35 kV Collector Circuits: How MV Cable Selection Actually Works

01What a collector circuit is, and why it is specified differently

Utility-scale generation is built as a hierarchy of voltages, and the collector circuit occupies the middle rung. At the bottom, PV strings run at up to 1,500 V DC and turbines generate at roughly 690 V to 1 kV. At the top, the point of interconnection may be 110 kV, 220 kV or higher. Between them sits the collector system, typically 33 kV or 35 kV, whose job is to aggregate tens or hundreds of megawatts across a site that can be several square kilometres in extent.

That geography is what makes the collector circuit a cable problem rather than a switchgear problem. There is one substation and many generators, so the cable route length is set by the site layout, and the route length is what drives the screen bonding decision, the induced screen voltage, and how much of the standard ampacity table you can actually use.

Three source types on one network

A hybrid or combined site may have all three of the following collector branches, and they behave differently under fault:

• PV inverter branches — current-limited by the inverter's own rating, so the MV cable's earth-fault current is largely set by the inverter contribution and the transformer impedance.
• Wind turbine branches — each turbine is an induction or doubly-fed machine with its own reactive behaviour, and the collector cable may also have to carry the reactive current the turbines exchange with the grid.
• Battery storage branches — a storage unit can deliver fault current from both the battery and the power conversion system, and it can also be a source when the rest of the site is de-energised.

For cable selection, what matters is that the earth-fault current and its clearing time are not the same on all three branches, and the metallic screen is sized for that current. A screen sized for an inverter-limited branch may be undersized for a storage branch on the same network.

Where the collector circuit sits between two standards worlds

This is the structural fact that shapes every MV procurement decision, so it is worth stating plainly. A 35 kV collector circuit can be built to one of two parallel standards families, and they are not interchangeable:

• The IEC system — IEC 60502-2 defines the cable itself for rated voltages from 6 kV (U_m = 7.2 kV) up to 30 kV (U_m = 36 kV), covering construction, dimensions and test requirements for extruded solid insulation in fixed installations. Above that ceiling the same family continues in IEC 60840, which covers rated voltages above 30 kV (U_m = 36 kV) up to 150 kV (U_m = 170 kV).
• The North American system — Type MV cable is defined in NEC 311.2 as solid dielectric insulated cable rated 2001 V up to and including 35,000 V nominal, and a North American MV catalogue typically names UL 1072 for the Type MV designation alongside ICEA S-93-639 / NEMA WC 74 and AEIC CS8, which set the insulation thicknesses and test voltages.

The two families use different rated temperatures, different insulation thickness tables and different test voltages for what a buyer may reasonably think is the same cable. On a given project the governing one is decided by the site's wiring regulations and by the specifications of the equipment the cable terminates into — not by preference. Our own medium voltage armoured power cable page is explicit that the ICEA and AEIC values are named for identification only and are not used or implied in the IEC construction we publish there.

02The voltage basis: U₀, U and U_m are three different numbers

If there is one error that propagates through an entire MV enquiry, it is collapsing the voltage rating into a single figure. A "10 kV" cable does not exist as a procurement item. What exists is a cable rated 6/10 kV, whose maximum system voltage U_m is 12 kV.

The three symbols mean this:

• U₀ — the rated power-frequency voltage between conductor and earth, or between conductor and metallic screen.
• U — the rated power-frequency voltage between conductors.
• U_m — the maximum value of the highest system voltage for which the cable is designed to operate in service.

IEC 60502-2 works in five standard ratings, and each has its own U_m and its own specified AC test voltage. Writing them as a single "6–30 kV" range is how the wrong insulation thickness and the wrong routine test value end up on a purchase order.

Ratings, U_m values and AC test voltages as published on our medium voltage armoured power cable page for the IEC 60502-2 construction. The AC test voltage column is the value specified by IEC 60502-2 for that rating — a standardised test value, not a service voltage and not a description of any manufacturer's in-house routine.

The practical consequence runs like this. A 33 kV collector network usually sits under the 18/30 kV rating, because its U_m of 36 kV covers the system. A collector network specified at 35 kV nominal does not: its equipment maximum voltage is 40.5 kV, so it is above the 36 kV ceiling of IEC 60502-2 and its cable is procured against IEC 60840, which covers rated voltages above 30 kV (U_m = 36 kV) up to 150 kV (U_m = 170 kV). On the North American side the type MV definition tops out at 35,000 V nominal, so the number that governs on a North American project is the nominal one, and on an IEC project it is U_m. Same physical system, two different numbers being checked — and on the IEC side, a standard boundary that a 35 kV designation crosses.

03Construction: what is fixed by the rating and what you choose

An MV cable is a layered dielectric system, not a conductor with more rubber on it. The construction chain is the same at every rating, but the responsibilities change at each layer.

• Conductor — annealed copper to IEC 60228 Class 1 (solid) or Class 2 (stranded), sized for both the continuous load and the conductor short-circuit duty.
• Conductor screen — an extruded semiconductive non-metallic layer bonded to the insulation, whose purpose is to eliminate voids at the conductor surface.
• Insulation — XLPE as standard, with EPR/HEPR available; thickness set by the voltage rating.
• Insulation screen — a second extruded semiconductive layer over the insulation, bonded or strippable.
• Metallic screen — copper wire (CWS) or copper tape (CTS), sized for the earth-fault current and its duration.
• Bedding or separation sheath — extruded layer under the armour.
• Armour — aluminium wire (AWA) or steel tape (STA), where mechanical protection is required.
• Oversheath — PVC, LSZH or PE depending on the environment and the fire requirement.

The two semiconductive layers are the reason an MV cable cannot be jointed like a low voltage one. The conductor screen and the insulation screen are what hold the electric field inside the insulation; a void or a protrusion at either interface becomes a site of partial discharge, and partial discharge is what eventually fails the cable. This is also why the finishing of the screen at a termination is an accessory-engineering question rather than a cable question — our cold-shrink termination and joint range for 10 kV to 35 kV exists for exactly that step.

Insulation thickness is fixed by the rating, not by the conductor size

The single most useful dimensional fact in the whole IEC 60502-2 system is this: from 6/10 kV upwards, the nominal insulation thickness is determined by the rating and does not change with conductor size. Only the 3.6/6 kV rating moves with the conductor.

Reference construction values published on our IEC 60502-2 medium voltage armoured power cable page, cross-checked against the dimensional data of the individual voltage ratings of a European MV cable range held in our library (240 mm² single-core: 2.6 mm at 3.6/6 kV, 3.4 mm at 6/10 kV, 4.5 mm at 8.7/15 kV, 5.5 mm at 12/20 kV, 8.0 mm at 18/30 kV). Sizes below 16 mm² and the 800–1000 mm² range are quoted per rating. The thickness applied to a specific order is confirmed on the drawing.

Read across the 6/10 kV column: 3.4 mm at 16 mm² and 3.4 mm at 1000 mm². The thickness is not a function of how much copper is in the conductor. It is a function of how much voltage the dielectric has to withstand, which is set by the rating alone. A buyer who asks for "thicker insulation for the bigger cable" is asking for something the standard does not provide, and a supplier who quotes it is quoting outside the standard.

04The metallic screen: the layer that decides the procurement

The metallic screen is where MV cable selection actually gets decided, because it is the one layer whose duty is set by the project's own fault study rather than by the voltage rating.

The NEC states the purpose of shielding directly. Under 311.44, nonshielded ozone-resistant insulated conductors with a maximum phase-to-phase voltage of 5000 V are permitted in Type MC cables in industrial establishments where maintenance and supervision ensure that only qualified persons service the installation; for other establishments, solid dielectric insulated conductors operated above 2000 V in permanent installations must have ozone-resistant insulation and must be shielded, and all metallic insulation shields must be connected to a grounding electrode conductor, a grounding busbar, an equipment grounding conductor, or a grounding electrode. Two exceptions carve out narrower cases — nonshielded conductors listed for use up to 2400 V under stated conditions, and airfield lighting cable in series circuits rated up to 5000 V. The code's own informational note lists what the shield is for: confining the voltage stresses to the insulation, dissipating insulation leakage current, draining off the capacitive charging current, and carrying ground-fault current so that ground-fault protective devices operate in the event of a cable fault.

That last function is the sizing criterion. The screen is a fault-current path, and it is a much smaller conductor than the phase conductor it sits around.

Copper wire and copper tape are not equivalent at the same nominal size

Both screen types are standard, and both are described by a nominal cross-section in square millimetres. Those two numbers do not produce the same short-circuit performance, because a tape screen and a wire screen of equal nominal area do not have equal effective cross-section once the tape's overlap and the wire's lay are accounted for.

Values for the reference construction as published on our medium voltage armoured power cable page. The copper wire screen cross-section is graded by conductor size — 16 mm² for 16–120 mm², 25 mm² for 150–300 mm², 35 mm² for 400–630 mm², 50 mm² for 800–1000 mm² — and that grading does not change between the five voltage ratings. The CTS figure is quoted for the 6/10 kV rating with a 16 mm² tape; tape cross-section is derived from the construction and rises with conductor size. Screen selection must be made against the earth-fault current and its clearing time for the actual installation.

Look at the first and last rows: 1.96 kA against roughly 0.86 kA for the same 16 mm² nominal figure. That is a factor of well over two, and it is the reason a specification that says "16 mm² copper screen" without stating wire or tape has not actually specified the screen at all.

Armour is a mechanical choice with an electrical consequence

Armour exists to protect the cable from mechanical damage — the load of backfill, a dig-in, a rodent, the pull load during installation. Two forms dominate MV work:

• AWA — round aluminium wire armour, applied helically over the extruded separation sheath. Non-magnetic, so it does not introduce the eddy-current and hysteresis losses a steel armour would in a single-core AC circuit, and it is the usual choice for single-core MV cable.
• STA — steel tape armour, generally on multicore constructions where the magnetic circuit is balanced around all phases and the tape's mechanical protection per unit cost is higher.

The electrical consequence is that armour is a conductive part, and therefore has a bonding requirement of its own, which the NEC states plainly in 311.36: the metallic shield, sheath or armour must be connected to a grounding electrode conductor, a grounding busbar, or a grounding electrode. The same article's 311.44 carries the shielding requirement itself. On a single-core cable the armour and the screen sit in the same field, so the bonding decision taken for the screen applies to the armour as well. That is the subject of the next section.

05Screen bonding: the decision that changes the cable's current rating

This is the part of MV collector design that is least visible in a procurement document and most consequential to the installed plant. The screen runs parallel to its own conductor for the whole route, which makes it, in part, the secondary of a 1:1 transformer. Alternating current in the conductor induces a longitudinal EMF in that parallel screen, proportional to the conductor current, the route length and the conductor-to-screen mutual inductance.

What happens next depends entirely on where the screen is earthed.

Both ends bonded: no standing voltage, but a circulating current

Earthing the screen at both ends closes a loop. The induced EMF drives a current around the screen–earth circuit, and that current produces an I²R loss in the screen — heat, and a correspondingly lower permissible current for the circuit. Sources differ on the magnitude: one cable-engineering source quotes 50–80% of conductor current on long circuits, another quotes 30–40% in some installations. Both figures come from practice at different voltage levels and constructions, so the honest statement is that the circulating current can reach a substantial fraction of the conductor current and that it must be computed for the actual route rather than assumed from either range.

The trade-off is that both-ends bonding leaves a standing screen voltage of only a few volts, requires no surge voltage limiters and no parallel earth-continuity conductor, and is the safest arrangement to touch. It is normal practice on short runs where the induced EMF and the resulting loss are affordable.

Single-point bonding: no circulating current, but a standing voltage at the open end

Earthing the screen at one end only breaks the loop, so no circulating current flows and the screen loss disappears — and the circuit keeps its full tabulated ampacity. The price is a standing induced voltage at the unearthed end, proportional to route length and load current. That voltage is what limits the length of a single-point-bonded section, because it has to stay below the project's touch-voltage limit.

Single-point bonding also needs two additional components: a parallel earth-continuity conductor so that fault current has a path back, and surge voltage limiters at the open end so the screen survives through-fault and transient overvoltages. Link-box-mounted SVLs in the range of about 0.8–4.8 kV MCOV, and outdoor units around 4–14 kV, are typical product practice — selected by an insulation-coordination study, not by the cable standard.

Cross-bonding: cancelling the EMF by transposition

Cross-bonding splits a major section into three electrically equal minor sections and transposes the screens at each intermediate joint, so that each screen in turn sees the induced EMF of each phase. Over a complete transposition cycle the three induced EMFs sum to approximately zero: no circulating current, no large standing voltage, and screens solidly earthed at both ends of the major section.

The cost is complexity. Cross-bonding needs sectionalised joints, a link box at each transposition point, and equal section lengths — unequal minor sections leave a residual circulating current, typically a few percent of conductor current, because the cancellation is incomplete. A sheath fault on a cross-bonded system also unbalances the transposition, and the same sheath test is the only thing that finds it.

Schemes and their operating characteristics are established cable-engineering practice, documented by Keentel Engineering, PTE (Power Transmission Engineering), QD Britop and Power & Cables, all held in our library. The circulating-current magnitudes quoted in the text are illustrative ranges from separate sources at different voltage levels, not a design rule; the SVL figures are link-box product practice rather than a standard requirement.

The NEC states the ampacity consequence explicitly

This is the point at which the bonding decision and the ampacity table meet, and the NEC is unusually direct about it. Under 311.60(D)(1), the ampacities given in four specific tables — 311.60(C)(69), (70), (81) and (82), which are the single-conductor tables for cables in air and directly buried respectively — apply only to cables whose shields are grounded at one point only. Where shields are grounded at more than one point, the ampacities must be adjusted to take into account the heating due to shield currents. The code then adds an informational note: the other tables in that group contain the ampacity of cables with shields grounded at multiple points.

So the tabulated ampacity is not a property of the cable. It is a property of the cable plus its bonding scheme. A collector circuit designed from the single-point-grounded table but installed with both ends bonded will not deliver the tabulated current, and the error is not detectable from the cable datasheet.

06Burial, ducts and the ampacity baseline

A collector circuit on a wind or solar site is almost always buried. That makes it worth knowing exactly what the standard ampacity tables assume, because the baseline conditions are specific and the corrections away from them are prescribed.

For direct-buried cable and underground duct banks, 311.60(F) sets out the basis on which the NEC ampacity tables are computed: an ambient earth temperature of 20 °C, an arrangement in accordance with the standard installation figures, a 100% load factor, a soil thermal resistance (Rho) of 90 (cm·K/W — the NEC states the figure without a unit), conductor temperatures of 90 °C or 105 °C, and a maximum depth to the top of the duct bank or direct-buried cable of 750 mm and 900 mm respectively. Minimum burial depths follow the general installation rules for underground conductors and cables under 300.50.

The relevant correction provisions are:

• Ambient earth temperature. Where the ambient differs from the 20 °C baseline, 311.60(D)(4) applies a correction factor from the standard's table. At 30 °C the factor is 1.10 for a 90 °C conductor and 1.07 for a 105 °C conductor; at 10 °C or below it is 1.26 and 1.21.
• Burial depth. Under 311.60(D)(2), no correction is required where the burial depth is decreased. Where it is increased beyond the tabulated depth, a derating of 6% per 300 mm of additional depth is permitted for all values of Rho.
• Multiple circuits. The tables are published per circuit count — one, three, six, and larger groupings — and the ampacity falls as more circuits share the same corridor.
• Two different ampacities applying. Under 311.60(A)(2), where more than one calculated or tabulated ampacity could apply to a circuit, the lowest value is used, with a narrow exception for short portions of lower ampacity not exceeding the lesser of 3.0 m or 10% of the total circuit.

The soil thermal resistivity term is where collector circuits most often go wrong, because it is a site measurement rather than a catalogue value. If the backfill or the native soil is drier or more insulating than the assumed baseline, the cable runs hotter at the same current, and no amount of cable oversizing fixes a thermal problem in the ground. Our separate treatment of soil thermal resistivity and its effect on direct-buried cable ampacity goes into the measurement side; this article only needs the point that the baseline is an assumption, and that a collector circuit running through a long dry section is a different thermal problem from the same cable in damp clay.

07IEC 60502-2 against UL Type MV-105: two procurement worlds

A collector circuit specification has to name one of these families, and the two are not convertible. Naming both, or assuming the values of one carry into the other, is a defect that surfaces at the type-test stage.

IEC column: IEC 60502-2:2014, Edition 3.0 — scope, voltage range and exclusions as recorded in our library from the IEC webstore record for the consolidated edition including Amendment 1:2024. North American column: NEC Article 311 (medium voltage conductors and cable, 2022 California Electrical Code) for the Type MV definition, the MV-90 / MV-105 temperature ratings, the 100% / 133% / 173% insulation levels and the ampacity tables; UL 1072 and the ICEA / AEIC families as named in the MV cable specifications of a North American cable manufacturer held in our library. The insulation thickness, test voltage and rated temperature values of the two families are not interchangeable, and a construction qualified to one is not qualified to the other without its own testing. Note that the 2023 edition of the NEC renumbers this article as Article 315; clause references in this article follow the 2022-based text we hold, and a project working to the 2023 edition should verify the numbering against its own copy.

The insulation-level convention you will meet in a Type MV specification

The North American system carries a concept the IEC system expresses differently. Shielded MV cable is specified at a 100%, 133% or 173% insulation level, and the difference is a protection requirement rather than a cable quality grade. The three levels are defined in 311.10(C):

• 100% insulation level — permitted under 311.10(C)(1) where the system has relay protection such that ground faults will be cleared as rapidly as possible and, in any case, within 1 minute. These cables are applicable to grounded systems, and are permitted on other systems provided the same clearing requirement is met in completely de-energising the faulted section.
• 133% insulation level — permitted under 311.10(C)(2) in situations where the clearing-time requirement of the 100% level cannot be met and the faulted section will be de-energised in a time not exceeding 1 hour. It provides additional insulation margin for the longer duration a fault may persist on an ungrounded or impedance-grounded system.
• 173% insulation level — permitted under 311.10(C)(3) only where all four conditions hold at once: the installation is in an industrial establishment where maintenance and supervision ensure that only qualified persons service it; the fault-clearing time requirements of the 133% category cannot be met; an orderly shutdown is required to protect equipment and personnel; and the faulted section will be de-energised in that orderly shutdown. It is the level for a system that is allowed to run with a fault present until it can be taken out deliberately, rather than being tripped immediately.

This is the clearest illustration of why the two families cannot be mixed: the IEC system already encodes the slow-clearing case in the rating itself. The 8.7/15 kV rating exists precisely for installations where earth-fault clearing is slower or the neutral is impedance-earthed, which is the same engineering situation the 133% level addresses in the North American system. Two different mechanisms, one underlying question — and the 173% level marks out the further case where the fault is not cleared at all until an orderly shutdown.

08What we publish, and what we will not pretend to publish

This article sets out the selection logic and the standard values that govern it. It is worth being explicit about the boundary, because MV procurement is full of documents that imply more certainty than the standards provide.

What this article does state, with sources: the voltage ratings, U_m values and test voltages; the fact that insulation thickness is fixed by rating from 6/10 kV upwards; the screen grading and short-circuit figures; the three bonding schemes and their trade-offs; the assumptions behind the buried ampacity tables and the correction provisions; and the scope of both standards families.

What this article does not do — and no honest article can: give you a current rating for your cable. The current rating is a function of the installation method, the ambient and earth temperature, the soil thermal resistivity, the circuit grouping, the burial depth and the bonding scheme. Any generic ampacity table for an MV collector cable is a table for some other site. That is why our MV cable page declines to publish one, and why the screen type is confirmed against the earth-fault duty you send us rather than offered as a standard option.

The other boundary is certification. IEC 60502-2 is a product standard: it specifies how the cable is constructed and tested. It is not a factory certification, and holding a cable built to it is not the same as holding an approval. Certification and test documentation for the construction supplied is issued with the quotation for that construction.