Melted ground wire in a junction box: causes and fixes

Short answer: that ground wire melted because it was carrying current the breakers never see — welding return current first, a damp high-resistance fault second. Replace the damaged 14 AWG run end to end, clamp the new wire to prove it carries nothing, and give welders a dedicated return clamp on the workpiece. Copper conducts about 41 times better than 304 or 316 stainless. A short copper ground wire can beat a huge welded steel frame as a return path and pay for it in heat. The sections below rank the causes, put numbers on the melting, and end with the checks that keep it from repeating.

Why did the ground wire melt when no breaker tripped?

Because the heat never passed through the breaker. Welding return current takes the ground path without crossing the overcurrent device at all. A damp fault self-limits to a few amps that a 10 A breaker holds indefinitely rather than tripping. Both causes leave the panel history spotless and the wire as the only evidence.

A 10 A IEC miniature breaker with a C curve trips magnetically at five to ten times its rating, so the magnetic range is 50 to 100 A under IEC 60898. The thermal element sits at 11.3 A (1.13 times rated) for at least an hour from cold. It must trip within an hour only once the current reaches 14.5 A. A fault under those levels never leaves a record at the panel.

Welding current takes a different road altogether. The arc draws from its own supply, finds its return at the work clamp, and flows home through whatever metal lies in between — frames, catwalks, conduit, and the ground wires inside them. The breaker protecting the melted wire sits nowhere on that loop, so it never trips and never records anything. The wire's real job, as the grounding and bonding guide sets out, is to carry a short circuit home fast enough to trip its breaker. Arc-current hitchhiking is not part of that job.

What melts a ground wire, most likely first?

Rank the causes this way: welding return current first wherever welded stainless frames and catwalks surround the conduit, a damp high-resistance fault second in cold humid rooms, a neutral landed on ground third, and a surge or lightning strike a distant fourth. In a cold, damp, welded-steel building the first two are the working suspects.

1. Welding return current. The arc takes the lowest-resistance loop home, and a direct conduit run beats a long walk around the steel frame. It leaves melted grounds in several boxes along one route, with the phase conductors untouched.
2. Damp high-resistance fault. Condensation or a wet, dusty film tracks across a terminal; the current sits below the trip curve while the heat lands on one spot. It points to one box, green crust, and water lines — the same local heating that drives a burned connector in a sealed enclosure.
3. Neutral on the ground terminal. Load current then flows on the ground wire around the clock. The wire carries amps with every load running and no welding anywhere on site.
4. Surge or lightning. A big pulse melts a conductor in a blink and usually leaves other casualties — dead electronics, a tripped main — that date the event to a storm.

The checks in the next sections separate them in under an hour.

How much welding current does it take to melt a 14 AWG wire?

Far less than you would expect. In the worked example below, 40 percent of a 150 A arc — 60 A — flows through 25 m of 14 AWG copper and dumps about 750 W of heat into it. That is thirteen times the heat the wire dissipates at its 15 A limit.

Start with the materials. Resistivity tables put 304/316 stainless at 1.45 × 10⁶ S/m against copper's 5.96 × 10⁷ S/m, so stainless carries about 2.4% of copper's current per equal area. A 14 AWG copper wire (2.08 mm²) therefore does the work of roughly 85 mm² of stainless, and its resistance is about 8.3 mΩ/m. Steel wins only through area: a 300 mm² frame member measures about 2.3 mΩ/m, better than the wire per metre, though not enough to ignore a copper wire running in parallel.

Now split the current between the two parallel loops. The example is hypothetical — replace the lengths, sizes and arc current with your own and the method still runs.

At its 15 A protection limit the same wire dissipates roughly 2.3 W/m, so the arc puts about thirteen times the rated heat into it. Insulation rated 90 °C sheds in minutes at that level, the copper softens and necks down, and the molten PVC welds itself to the box cover exactly as this kind of damage shows. Copper itself only melts at 1083 °C, but a buried run gets nowhere near that — the insulation is long gone first.

The frame shrugs off its share because of current density. The wire carries 60 A through 2.08 mm² (about 29 A/mm²) while the frame carries the other 90 A through 300 mm² (0.3 A/mm²), roughly a hundred times lower. That also answers the classic question: the steel does take the majority, but the majority of 150 A is still 90 A, and the wire's minority is still 60 A.

Can moisture melt a wire without tripping anything?

Yes, in two ways. Eight amps of leakage through a corroded or wet terminal puts 32 W into one point — 8² × 0.5 Ω — while a 10 A breaker holds 0.8 times its rating forever. Intermittent arcing pulses under the thermal curve char the insulation burst by burst.

The trip curve leaves a wide dark zone. An 8 A leak spread along 25 m of ground wire adds up to just 13 W (8² × 0.21), warm but harmless. The same 8 A across half an ohm of corrosion at one terminal is a 32 W soldering iron that runs all summer. The damage lands where the resistance is — one crusty terminal in one box, with the panel quiet.

The moisture that starts it comes from air, not leaks. Cellars run cool and humid, and cold air travelling through flexible conduit chills the box and the wire. When humid air meets a surface below its dew point, water films out. Air at 25 °C and 80% relative humidity has a dew point near 21 °C, so a box fed by cold cellar air sweats on exactly the warm days when doors stand open. Repeated wet-dry cycles grow the green crust behind corroded connections inside a junction box, and eventually a carbonized track that conducts. The condensation guide walks through the dew-point arithmetic.

With the circuit dead and loads disconnected, an insulation-resistance meter says how wet it is: dry 120 V control wiring reads well above 1 MΩ to ground, while a damp or carbonized path reads in the kilohms. A kilohm reading means moisture is finding ground. The run needs replacing rather than drying and refilling.

How do you find which current it was?

Clamp a DC-capable clamp meter around the new ground wire and read it three times: with everything idle, with all loads running, and while any welding on site strikes an arc — the reading that shows amps is your cause.

• Idle reading above zero. Something feeds the wire right now. Throw breakers one by one with the clamp in place; the reading that collapses names the circuit with the fault or the mixed neutral.
• Amps only while welding. Stray return confirmed. Walk over and look at the welder's work clamp — sitting on a distant column or a catwalk is your answer.
• Insulation resistance in the kilohms. A wet or carbonized path to ground exists; dry the box, replace the run and reseal the entries.
• Damage pattern. Melted grounds in several boxes along one conduit route point to return current. One box with green crust and water lines points to a damp fault. Untouched phases beside melted grounds point to welding, because a genuine fault usually burns the phase conductor too.
• Neutral audit. Every white lands on the neutral bar and every green or bare on the ground bar. Beyond the service disconnect the two stay separate, as the neutral-ground bonding section of the grounding guide explains.

One safety rule before any of it: lock the circuit out and prove absence of voltage before opening a box, and keep it out until the melted run is replaced, because that run is the circuit's effective ground-fault current path under NEC 250.4(A)(5).

What do you replace, and what do you leave alone?

Replace the whole heated run of ground wire between boxes, not just the melted end. Heat that melted insulation also annealed the copper and damaged it beyond what you can see, and a patched-in tail leaves the fault path weak.

Pull new 14 AWG green-insulated copper and land it on the ground bar. NEC 250.119 reserves green, or green with one or more yellow stripes, for insulated equipment grounding conductors 6 AWG and smaller. NEC Table 250.122 sizes the conductor from the breaker ahead — a 15 A device takes 14 AWG copper. Give every ground its own bar terminal rather than chaining through device screws; the box-by-box habit is in the grounding guide.

Leave the welded stainless frame alone as a fault path. NEC 250.96(A) requires metal parts serving as grounding conductors to be bonded so they can safely conduct the fault current likely to be imposed on them, and welded joints do that superbly. The frame misbehaves only when a welder asks it to carry arc current home — a work-lead habit, not a bonding defect — so the cure lives at the welder, not in the steel.

How do you keep it from happening again?

Three habits stop a repeat: a welding return clamp on the workpiece within reach of the arc, conduit entries sealed against cellar air, and a gasketed stainless box in place of every corroded PVC one — delivered in 10 business days from your approved drawing.

Put the welding return on the workpiece itself, as close to the arc as the cable reaches, and run dedicated welding cable sized for the arc current. A stray-current interrupter in the work lead adds a tripwire: it opens the circuit when return current tries to leave through building steel. Post the rule where welders plug in, because one clamp on the wrong column can cook a control run across the building in an afternoon.

Seal the conduit where cold air travels: duct seal at the box entries, wet-location connectors on every termination, and a drip loop so water runs down the outside of the conduit rather than into the box. Inside a gasketed box, a 15-30 W anti-condensation heater — common panel practice in damp cellars — holds the interior a few kelvin above the dew point and breaks the wet-dry cycles that grow corrosion.

Then replace the corroded PVC boxes with gasketed 304 stainless wall-mount enclosures; the stainless size finder filters every line, size and grade. In a winery without salt or brine, 304 is the grade; 316L earns its keep where brine, salt air or chlorides wash the walls. Our boxes come rated NEMA 4X / IP66 by the manufacturer, not UL listed, in outside sizes from 6 x 4 x 4 in to 60 x 36 x 16 in (H x W x D), built from one piece with your cutouts and delivered in 10 business days. For a rating the inspector needs by type, the NEMA enclosure types guide maps the types to the conditions. Keep every entry worthy of the rating: an IP66-rated cable gland preserves it, while a field-drilled drain or a bargain connector drops the whole assembly to that part's rating. Where you need a size or cutout off the standard grid, custom enclosures take your width, height, depth and every opening from one piece.

The decision, by cause: welders on site — move the return and re-check with the clamp. A wet box — dry it, seal it and swap PVC for gasketed stainless. Load current on the ground — find the mixed neutral and land it right. In all three cases the melted run gets replaced, not spliced, before the circuit runs again.

Catalog dimensions to compare

Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.

Questions