Short answer: rain ran down the corrugated siding, crossed the open gap behind a box that was never flush, and dripped through the box's own drain holes onto live terminals. A joint that had loosened or corroded then cooked in place — 16 A across 0.05 Ω makes 12.8 W of heat in a few square millimeters, while the 40 A breaker sees only normal load current. Leave the circuit off at the panel until the repair is done. The job splits in two: replace what the heat touched, then mount the new box gasketed and flush so the water stays outside.
Why did the outdoor electrical box melt?
The heat was made at one bad connection. Water that ran behind the box found a terminal that had loosened or corroded, and the joint turned into a 12.8 W heater at 16 A. That is hot enough to char the plastic around it for months without troubling any breaker upstream.
The parts failed in a set order. The mount came first: a flat box bolted to corrugated siding touches only the crests, so an open channel runs behind it from eave to grade. Rain sheets down that channel and reaches the drain holes. Then the joint: moisture on terminals grows oxide and leakage paths, contact resistance climbs, and I²R heating starts. Aluminum conductors speed this up, because they creep under thermal cycling and oxidize under water. A copper joint fails the same way when water sits on it.
The damage itself tells you which fault you had:
- Melt concentrated at one terminal: a loose or corroded joint at that point.
- Melt spread across the board with rust trails: water bridging the live parts.
- Browned, stiff insulation along the conductors: months of heating at the joint.
- A tripped breaker on top of the melt: the fault finally went low-resistance.
Here is the arithmetic, with example numbers. Take a 240 V supply, a 40 A breaker and a heat pump drawing 16 A steady; replace them with your nameplate values and the method holds. A corroded terminal presents about 0.05 Ω of contact resistance. Power at the joint: P = I²R = (16 A)² × 0.05 Ω = 12.8 W, concentrated in a contact spot a few square millimeters across. That is a soldering-iron element running inside a plastic box, around the clock. The circuit feels it as a 0.8 V drop — 239.2 V of the 240 V — so nothing downstream changes.
Two inputs move the heat quickly. Resistance: at 0.1 Ω the same 16 A makes 25.6 W. Current: heat goes with the square, so 20 A at the same joint makes 20 W, and half the current makes a quarter of the heat. This is why a box runs clean for months, then collapses within a week once corrosion takes hold.
Why didn't the breaker trip?
Because the fault drew no extra current. Sixteen amps on a 40 A breaker is 40 % of rating, and a thermal-magnetic breaker holds that load indefinitely. The curve only moves above roughly 1.25× rating on the thermal element and around 5-10× on the magnetic instant trip.
A breaker is a current device; it has no wire that senses temperature at one terminal. Typical thermal-magnetic breakers hold the rated load indefinitely, start answering somewhere above about 1.25×, and trip instantly at 5-10×. Your 12.8 W joint sat outside that whole curve while it charred the box. Even the leak drew little: a few amps of stray current through wet debris is about a tenth of the breaker's rating, while the arc energy lands in one spot.
The trip arrives only when the fault escalates. A joint that finally bridges to the other leg or to ground draws hundreds of amps and clears in a cycle or two. By then the plastic has already burned. An arc-fault breaker, where the circuit carries one, watches for the arc signature instead and gives the earlier warning on a tracking fault.
Is it dangerous to leave the box in service?
Yes — treat a melted box as a live fire risk until the circuit is off and verified dead. Carbonized plastic conducts, the failure ends as a phase-to-phase arc, and the enclosure or the siding behind it can become the path.
Before anything else: switch the breaker off at the panel and verify the terminals dead with a meter. Heat-damaged insulation can fail when the wires are moved.
The damage does not reverse itself. Plastic that has melted keeps carbonized, conductive tracks after it re-hardens. Lugs that ran hot lose clamp force. Conductor insulation that stiffened or browned has lost its rating — common THHN is 90 °C insulation, and a cover that melted ran far past that. Every part that saw the heat is replaced, not cleaned and reused.
Two conditions raise the stakes. Aluminum conductors creep loose and oxidize fast under water. Standing water left in the box restarts the tracking the day after the repair if the mount stays unchanged.
How does water get into a rain-tight box?
Type 3R — the rating most outdoor disconnects carry — is defined in NEMA 250 for outdoor use against falling rain and ice formation. It holds rain off the front, drains through weep holes by design, and never planned for water fed in from behind.
The rating did what it says. Rain hitting the front and top of a 3R box is held out, and the drain openings are deliberate: NEC 314.15 permits approved drainage openings of 3 mm (1/8 in) to 6 mm (1/4 in) in boxes and conduit bodies listed for damp or wet locations, so intruding water has an exit. The unsealed mount turned those exits into inlets. Water arriving from behind, through holes sized around 3 mm (1/8 in) to 6 mm (1/4 in), landed inches from live terminals.
Corrugated siding set the trap. A flat box on a corrugated sheet meets two contact lines per panel, with open channels between them that drain downward past the back, and every rain fed those channels. Wind-driven rain wraps upward and sideways too, so a bead of sealant across the top alone guards one side of a three-sided problem. The same entry logic — gasket, glands, field holes, condensation — applies to any outdoor box, and the water in a weatherproof junction box article walks each path. Where the rating ladder matters, the NEMA 3R vs NEMA 4 guide lays out what each type is tested for.
What should be checked before ordering parts?
Read the conductors, not the melted cover. Conductor material, insulation condition back past the heat, terminal grip, water traces and entry sealing are five checks that decide how much of the circuit gets replaced.
- Conductor material. Copper or aluminum? Aluminum strands take lugs marked AL/CU plus joint compound; a copper-only lug loosens within months.
- Insulation. Press the covering about 6 in away from the box; stiff, shiny or cracked insulation gets cut back to bright copper and re-terminated.
- Terminal grip. A lug that turns by hand or shows a bright wear ring has lost clamp; every re-termination goes to the torque stamped on the lug.
- Water traces. Rust streaks under the drain holes, mineral trails and green or white corrosion on copper confirm back-fed water and point at the mount as the root cause.
- Entry sealing. The liquid-tight whip and its sealing ring at the box wall are the other regular leak; the unsealed conduit entry repair checks list what to look at.
Order the parts after the checks, not before. Conductor length and size decide whether a cut-back re-termination is enough, and the mount hardware decides which box you buy.
What is the fix for the melted box?
Replace what the heat touched, then fix the mount: a new disconnect or enclosure, fresh terminations torqued to the lug markings, and a gasketed flush mounting that closes the channels behind the box for good.
The electrical work belongs to a licensed electrician: 240 V, possibly aluminum conductors, and the heat pump manufacturer's wiring instructions on top. A disconnect that has melted is replaced whole as the listed assembly its manufacturer built. Its enclosure is part of the device, so it goes back as one unit. Conductors with damaged insulation are cut back past the heat or run new, lugs are new, and every joint is tightened to the torque stamped on the lug.
Then close the water path. A mounting block or a flat back plate bridges the corrugations so the box compresses its gasket on a continuous surface; the siding-gap mounting block fix shows the build, and mounting a weatherproof box without losing its rating covers fasteners and gasket handling. Sealant goes on as a bead along the top and up the sides, as a second line behind the gasket. The drain holes stay open, and the box bonds through its ground screw to the equipment grounding conductor: NEC 250.96(A) requires the metal parts to be bonded together so they can carry fault current safely, and wall contact is not that bond.
Three checks close the job: the door gasket compresses evenly all the way round, a continuity meter reads near zero ohms from the enclosure to the ground bus, and a hose run over the siding above the box leaves the inside dry while water exits the drains below.
What should the installer have done the first time?
Three things at installation: mount on a flat surface or bridge the corrugation with a block, compress a gasket all round the back, and torque every termination to the value stamped on the lug.
A box set straight onto corrugated sheet pumps water behind itself for as long as it hangs there. Silicone added at the top six months later patches one side of a three-sided problem. The gasket does the sealing, the block makes the gasket possible, and the top bead is backup. Terminations torqued at install and re-checked after the first heating season handle the thermal cycling that loosens joints.
One look after the first storm season pays for itself: an even gasket line, no rust trails below the box, drain holes clear. Rust there means water has been working, and the corrosion repair sequence starts the same way this one does — stop the water first. The replacement still obeys the mounting rules: with breakers or fuses inside, the center of the highest operating handle stays within 2.0 m (6 ft 7 in) of the floor or platform under NEC 240.24(A), and the enclosure mounting guide collects the clearances.
When is a different enclosure the better answer?
Change the box when the environment, not the install, drove the failure. Salt air, brine or washdown chemistry points to 316L over 304, and a gasketed Type 4X stainless enclosure with a flat sealing back beats a bare 3R shell.
Type 3R covers falling rain and ice, and that is all it claims. Type 4X keeps that protection and adds corrosion resistance; UL 50E gives the 4X design an extra 200 h of salt spray beside an AISI 304 test specimen. For coastal air, brine or washdown chemistry, 316L is the grade. Its molybdenum raises pitting and crevice resistance where chlorides sit in gasket lines.
That is the shape our wall-mount enclosures come in: gasketed door, removable mounting panel, wall-mount feet, rated NEMA 4X / IP66 by the manufacturer; not UL listed. A 10 x 8 x 6 in (H x W x D) 304 stainless enclosure has the depth for a whip splice, terminals and a DIN strip, and the same 10 x 8 x 6 in box in 316L takes the salt-air sites. When the melted part is the disconnect itself, its listed replacement comes from the disconnect manufacturer, and a box like this houses the terminals and control gear around it. For a back that has to seal on corrugated or uneven walls, custom enclosures build the plate and mount into one order, delivered in 10 business days from your approval of the drawing.
If sizing the box is your next step, the enclosure selector filters by environment, listing and inside dimensions and hands you a shortlist or a quote request.
Enclosures in this article
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 10 × 8 × 6 in | 304 stainless | 10 x 8 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 10 × 8 × 6 in | 316L stainless | 10 x 8 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
How much heat does one loose terminal make?
Use I²R, and watch how fast it climbs. The example in the body makes 12.8 W at 16 A and 0.05 Ω. Raise the draw to 25 A on the same joint and the heat reaches 31.3 W; at 0.1 Ω it doubles again to 62.5 W. That is a soldering iron that never switches off. No breaker element watches joint temperature, so the heat runs until the joint fails short or arcs.
Can silicone sealant alone keep an outdoor box watertight?
Silicone is a gap filler, not the seal. An elastomer gasket does the sealing at about 20-30 % squeeze, the range gasket handbooks give for elastomer flat gaskets, and a gasket needs a flat, continuous back to press against. On corrugated siding a bead across the top leaves the side and lower channels open, so water keeps running to the weep holes with the mount unchanged.
Why does a Type 3R box have holes — should they be sealed shut?
They are weep and drain openings, and the design counts on them. NEC 314.15 permits approved drainage openings of 3 mm (1/8 in) to 6 mm (1/4 in) in boxes and conduit bodies listed for damp or wet locations, and any water that gets in needs the exit. Leave them open and pointing down; plugging them turns the next leak into a full box. The fix is keeping water off the back plane.
Would a GFCI have caught this fault?
Only if current leaked to ground. A US Class A GFCI trips at 4-6 mA of ground-fault current, but water bridging the two hot legs passes line-to-line current that cancels at the GFCI core, so it stays closed. The same melt can therefore happen on a GFCI-protected circuit. An AFCI breaker reads the arc signature instead and gives the earlier warning.
