-in stress.
That hard spot is a crack waiting for an excuse. Under restraint, or with hydrogen from moisture in the atmosphere, it can crack within hours; under cyclic load it acts as a fatigue initiation site even if it never cracks — the notch concentrates stress. On stainless steel the mechanism is different but equally damaging: the local heating sits in the sensitizing range, chromium carbides precipitate at grain boundaries, and the corrosion resistance of the area is gone.
Two habits and one ground clamp stop nearly every arc strike on the floor.
Check the ground: clean contact point, tight clamp, adequate cable. A good ground means the current goes where you want it — and only there.
Strike the arc in the joint or on a run-off tab — never on the workpiece. If you need to test an electrode or set the machine, use a scratch plate.
Never rest the torch or electrode on the work. Walk the joints and clamps for strikes — and for grind marks that hide them — before the part moves on.
If you're getting "accidental" strikes you can't explain, fix the grounding before you blame your hands. A failed work clamp is the most common invisible cause — the current finds its own path, and the path is the strike.
Arc strikes are small, which makes them easy to fix — and easy to fix wrong. The rule: remove the metal, don't bury it.
| Step | Action | Why |
|---|---|---|
| 1 | Locate every strike — walk the whole work, look for heat tint and polished-over spots. | Hidden strikes fail inspection later; find them now. |
| 2 | Grind out the crater completely with the right disc, feathering the edges smooth. | The hard martensite lives in the crater; surface blending alone leaves it. |
| 3 | Verify removal with MT on ferritic steel, PT on non-ferritic. | Micro-cracks radiate from the strike and must be found. |
| 4 | If the excavation is deep, repair with a qualified weld after QC approval, per the code. | A deep cavity needs fill, not just grinding. |
| 5 | Re-inspect the repaired area. | The repair is a weld like any other — it gets inspected like one. |
Welding over an arc strike buries the hard spot instead of removing it — the crack risk and fatigue site are still there, now hidden from inspection. Grind first, verify, then weld only if required.
Every process touches the arc differently. Here's where each one strikes.
No strikes on the plate — trigger on scrap or a run-off tab. Beware nozzle contact with the work while the trigger is live; it lights an arc in an instant.
Use high-frequency start — the arc ignites without touching the tungsten to the work, which eliminates contact strikes by design. Scratch-starting TIG is making strikes on purpose.
Scratch- and tap-start technique — strike in the joint or on a run-off tab. Never test electrodes on the workpiece; that's the most common intentional strike on the floor.
Program start and stop outside the joint — on run-off tabs. Check the work-return cable path for accidental touch points; a robot crash or a bad ground creates strikes in seconds, over and over.
The dials are mostly innocent here — arc strikes are a grounding, technique, and habit problem. Check these before anything else.
| Parameter | Check | Typical Fix |
|---|---|---|
| Grounding / work clamp | Clamp condition, cable size, connection point | Clean, tighten, and clamp to the work — not to a painted fixture |
| Start technique | Where the arc is struck | Strike in the joint or on a run-off tab — never on the workpiece |
| Start habits | Scratch plates and run-off tabs in place | Keep scrap handy; replace the tab when it's worn |
| Open-circuit voltage | Machine setting where adjustable | Lower OCV reduces the chance of accidental ignition on contact |
No calculator applies here — arc strikes are a grounding and technique problem, not a numbers problem. The Welding Parameters guide covers start/stop practice and ground connections in depth.
Arc strikes live in the ground circuit — walk it before you look anywhere else.
Power sources matter for two things: the ground circuit and open-circuit voltage. Check the work clamp, the cable, and the contact point — a bad ground is the classic hidden cause. PPE matters because grinding out strikes on hard plate throws sparks and debris; wear the right eye and face protection.
Arc strikes are found with the eye — and confirmed with surface NDT after grinding.
Acceptance depends on the code and project: AWS D1.1, ISO 5817, API 1104 and project specifications each set their own limits for this discontinuity. Check the governing document before judging a weld acceptable — there is no universal pass/fail rule.
The questions welders actually ask about arc strikes — answered straight.
Yes, by most codes. Even though no weld metal is deposited, the quenched crater is a discontinuity — a hard spot that can crack and a fatigue initiation site — and codes like AWS D1.1 require it to be removed.
On code work, yes. The strike leaves hardened, often cracked metal that surface blending alone won't remove. Grind to sound metal, feather the edges, and verify with MT on ferritic steel or PT on non-ferritic.
No. Welding over a strike buries the hard spot and its cracks under weld metal — it's still there, now hidden from inspection. Grind out first; weld only if the excavation depth requires fill.
The tiny pool quenches almost instantly against cold metal, forming hard, brittle untempered martensite in carbon and low-alloy steel. It is stress-locked and hydrogen-sensitive, so restraint and moisture can crack it — and it acts as a fatigue starter under cyclic load.
Check the ground first. A dirty, loose, or undersized work clamp forces the current to find another path through the work — and the current always finds one. Fix the ground, then look at technique and start habits.