Cold Shrink vs Heat Shrink Cable Terminations: IEC 60502-4 Explained

01Why the joint fails before the cable does

There are three interfaces where an accessory can lose the argument with its environment, and only one of them is about the material.

• The electrical interface at the screen cut. The semiconductive screen cannot simply be stopped; the field has to be graded back to the insulation surface. Cold shrink does this with a geometric stress cone moulded integrally into the body. Heat shrink does it with a zinc-oxide (ZnO) stress-control material compounded into or applied under the tubing, which becomes conductive as field rises and smooths the stress distribution. Either approach works, and both are type-tested; what differs is how tolerant they are of a cable end prepared slightly outside the ideal.
• The moisture seal at the jacket and lug. Water enters at the place where the new insulation system meets the old cable jacket, and at the lug. This is why vendor kits ship sealing mastic as a distinct, named component rather than as an afterthought: cold shrink QT-III terminations are built with an integrated moisture-sealing mastic layer, and heat shrink termination kits carry yellow stress-control mastic and red moisture-seal mastic as separate items. A kit that treats mastic as optional has told you something about its field record.
• The mechanical interface under load cycling. As the conductor heats and cools, the cable expands and contracts radially and axially, and the accessory has to follow or open a gap. This is the difference that a datasheet rarely quantifies and a specification writer has to reason about.

What this article will not do is quote a failure rate. Comparative annual failure figures for cold shrink and heat shrink circulate freely on vendor blogs, but in this research we did not find a primary source — a utility report, a CIGRE or IEEE study, a testing laboratory result — that supports any of the specific percentages we saw repeated. We are not going to pass them on as if we had. What we can support, and what the rest of this article uses, is standard scope, type-test sequences and manufacturers' published installation and storage limits.

02How each technology actually makes its seal

The three families on the market are distinguished less by what they are made of than by how the sleeve is brought into contact with the cable.

Heat shrink begins as a tube of cross-linked polyolefin that is expanded at the factory and cooled in the expanded state. On site it is heated until the polymer's memory pulls it back onto the cable, and the hot-melt adhesive inside flows into the interfaces. The recovered sleeve is dimensionally stable and rigid — an excellent barrier, and one that no longer moves once it has cooled. Complete recovery for polyolefin tubing falls in the 120–135 °C range depending on the compound, which is why the technique depends on a trained hand holding a torch or hot-air gun at the right distance. Commercial heat-shrink MV termination kits, such as the ZnO-based outdoor termination families rated from 12 to 42 kV, are tested to IEC 60502-4, CENELEC HD 629.1 and IEEE Std 48.

Cold shrink starts as a tube moulded at its final diameter, then expanded and held open on a removable spiral support core. Installation is mechanical: position the body over the prepared cable end, then unwind and withdraw the core. The elastomer's elastic recovery does the rest, applying pressure uniformly around the circumference with no heat, no flame and no hot-work permit. For medium voltage this body is silicone rubber. It is worth being precise here, because the material story is layered and it is easy to get wrong: the cold shrink splice body itself — with its integrated stress-control device, conductive electrode and outer semiconductive layer — is silicone, while the thick-walled tube that rebuilds the cable's outer jacket is EPDM rubber. A specification that asks for "an EPDM cold shrink termination", or for "a silicone one", may be describing only half of the product in front of it. Ask for the layer.

Pre-moulded slip-on components are dimensioned at the factory to an interference fit and pushed onto the cable over lubricant. Dimensional precision is their strength, and it is why they dominate the higher voltage classes where partial-discharge performance at the interfaces is least forgiving. Their limitation is obvious from the same fact: the fit is fixed, so a wider range of cable diameters cannot be covered by one body.

03What each standard actually covers

Four documents come up in every MV accessory conversation, and they do different jobs. Confusing them is the most common specification error we see, because a kit can legitimately claim compliance with one and not another, and the claim will be true.

IEC 60502-4:2023 is the accessory side of the IEC 60502 series. Its scope is type-test requirements for accessories used with extruded-insulation power cables from 3.6/6 kV (Uₕ = 7.2 kV) up to 18/30 kV (Uₕ = 36 kV), and the test methods it applies are those of IEC 61442. Accessories for special applications — aerial cables, submarine and shipboard cables, and hazardous situations such as explosive environments, fire-resistant cables and seismic conditions — are outside its scope. The 2023 edition is not a light revision: it introduces a range-taking versus non-range-taking classification that decides how far an approval extends, adds dynamic short-circuit classes 0 to 3, offers partial-discharge testing options including at 2 U₀, removes specific DC tests from several test tables, adds a 240 mm² conductor size for accessory testing, and brings in a 2,500 A separable-connector size with its own methods. If a specification still cites an older edition, the test tables it points at are not the current ones.

CENELEC HD 629.1 is the European harmonisation document that many MV accessory type tests on the European market are actually run against. It is the reason a datasheet can say "meets and exceeds HD 629.1" while naming IEC 60502-4 in the same breath: the European sequence and the IEC sequence overlap heavily, and manufacturers test to both. When a test report is dated and numbered, as the good ones are, the report is the artefact worth asking for — the standard number printed on a brochure is a claim, not evidence.

IEEE Std 48-2020 is the North American counterpart for terminations, and it is broader in voltage than the IEC accessory standard: it covers indoor and outdoor cable terminations on shielded AC cables with laminated insulation rated 2.5 kV through 765 kV, and with extruded insulation rated 2.5 kV through 500 kV. Separable insulated connectors are excluded, because they are covered by IEEE Std 386. The 2020 revision added definitions for basic lightning impulse insulation level (BIL), basic switching impulse insulation level (BSL), termination families and insulation classes; updated the temperature range under service conditions; and — relevant to anyone comparing it with older test reports — eliminated DC voltage design tests and added requirements for termination families and the scope of an approval.

IEEE Std 404-2022 is the joint-side standard: electrical ratings and test requirements for cable joints used with extruded and laminated dielectric shielded cable rated in preferred voltage steps from 2.5 kV to 500 kV, plus test requirements for joint metallic-shield and jacketing components. If you are specifying a straight joint rather than a termination, this is the document that governs the acceptance test, not IEEE Std 48.

04The type-test sequence, in the order it is run

The reason to read a type-test sequence rather than a pass/fail certificate is that the order encodes the physics. A sequence does not just check that a joint works when new; it works the joint first, then checks it again, because the failure mode being screened for is the one that appears after thermal cycling has opened micro-voids at the material interfaces.

The table below reproduces the type-test sequence that 3M publishes for its QS200 three-core cold shrink inline joint series, tested to CENELEC HD 629.1 sequence S2 B1, at 6.35/11 (12) kV. We reproduce it because it is a worked example of the whole method from a primary source — the manufacturer's own technical data sheet, which cites an independent test report — and because reading it makes the partial-discharge discussion in the next section concrete.

Two things stand out when the sequence is laid out in order. The first is that the electrical checks are run twice — once on the fresh assembly and once after 126 cumulative thermal cycles in air and water. The second is that the repeated test is the one that matters most for a moisture-sealing accessory. A joint that passes a room-temperature partial-discharge measurement on the day it is made has demonstrated that it was assembled correctly. A joint that still measures below the limit after 63 cycles in water has demonstrated something about the material interfaces, which is the claim a buyer is actually paying for.

05Partial discharge: the threshold depends on which standard you quote

Almost every MV accessory argument we have been pulled into comes down to one line on a datasheet: a partial-discharge figure in picocoulombs. It is worth being careful with it, because a pC number without its test voltage and its source standard is not a specification — it is a number.

Two thresholds that are not interchangeable

Run the comparison as a buyer would, on published data rather than on sales language. The QS200 cold shrink joint series is tested, per the sequence in Table 3, at 2 times the rated phase-to-earth voltage with an acceptance limit of 10 pC. IEEE Std 48-2020, for terminations, specifies a partial-discharge test at a voltage raised to 120 % of its tabulated value in the relevant table, and states a limit no greater than 5 pC.

Why the “5 pC at 1.5 U₀” shorthand was removed from this article

An earlier version of this page stated that both technologies meet IEC 60502-4 acceptance with partial discharge below 5 pC at 1.5 U₀. On re-checking, that sentence was a composite: it took a threshold that belongs to one standard's test table and attached it to a test voltage from a different context. It read as precise and it was not sourced to anything. It has been replaced with the figures above, each tied to the document it comes from.

What the mastic and the cone are doing, in discharge terms

Partial discharge in an accessory almost always begins at a void or a surface path, and voids appear where two materials do not fully conform. This is the mechanism behind the three sealing components that vendors name explicitly in their kit contents. The stress-control material in a heat-shrink termination is formulated to fill irregular gaps and bond to the insulation as the tubing recovers, so that the region around the screen cut is not an air gap. In a cold shrink termination, the geometric stress cone is moulded into the body, so the field-grading element arrives already bonded to the insulation rather than being assembled in the field. The moisture-sealing mastic is what keeps the interface from becoming a water path once thermal cycling has been at work on it. A kit is only as good as the mastic and the cone in it, which is why the acceptance test in Table 3 is run again after the thermal cycles rather than only at the start.

06Installation: the temperature window, the tools and the site conditions

Installation conditions are where cold shrink and heat shrink diverge most sharply in practice, and they diverge on facts a project planner can check before ordering.

The two installation failures that come from the installer, not the product

With heat shrink, the failure that recurs in the field is incomplete recovery. When a sleeve is heated in the middle and not worked outward to both ends, the ends flare instead of conforming — the shape installers call fish-mouthing — and each flare is a funnel for water. The opposite error is a stationary heat source, which takes the material past its degradation point and chars the surface; a charred sleeve has lost the property it was installed for. Neither is a manufacturing defect, and neither shows up in a factory test.

With cold shrink, the corresponding errors are positional. The spiral core has to come out in the direction and order the instructions specify, and the body has to be sitting on the marked position before the first turn is taken, because once the elastomer has gripped the cable it cannot be slid. A support core pinched or scored during storage and handling is a third variant of the same problem: the body will still look right on the outside and will not have applied pressure where it matters. This is the practical reason cold shrink kits carry storage instructions on the packaging, and the reason a delivery that has been sitting in a hot container for a season deserves a look before it goes into a trench.

07Storage and shelf life: the discipline cold shrink demands

There is a tidy way to think about the trade. Heat shrink buys you storage patience and a lower unit cost, and asks you to control a flame on site. Cold shrink buys you a flame-free, technique-tolerant installation and a seal that keeps applying pressure, and asks you to manage your inventory. Neither is free; the cost simply lands in a different part of the project. A site with an unreliable laydown yard, or a spares policy that keeps accessory kits on a shelf for years, is a site where that second cost is easy to underestimate.

08Fitting the accessory to the cable you actually bought

Accessory selection is a diameter problem before it is a cross-section problem, and this is the point on which most imported-datasheet comparisons go wrong. Cold shrink bodies are chosen against the range of diameters they can grip, not against the conductor size alone, and the conductor size is often quoted as a guide because that is what buyers ask for.

The published data makes the relationship visible. A single-core outdoor QT-III termination series for the 12/20 (24) kV class is offered in three bodies covering insulation diameters of 16.3–27.4 mm, 21.1–38.9 mm and 26.7–45.7 mm, with maximum accepted lug diameters of 30 mm, 39 mm and 45 mm respectively; the same bodies appear against conductor ranges from 25–120 mm² up to 500–630 mm² depending on voltage class. Overlap between adjacent bodies is deliberate, and it is what allows a splice body to cover a cross-section transition within one kit. A buyer comparing two suppliers on "25–630 mm²" is comparing a headline that both can print while the bodies behind it cover different diameters.

The cable itself sets the ceiling on accessory performance

An accessory seats against the surface it is given. Where the cable's insulation diameter wanders along the drum, or its layers are eccentric, the recovery pressure around the circumference stops being uniform, and the interfaces that partial-discharge testing is designed to protect are the ones that suffer. This is why a specification that treats the cable and the accessory as two separate purchases tends to produce arguments later. Medium-voltage armoured power cable to IEC 60502-2 is built to the same standard family as the accessories that terminate it, which is the point at which the two halves of the specification can be checked against each other. If the route also involves direct burial, the thermal and moisture conditions that the joint has to survive are set by the installation method rather than by the accessory — direct burial versus duct installation works through how installation method changes the conditions at the joint, and soil thermal resistivity covers the burial side of the same problem. On a utility-scale site the same chain starts further upstream, where 35 kV collector circuit cable selection works through the voltage basis, the screen's earth-fault duty and the burial conditions that decide the diameter and screen type arriving at the termination.

09What to put on the purchase order

Six lines, each of which prevents a specific argument later.

• The full type-test report, not a summary certificate. A report identifies the laboratory that ran the test and the test sequence applied; a certificate that says "type tested" without either is a statement about the supplier's filing system. Where the report is referenced by number and date, as in the QS200 data quoted above, that is the marker of a document worth reading.
• The voltage class and the cable's insulation diameter range, not only the conductor size. These are the two numbers that decide which body is in the box.
• Indoor or outdoor, and termination or straight joint. The governing standard changes with the answer, and so does the physical product. A termination is tested to the termination standard; a joint to the joint standard.
• Production date and remaining shelf life, for cold shrink. Take the figure from the supplier's own data sheet rather than from a general rule, because it varies with compound and packaging: the kits quoted in section 07 state three years from date of manufacture, stored between −40 °C and +50 °C at below 70 % relative humidity, and other suppliers publish shorter periods. On a project with a long procurement lead, this is the line that stops a kit arriving with a year of its life already spent.
• Batch traceability. Kits are marked with cross-section ranges, voltage class, cable type, storage conditions and manufacturing codes. Recording that marking against the joint's location is what allows a fault to be traced to a batch rather than to a suspicion.
• The earthing and screen-continuity components. Constant-force springs and copper screen sleeves are named kit contents, not universal fittings. If the joint must maintain a metallic screen path, confirm which components the kit carries.

One closing note on how to read a claim. A datasheet that states a partial-discharge figure should also state the test voltage and the standard. A datasheet that states an installation temperature range should also state it for the product rather than for the compound it is moulded from. And a supplier who tells you which of their products is not right for your application has told you more than one who says everything is. Our own position is that accessories are specified from the cable outwards: the cable's construction, diameter and screen type come first, and the accessory that fits them is not always the one that was on the comparison table.