After the fault
When a cable accessory fails, proving whose fault it was
Three verdicts are available, each needs different evidence, and most of that evidence stops existing within a week.
Three verdicts, and only one of them is the product
A joint goes down at three in the morning and by breakfast there are three theories. The utility says the kit was defective. The contractor says the kit was wrong for that cable. The supplier says the joint was made badly. All three describe the same hole in the ground, and only one of them ends with somebody being paid.
An investigation can reach exactly three verdicts. The product was defective. The specification was wrong. The workmanship was bad. Anything else on the report is commentary.
No published figure splits Indian field failures across those three, and this reference will not invent one. What the paperwork shows is simpler. Two of the three verdicts land on your own side of the table, and those two are the hardest to disprove afterwards, because the documents that would settle them are exactly the documents nobody was told to keep.
That is the useful way to read this page. Whose fault it was gets decided by which documents exist, and which documents exist was decided by you, in the enquiry and in the order, months before anything went bang.
What each verdict has to be proved with
- The product was defective
- You need three things: the batch marking off the failed part, the carton it came out of, and the type test report the batch was built against. Then read what that report covers rather than that it exists. The CPRI certificate on the file behind this reference is dated August 2012 and covers 33 kV heat-shrink joints and terminations. It records what one stated construction did on one date. It is not a statement about the lot that reached your store years afterwards, and no test house has ever claimed it was. The certificates and approvals that exist, with the dates printed on them, are listed on the manufacturer's own certifications page.
- The specification was wrong
- You need your own enquiry and the supplier's quoted scope, printed and laid side by side. The comparison is mechanical: rated voltage class, conductor cross-section range, cable outside diameter range, indoor or outdoor, and the service environment. Anything the supplier shipped that your enquiry never asked for is a finding against your document, not against theirs. This verdict is decided in the RFQ, and it is decided in your favour or against you before a single quote comes back.
- The workmanship was bad
- You need the jointing record: who made the joint, on what date, at what ambient temperature, against which revision of the installation instruction sheet, and what qualification that person held. The instruction manual is a numbered line on the kit contents list, alongside the cleaning tissue, the emery tape and the mopping cloth, so we did not have instructions is not available to anybody. Where no record exists, this verdict cannot be excluded — and a claim that cannot exclude it settles at a discount, if it settles.
The sequence below is a reconstruction, not a published procedure
No manufacturer and no standard publishes an evidence-recovery procedure for a failed cable accessory. What follows is this reference's own reconstruction of what a claim later turns out to have needed. Your utility's incident procedure, your permit-to-work rules and the safety instructions supplied in the kit override every line of it. Nothing here is a working instruction for an energised circuit.
Four things to secure in the first 48 hours
Every one of these is routinely destroyed by the crew restoring supply, which is their job and not their mistake. Somebody has to be told to preserve them, in advance, in writing.
Recover the failed assembly whole
Cut generously on both sides and lift it intact. A joint sectioned on site by whoever reached it first with a hacksaw has destroyed the only physical evidence there was. Bag it, seal it, and label the bag with the circuit, the date, and the name of the person who lifted it. Nobody opens it until the parties have agreed who opens it and where.
Find the carton and the batch marking
The batch marking is the only thing that connects a failed part to a production record. Without it the supplier is answering questions about a product; with it, about a batch. Cartons are normally in the site skip by the second morning, which is why this belongs in the first 48 hours rather than in next week's meeting.
Photograph the failure surface before anyone cleans it
Photograph in position first, then again after removal, with a scale in frame and the supply end marked. Do not wipe the surface. On a surface fault the deposit and the burn track are the finding, and a cleaned specimen has had the answer rubbed off it.
Pull the commissioning test record for that circuit
The test result after jointing, its date, and the instrument it was taken on. A circuit that was never tested after jointing, or tested with no record kept, has already removed the workmanship verdict from anybody's reach, including yours.
The one mechanism buyers misattribute most often
An outdoor termination flashes over during the first heavy rain after a long dry spell. It is reported as a product failure more often than as anything else, and it is usually neither a defect nor bad jointing.
The path is across the surface, not through the insulation. Airborne salt, cement dust or industrial deposit builds on the exterior through the dry months. Light wetting turns that layer conductive, leakage current dries it in bands, and the discharge across a dry band walks the length of the termination. Nothing punctures. What decides whether it happens is the length of the surface path, and that length is a specification quantity.
Look at how the shed count is actually published. In the manufacturer's combined joints and terminations catalogue, the three-core outdoor XLPE and EPR termination kits list six rain sheds at 12 kV (GXO/E-1112), nine at 24 kV (GXO/E-1524) and twelve at 36 kV (GXO/E-3336) — two, three and four per core. The 12 kV indoor kits are supplied with none at all, the 24 kV indoor kits with one per core, the 36 kV indoor kits with two.
That chart is indexed on voltage class and on indoor against outdoor. It is not indexed on pollution. Your coastline, your cement plant, your fertiliser unit and your six rainless months appear nowhere in it. So if the enquiry named a voltage and a cable size and said nothing about the environment, the standard shed count is what shipped, and the standard shed count is what you ordered. That is a specification verdict, and it was written by the buyer.
Where a specification has to name the requirement, IEC 60815 is the document that sets a creepage distance against a site pollution severity class. It is cited here as a well-known published standard rather than from the manufacturer's literature, so establish the current edition and the part that applies to your insulator type before pasting the number into a tender.
The material side is separately published and is rarely where this argument ends. The creepage extension skirt is a cross-linked polyolefin qualified against tracking and erosion to 3.25 kV for 20 minutes under ASTM D2303, with a dielectric strength of 15 kV/mm minimum under ASTM D149 and a shrink temperature of 125 °C. Those are material qualifications. None of them tells you how many sheds your site needs, and no claim has ever been won by quoting them.
Sources disagree: the sheet you would cite has the wrong class in its heading
In the manufacturer's combined joints and terminations catalogue, the electrical specification table for the GXO/E-3336 series is headed Electrical Specifications (24 kV Rated Accessories) — the same heading printed above the GXO/E-1524 table earlier in the document. The figures beneath the two headings are not the same. The 1524 table gives 50 kV AC for one minute, 96 kV DC for thirty minutes and 125 kV impulse crest; the 3336 table gives 75 kV, 144 kV and 170 kV. Read exactly as printed, a 36 kV series is certified under a 24 kV heading. If that page is going to be attached to a specification or produced in a claim, get the rated class confirmed on letterhead first.
Rule these out before you name the accessory
A claim aimed at the wrong component dies slowly and bills the whole way down. Five questions, in this order, before anybody writes to the supplier.
- Was the cable itself faulted? A fault that started a metre away in the cable arrives at the joint and is found at the joint, because that is where the crew dug. Ask for the fault-location record and the distance it reported, not the conclusion somebody drew from it afterwards.
- Was it the crimp? A lug or ferrule pressed with the wrong die runs hot under load and cooks the insulation around it from the inside out. That is a tooling and workmanship finding on the connector. The accessory that melted over the top of it is the victim, not the cause.
- Was the earthing actually continuous? Screen and armour earthing is assembled on site from the tinned copper braid, back-up ring, worm-drive clips, solder and flux listed in the kit contents. If the armour earth was never made off properly, fault current went somewhere nothing on that circuit was designed for.
- Did anything happen to the network that night? A switching surge, a lightning strike or a sustained over-voltage loads every accessory on the feeder at the same instant. Three joints down in one night is a system event, not a batch defect, and the two are argued completely differently.
- And if one of those is the answer, say so early. The accessory claim closes and a different one opens — against a contractor, against an installer, or against nobody. Closing a wrong claim in week one is usually worth more than the claim was.
All four of these were decided in the purchase order
Every verdict on this page depends on a document that somebody had to have been obliged to produce. That obligation is a clause. Clauses cost nothing before award and cannot be bought at any price afterwards.
Acceptance criteria fix what the goods must satisfy on your dock, tested how, by whom, and what happens to the rest of the lot when a sample fails. Without them, the goods were accepted is the supplier's first and often last answer.
Sample retention obliges the supplier to hold a sealed sample from the batch supplied, for a stated period, released to a named laboratory on your instruction. Skip it and every product-defect claim collapses into an argument about whether the burnt article in the bag is representative of anything.
Batch traceability obliges the marking on the goods to map to a production record the supplier will produce on request. Ask what the marking looks like and where on the part it is applied before the first delivery arrives, not on the morning you need to read one off a charred sleeve.
Retest on failure decides in advance what happens next: the retest sample size, who pays for it, and whether its result is final. Left unwritten, this is the clause both parties re-argue for a month while the lot sits in your store.
The wording for all four is on the acceptance tests page. Read it with how a type test report is read beside it, because the first clause is worthless if you have accepted a report that covers a different construction.
Not published — ask for it in writing
No warranty period, claim procedure or liability limit appears anywhere in the manufacturer's published literature behind this reference, and this site will not infer one from silence. Ask for three things on letterhead before award: the warranty period and the event that starts it, the address and format for a valid claim notice, and the name of an independent laboratory the supplier will accept for a joint post-mortem. Agree that third one while nothing has failed and nobody is angry.
Write the clauses before you need them
Nothing on this page can be arranged after a failure. Every one of these documents exists because somebody was contractually required to keep it, and that requirement is created once, in the order.