Guide
Six ways a cable joint fails
Six mechanisms, the mark each leaves on a joint cut open, and the part in the box that was meant to stop it.
Reading a joint after the event
A joint fails in the ground, long after the jointer went home. What comes back to the depot is a metre of cable on a bench. Then an argument about whether to change the kit, the crew or the specification.
Two kinds of statement follow, and they are not the same weight. The figures — kit codes, part counts, test voltages, test conditions — are quoted from the manufacturer's Heat Shrinkable Cable Accessories catalogue and the product sheets behind it. The descriptions of what a fault looks like once the joint is opened are the trade's account of it. Nobody publishes those.
The catalogue is unusually blunt about what counts as failure. After the electrical sequence the joint is opened and examined. It passes only if the filling has not cracked, no moisture has crossed the primary seal, and nothing has corroded or tracked. Read that backwards and four of the six mechanisms below are already named by the row that hunts them.
They are set out here in the order a joint is built, not in order of frequency. Nothing in the record ranks them, and the box further down says so plainly.
Six mechanisms, and the part that answers each
1 · Water through the outer seal
The seal is not one part. On a GXS/1236 kit it is a wraparound metallic protection case, three heat-shrinkable outer jacketing sleeves — 1 centre plus 2 side sleeves — 4 worm drive clips and a roll of mastic sealing tape. Water finds the side sleeve whose adhesive never reached clean sheath. Opened up, the copper woven flexible mesh tape is the tell: green where it should be bright. GI mesh is offered in place of the case, and swapping it changes the mechanical protection, not the sealing.
2 · Stress control missed or misplaced
Field lines crowd at the screen cut-back. GXS/1236 answers that with 3 black heat-shrinkable stress control sleeves, 6 or 9 pieces of stress control mastic and 6 grey or yellow strips. Leave the mastic short of the cut edge and a void survives the shrink. The joint then passes handover and degrades for months, because discharge erodes before it punctures. The catalogue's limit is ≤ 5 pC at 11, 22 or 33 kV by class. Below 3.3 kV the mechanism does not exist: a GXLT/GPLT kit lists no stress control item at all.
3 · Contamination on the XLPE
A GXS kit ships 1 nylon string for cutting the XLPE insulation, 3 pieces of aloxite emery tape, 6 cleaning tissues and a mopping cloth. That list is the sequence: cut, abrade, clean, then nothing touches the surface. Semi-con is carbon-loaded, so a smear left behind is a conducting path lying flat against a graded surface. The discharge extinction voltage the GXS sheet demands — 12.7 kV minimum on the 12 kV class — is measured on a clean joint. A bare hand on cleaned insulation undoes it and leaves no mark you can see at the time.
4 · Under-shrink
Heat-shrink is shape memory. The material recovers only above its crystalline temperature; short of that it softens, moves, and never pulls down. The adhesive does not flow either, which is the half that actually loses the joint. Cut open, an under-shrunk sleeve peels instead of tearing and shows dry adhesive on one face. Do not wait for a colour change to confirm it. No joint-kit sleeve in this range publishes an indicator, and the one the manufacturer does publish sits on a different product for a different job.
5 · The ferrule
Three ferrules on a GXS/1236 kit, one on a GXSS single-core kit. GPS/1112 supplies a crimping type in copper or aluminium to suit the conductor. Transition kits use a barrier type instead, oil migration from the paper side being the reason. Undersize it, mismatch the metal or crimp it with the wrong die and the ferrule runs warm. Handover tests see none of that. The qualifying regime does. It puts the joint through 63 heating cycles in water, conductor at 95°C, before the kit is called fit.
6 · The earth path
Continuity arrives as listed parts with quantities against them. GXS/1236 carries 3 rolls of copper woven flexible mesh tape and a solder tack or roll spring. Then 4 worm drive clips and 1 tinned copper braid as the main earthing. Paper cable adds plumbing metal and jubilee clips. Break any of it and the joint still looks finished, still tests clean, still carries load. It is only wrong during a fault, and that is the worst moment to find out. The published sequence has no row that measures it — this one is caught on site or not at all.
The failure that was never a joint
Before any of the six applies, settle one question: was the conductor ever damaged? A torn sheath over sound cores is not a jointing job. Cutting the cable there manufactures a joint where none existed, and every mechanism above then applies to something that was fine an hour earlier.
The manufacturer's own answer to that case is not a kit. It is the wrap-around repair sleeve. Fold it around the cable, zip it shut on a stainless steel channel, shrink it in place. The sheet states that shutdown of the system is not required. This is the one case on the page where reaching for a joint kit is the wrong call.
Sealing and mechanical protection are also the same subject, not two. The impact test says so in its order of operations: the joint is struck, then immersed, then insulation resistance is read at 500 V DC for one minute. Backfill a joint onto a stone and you have set up the first mechanism on the list months in advance.
Which published test row would catch which failure
| Failure | Test row | Parameter | Stated result |
|---|---|---|---|
| Water through the outer seal, MV | Heating cycle test in water, straight-through joint | 63 cycles of 8 hours, 5 hours heating and 3 hours cooling, 16 / 30 / 45 kV AC by class | No breakdown and no flashover |
| Water through the outer seal, 1.1 and 3.3 kV | AC high voltage test, joint immersed in water | 5 minutes at 3.5 kV and 6 kV | No breakdown and no flashover |
| Seal, corrosion and tracking together | Examination, after the electrical sequence | Joint opened and inspected | No cracked filling, no moisture path crosses the primary seal, no corrosion, no tracking |
| Stress control missed or misplaced | Partial discharge test, at ambient and at 95 to 100°C | 11 / 22 / 33 kV by class | ≤ 5 pC |
| Contamination on the insulation | Discharge extinction voltage, GXS/1236 sheet | 12.7 / 25.4 / 38 kV minimum by class | < 5 pC |
| Under-shrink and the voids it leaves | Impulse withstand at elevated temperature | 10 positive and 10 negative impulses, conductor at 95 to 100°C, 95 / 125 / 170 kV | No breakdown and no flashover |
| A ferrule that runs warm | Thermal short-circuit test | Two short-circuits raising the conductors to the cable's θsc | No visible damage |
| Mechanical damage that lets water in | Impact test, then insulation resistance immersed | 1 minute at 500 V DC | >10³ Ω |
| Armour, screen and earth continuity | No row in the published sequence | — | Verified on site before the case goes on |
Test rows quoted from the guaranteed technical particulars printed in the manufacturer's catalogue — the MV sequence declared against IEC 60502-4 and CENELEC HD 629.1, and the 1.1/3.3 kV sequence declared against IEC 60502, IEC 60060 and IEC 60230. Pairing each row to a failure mode is this site's reading of them, not the manufacturer's.
No source for that
Five things this page would have to invent to tell you. Which of the six is commonest. No failure-rate distribution appears anywhere in the record this site works from, so the six are ordered by build sequence and by nothing else. A shrink temperature for the joint-kit sleeves. The semi-conductive breakout sheet prints 125°C; no joint kit sheet prints one. Every kit ships an installation instruction manual, and that is the document which governs. A crimp die, tool or torque for the inline ferrule. Not published. A minimum overlap for the outer jacketing sleeves. The wrap-around repair sleeve publishes ≥25, ≥50 and ≥75 mm by size — a different product, and those figures do not transfer to a joint. IP68 on a joint kit. IP68 is claimed for the wrap-around repair sleeve and for the semi-conductive breakout. The joint kit sheets do not carry it. If any of those decides your case, ask through contact rather than working from an assumption.
What gets asked after a failure
Is a heat-shrink cable joint actually waterproof?
Does IP68 apply to a joint kit?
How long does a bad crimp take to show?
Can you tell from outside that a sleeve is under-shrunk?
Was it the kit or was it the jointing?
Where each of these goes next
- The order an MV joint goes togetherSetting-out, cut-back, the earth stage, and the two mistakes you cannot undo.
- What a joint kit's type tests actually proveThe full sequence by class, with the limits quoted above put back in context.
- Joints that live in waterDuct, draw pit and permanent immersion, and where the sealing chain gives way.
- Every part in a cable joint kit, namedThe manifest with quantities, so a short kit is caught before anything is cut.
- The sleeves inside a joint kitWhich sleeve does which job, and the order they are meant to recover in.
- What a cable joint has to put backThe four losses, if you want the mechanism before you want the failure.
What to put in the first message
The kit code marked on the box. The cable construction and conductor area, the system voltage, and whether the joint is buried, ducted or sitting in a pit. Roughly how long it ran. Everything named on this page comes out of kits made by Gala Thermo Shrink Pvt. Ltd. of Mumbai; its GXS/1236 joint kit and GXLT/GPLT splicing kit pages list the components named above.