Lightning protection for a cast resin dry-type transformer is not one device — it is a chain of three jobs: intercept the strike before it reaches the equipment, clamp the overvoltage that arrives on the conductors anyway, and give the current a low-impedance path to earth without lifting the local ground potential. If any one of the three is missing, the other two do not save the transformer. This page walks through all three for SCB18 epoxy cast resin units, shows the separation-distance calculation that decides where the arrester physically has to sit, maps the Chinese specification numbers onto North American practice (IEC 62305 vs NFPA 780, IEC 60099-4 vs IEEE C62.11, UL 1449 SPDs), and gives you a spec checklist you can send to a supplier. It applies to 10 kV / 20 kV / 35 kV primaries — 12.47 kV, 13.8 kV and 34.5 kV for North America — whether the unit sits in a basement vault, a rooftop plant room or an outdoor enclosure.

Bottom line for buyers: the "18" in SCB18 is a loss-level code. It says nothing about lightning withstand. What protects the winding is the BIL (basic impulse insulation level) declared on the nameplate, an arrester with a protective level comfortably below that BIL, and — the step that gets skipped most often — a short enough distance between the two.

---

1. What Actually Fails When Lightning Reaches a Cast Resin Transformer

Two different things damage a transformer in a lightning event, and they need two different countermeasures.

• Amplitude. The incoming surge drives the terminal voltage above the winding insulation's impulse withstand. If the peak exceeds the insulation level, the insulation is punctured.
• Steepness. A lightning wavefront reaching the terminal in a fraction of a microsecond does not distribute itself evenly along the winding. At very fast rise times the voltage drops across the first few turns rather than across the whole winding, so the inter-turn and line-end insulation sees a far higher stress than the terminal voltage alone suggests.

Cast resin construction raises the stakes on both counts:

• Epoxy insulation is non-self-restoring. A puncture or a tracking path caused by a surge is permanent; there is no oil to re-impregnate and no recovery after the event.
• Cast resin windings have lower series capacitance to earth than oil-immersed windings. That makes the initial voltage distribution more non-linear and concentrates stress at the line end — which is exactly why the line-end turn insulation and any electrostatic shielding at the line terminal matter more, not less, on a dry-type unit.
• Void or delamination inside the casting is where partial discharge starts. A surge does not usually create the void; it finds the one that was already there. So the partial discharge level on the type test report and the surge protection are two halves of the same problem.

The wider cast resin dry-type transformer family is chosen for fire performance and indoor placement, not for surge ruggedness; surge performance has to be specified separately.

2. Why the "18" in SCB18 Tells You Nothing About Lightning

SCB18 is decoded as: S = three-phase, C = cast (epoxy resin) insulation, B = foil-wound low-voltage winding, 18 = the loss-level / performance code.

That last digit is a loss code tied to China's minimum energy performance standard (GB 20052). It governs no-load and load loss. It has no relationship to:

• the BIL of the high-voltage winding (set by IEC 60076-3 / IEEE C57.12.01);
• the clearances and creepage inside the terminal box;
• the impulse voltage distribution along the winding;
• whether the unit is rated for outdoor exposure.

So when you write the RFQ, ask for two separate lines: the loss figures (SCB18 / GB 20052 Grade 1 — [Confirm] the grade and current standard version per rating), and the insulation levels (BIL and power-frequency withstand per IEC 60076-3). Our SCB18 cast resin transformer page covers the loss and efficiency side; everything below is the insulation-coordination side that has to be added on top.

3. Job One — Intercept the Strike: External Lightning Protection

3.1 Air terminals and the building's LPS

If the transformer is indoors, it is protected by the building's lightning protection system (LPS), not by anything you bolt to the transformer. What you have to confirm is that the LPS exists, is maintained, and is class-matched to the local lightning risk.

Two standards families describe it, and they are not interchangeable:

Item	IEC world	North America
LPS design and assessment	IEC 62305-1 to -4 (risk assessment -1, physical damage and life hazard -2, surge protection for electrical/electronic systems -3/-4)	NFPA 780 (Standard for the Installation of Lightning Protection Systems)
Component certification	—	UL 96 (lightning protection components) and UL 96A (installation requirements, Master Label)
Air terminal placement method	Rolling sphere, protection angle, or mesh method (IEC 62305-3)	Rolling sphere and side-strike rules (NFPA 780)
Typical rolling sphere radius	20 / 30 / 45 / 60 m for LPS classes I–IV	NFPA 780 commonly works to a 150 ft (45 m) rolling sphere for ordinary structures [Confirm]

Two corrections to the common Chinese-language specification wording, both worth catching before you copy it into an English document:

• "Protection angle ≤ 45°" and "rolling sphere radius 30 m" are two different methods, and you do not apply both. IEC 62305-3 permits the protection-angle method only within height limits; for anything but a very small, simple structure the mesh method (roof conductors on a defined grid, down conductors at defined spacing) is what a designer actually uses. State the method you want, not two partial halves of two methods.
• 30 m rolling sphere radius corresponds to LPS class II in the IEC scheme, not a universal default. Class is an output of the IEC 62305-2 risk assessment — you do not pick it by preference. [Confirm] the risk assessment result for the site.

3.2 Protecting the incoming line, not just the box

The transformer is far more likely to be hit from along the cable or overhead line than from directly above. Two consequences:

• On an overhead MV feeder, the incoming line is the collector. Expect induced and direct strikes on the line to be the dominant exposure, and design the line's shielding / arrester density accordingly.
• Any transition from overhead line to cable is the highest-risk point on the whole installation. Put the arrester at that transition, and put a second arrester at the transformer.

4. Job Two — Clamp the Overvoltage: Surge Arresters on the MV Side

4.1 What the arrester has to achieve

An arrester does not "stop" lightning. It shunts current and holds the voltage at its terminals down to its protective level (residual voltage, Ures, at a stated discharge current and waveshape). Your insulation coordination statement is then:

IEC 60071-2 practice is a minimum margin of roughly 20 % for lightning impulse on equipment insulation — [Confirm] the margin your client's standard or the AHJ requires.

Applied to the numbers in a typical SCB18 specification, for a 10 kV (Um = 12 kV) unit:

Quantity	Value	Where it comes from
System voltage	10 kV (Um = 12 kV) — 12.47 kV or 13.8 kV, 15 kV class, in North America	Project single-line
Winding lightning impulse withstand (BIL)	75 kV (IEC 60076-3 for Um = 12 kV)	Nameplate / type test
Arrester residual voltage Ures	≤ 50 kV [Confirm] at specified In and 8/20 µs	Arrester type test per IEC 60099-4 / IEEE C62.11
Margin	(75 − 50) / 75 = 33 %	Calculation
Maximum protective level allowed at 20 % margin	60 kV	75 / 1.2

So a 50 kV residual on a 75 kV BIL is a comfortable margin — on paper. The next section is where that margin usually gets spent.

4.2 The 20 kV and 35 kV equivalents

Um (kV)	IEC 60076-3 standard LI withstand	North American class	Commonly specified BIL in North America
12 kV	75 kV	15 kV class	95 kV or 110 kV [Confirm]
24 kV	125 kV	25 / 27.6 kV class	125 kV or 150 kV [Confirm]
36 kV	170 kV	34.5 kV class	150 kV or 200 kV [Confirm]

Note the trap for exporters: a 35 kV / 170 kV BIL unit is not automatically acceptable on a 34.5 kV North American system, where a 200 kV BIL is frequently specified. Same voltage class, different insulation expectation, driven by different earthing and exposure practice. Check the BIL the client's single-line calls out before you quote — the [Confirm] values above must be read off the actual project specification and confirmed per IEEE C57.12.01 / NEMA for dry-type.

5. Why "10–15 m From the Transformer" Is Too Far

This is the single most consequential number in the whole topic, and the source specification is wrong in a way that costs transformers.

An arrester clamps the voltage at its own terminals. Between the arrester and the transformer terminals, the surge still has to travel along a conductor. On that run, the travelling wave reflects and the voltage at the transformer can rise well above the arrester's residual voltage. The standard first-order expression is:

Illustrative arithmetic, using deliberately round illustrative inputs ([Confirm] the steepness assumed for your site; it depends on strike distance, line shielding and the presence of other arresters upstream):

Separation d	Conductor	Added voltage 2·S·d/v with S = 500 kV/µs	Voltage at transformer	Against a 75 kV BIL
10 m	Cable (v = 150 m/µs)	+ 66.7 kV	116.7 kV	Fails — 56 % over
10 m	Overhead (v = 300 m/µs)	+ 33.3 kV	83.3 kV	Fails
3 m	Cable	+ 20.0 kV	70.0 kV	Passes, but margin is only 6.7 %
1.5 m	Cable	+ 10.0 kV	60.0 kV	Passes with 20 % margin

Two things follow, and they should both go into the drawing notes:

1. Mount the arrester at the transformer terminals, or on the terminal box if the design allows. A separate arrester gantry 10–15 m away is a different design, and it needs its own calculation to prove it works.
2. The arrester lead itself counts. Lead inductance gives roughly 1 kV of additional drop per metre of lead at lightning-current steepness [Confirm per IEEE C62.22]. A "5 m grounding lead" — which the source specification explicitly allows — can add on the order of 5 kV on top of the residual, and it adds to the effective protective level. Keep leads straight, short and direct; avoid loops and sharp bends, which raise inductance.

If the arrester genuinely cannot be close — a long cable run to a remote switchgear, for example — the standard mitigations are:

• a second arrester at the transformer (the usual answer),
• a surge capacitor at the transformer terminals (typically 0.25–0.5 µF, [Confirm] rating) to flatten the wavefront and reduce S, which also reduces stress on the line-end turns,
• and accepting that the cable's own capacitance provides some damping, which the calculation above ignores.

6. Specifying the Arrester: IEC vs North American Practice

The Chinese-style designation "17/45 kV" (rated voltage 17 kV, residual 45 kV) does not map one-to-one onto a North American catalog number. North American metal-oxide surge arresters are selected by MCOV (maximum continuous operating voltage) and duty-cycle rating, to IEEE C62.11 / NEMA, and the driver is the temporary overvoltage (TOV) the arrester must survive during an earth fault.

That TOV depends on system earthing — and this is the biggest single difference between a typical Chinese 10 kV system and a typical North American 12.47/13.8 kV circuit:

System earthing	TOV during a single line-to-earth fault	Consequence for arrester selection
Unearthed / resonant (arc-suppression coil) earthed — common on Chinese 10 kV and 35 kV networks	Approaches full line-to-line voltage and can persist for hours	Arrester rated voltage must be high; this is why a 17 kV rated unit appears on 10 kV systems [Confirm]
Solidly earthed / multigrounded wye — typical North American 12.47Y/7.2 or 13.8Y/7.97 kV	Line-to-earth voltage, roughly 1.4 × or less	Much lower MCOV acceptable — for a grounded 13.8Y/7.97 kV circuit, MCOV ≈ 8.4 kV [Confirm]
Ungrounded or impedance-grounded North American 13.8 kV	Approaches line-to-line	MCOV ≈ 15.3 kV [Confirm]

Practical instruction: give the arrester supplier three things — system voltage, system earthing arrangement, and the required protective level — and let them select the rating. Do not carry a "17/45" over into an English document as if it were a universal part number; a North American reviewer will read it as a mismatch with the circuit.

Also decide and state:

• Polymer-housed vs porcelain-housed. Polymer is standard for MV distribution now; it behaves much better on failure (no shrapnel) and is lighter. Only specify porcelain where the client insists.
• Disconnector / fault indicator. A failed arrester that stays connected is a permanent earth fault. Specify a disconnector that visibly operates.
• Insulating covers on the arrester and on exposed terminals — cheap, and they stop the small-animal and bird contacts that are a real cause of MV faults.

The IEC 60076 vs ANSI/IEEE comparison covers the wider standards gap that this sits inside; your BIL and insulation levels are printed on the nameplate, and the nameplate reading guide shows where to find them before you write the arrester spec.

7. Job Two (b) — The Low-Voltage Side

Three mechanisms put surges on the 400 V / 480 V side, and they are worth separating because they are mitigated differently:

Mechanism	What actually happens	Mitigation
Capacitive transfer through the transformer inter-winding capacitance	A surge on the MV winding couples to the LV winding. Magnitude depends on the ratio of inter-winding capacitance to LV-side capacitance to earth	LV surge protective device (SPD) at the transformer LV terminals
Inductive coupling to LV circuits running parallel to the struck MV circuit	Induced voltage on the LV conductors themselves	Routing discipline + SPD at both ends of long runs
Ground potential rise	The LV neutral and the LV equipment frames are lifted together with the local earth; the dangerous voltage appears between this site's ground and a remote ground reached by a metallic cable (telecom, control, data)	Equipotential bonding + SPDs on the signal/control circuits, not just the power

A correction worth making: the claim that a strike on the MV line induces "tens of thousands of volts" on the LV side is overstated for the capacitive-transfer case. Transferred overvoltage through a distribution transformer is typically in the low kilovolt range — enough to destroy a PLC, a VFD input stage or a metering card, but not tens of kilovolts. The very high LV voltages occur when the LV line itself is struck directly, or when a ground potential difference appears across equipment. [Confirm] per IEC 62305-4 for the actual coordination levels.

Specification wording that works in both markets:

• IEC markets: SPDs to IEC 61643-11, Type 1 where the structure has an LPS and the SPD carries partial lightning current, Type 2 downstream, coordinated in stages with the required voltage protection level (Up) below the equipment's withstand category (IEC 60664-1 impulse categories I–IV).
• North America: SPDs listed to UL 1449 (4th edition), with the SPD type matched to the location in the NEC (Type 1 field-installed on the line side of the service disconnect, Type 2 on the load side), and a nominal discharge current (In) and voltage protection rating (VPR) appropriate to the panel.
• In both cases: short leads, and coordination between the upstream and downstream SPDs so the downstream unit does not take the whole event.

8. Job Three — Getting the Current Into Earth

8.1 Grounding: what the standards actually require

The familiar "≤ 4 Ω, or ≤ 1 Ω in high-keraunic areas" comes from Chinese practice (GB 50057-family wording). In North America you will be asked different questions, and you should not paste the ohm value into an English specification and expect it to be accepted.

Question	IEC / Chinese practice	North American practice
Electrode resistance trigger	≤ 4 Ω typical; ≤ 1 Ω for lightning-exposed sites [Confirm]	NEC 250.53(A)(2): a single rod/pipe/plate electrode exceeding 25 Ω requires one additional electrode. No "1 Ω" rule for lightning in the NEC [Confirm: AHJ]
System / substation grounding design	—	IEEE 80 (substation grounding, step and touch voltage limits) — design is driven by human safety limits and soil resistivity, not a single ohm target
Measurement	—	IEEE 81 (soil resistivity, fall-of-potential testing); Wenner four-pin method [Confirm]
Power quality / industrial grounding reference	—	IEEE 142 (Grounding of Industrial and Commercial Power Systems)

The right way to write it for an export project: give the soil resistivity survey and the step/touch criteria as the design basis, and give the ohm figure as a target to be confirmed against the AHJ. A fixed number with no soil data behind it is not a design.

8.2 Impulse impedance is what matters, not DC resistance

Here is the point that most specifications miss entirely. Lightning current is not DC. Its energy sits largely below a few hundred kilohertz, and a grounding electrode that measures 1 Ω on a fall-of-potential tester can present tens of ohms to a lightning current, because:

• the conductor's inductance dominates at fast rise times (again, roughly 1 kV per metre of lead at lightning-current steepness), and
• a horizontal or vertical electrode has an impulse effective length — beyond that length, adding more conductor does not reduce the impulse impedance, because the far end simply does not participate during the wavefront.

The design consequences:

• Mesh beats length. A grid with many shorter electrodes, bonded at every crossing, outperforms one long radial. A 5 m × 5 m mesh is a sensible starting point; IEEE 80 grid design then tells you the conductor size, mesh spacing and rod depth needed for the step/touch limits.
• Keep the arrester's earth lead short and straight — this is the same inductance argument as in section 5, and it is the reason the "lead length ≤ 5 m" in the source specification should be tightened to as short as physically possible, ideally under 1.5–2 m.
• Bond, do not daisy-chain. The arrester earth, the transformer enclosure, the core and clamp earthing, the LV neutral bond and the room's earthing bar all land on the same equipotential network, individually and directly.

Conductor sizing equivalents, for converting a Chinese metric spec into a North American one:

Source specification	Approximate North American equivalent
Copper earthing lead ≥ 25 mm²	≈ 3 AWG copper [Confirm]
Galvanised flat steel ≥ 40 mm × 4 mm (160 mm²)	≈ 300 kcmil copper equivalent [Confirm]; note that bare copper in direct contact with galvanised steel needs a bi-metallic transition
Copper tape ≥ 25 mm²	≈ 3 AWG equivalent cross-section, but tape behaves better at high frequency than a round conductor of equal area [Confirm]

8.3 Equipotential bonding is the real fix for "back-flashover"

"地电位反击" (back-flashover / back-strike) is usually described as a grounding-resistance problem. It is mostly a bonding problem. When lightning current enters the earth, the local ground rises with it. Nothing breaks unless a piece of equipment is connected to a distant ground at a different potential — through a data cable, a control cable, a metallic pipe or a signal line.

So the countermeasures, in order of effectiveness:

1. Bond everything in the room to one network — transformer frame, core and clamp earthing, arrester earth, LV neutral, switchgear frames, cable trays, cable armour, the building's earthing system. IEC 62305-4 calls this the equipotential bonding network; the NEC calls it bonding per Article 250.
2. Put SPDs on every metallic service that leaves the room — power, control, telecom, instrumentation. Signal SPDs are the ones that get forgotten and the ones that fail first.
3. Route LV and control cables away from the down conductors and away from the earthing leads that will carry lightning current.
4. Then, and only then, worry about lowering the resistance.

8.4 Cable sheath and shield bonding — one correction

The instruction to earth the MV cable's metallic sheath or shield at both ends needs a qualification. For single-core MV cables, both-ends bonding creates a closed loop that carries circulating sheath current whenever load flows. That current is wasted energy and it forces a de-rating of the cable. Standard practice is:

Bonding arrangement	When it is used
Single-point bonding	Short runs; sheath voltage limiter (SVL) at the free end to clamp the induced voltage. No circulating current.
Cross-bonding	Long runs (typically above about 1 km per section [Confirm]); sheath transposed in sections so induced voltages cancel while the loop is broken in the circulating-current sense.
Both-ends bonding	Short runs, or where the sheath is also the earth-fault return path; accept the loss and de-rate.

The device to specify is a sheath voltage limiter (SVL) — 电缆护层保护器 in Chinese — not a generic "cable protector". And note the North American difference: MV distribution cable in North America more often uses a concentric neutral (a layer of copper wires) rather than a metallic sheath or tape screen, and the concentric neutral is normally earthed at both ends and sized as the fault return path. Write the bonding requirement for the cable type you actually have. [Confirm] the bonding study for any run where the cable is single-core and more than a few hundred metres.

9. Special Situations

9.1 Outdoor installations

A cast resin transformer outdoors needs an enclosure that does three things at once, and they pull against each other:

• Keep water off the winding. Rain ingress is the dominant environmental failure mode for an outdoor dry-type unit. The enclosure needs a real roof, gasketed doors and, critically, cable entries that seal.

IP ratings are not outdoor ratings. IP23 (the "rain" figure in many dry-type datasheets) only covers water sprayed at up to 60° from vertical — it is not a weatherproof rating, and IP20's second digit (0) means no water protection at all. For a genuinely exposed site you want IP54 or, in North American terms, a NEMA 3R enclosure as the minimum. See our outdoor enclosed cast resin units for what that enclosure has to include, and remember that a sealed enclosure restricts cooling — so the derating, the ventilation and the surge protection have to be designed together. The same three-way tension — weatherproof, cool enough, and close-coupled to its arresters — is what drove the layout on our 50 MW solar PV step-up transformer project, where every unit sits fully exposed.

• Anti-condensation heaters. A cast resin unit that sits idle in a humid climate will pull moisture onto cold windings. Heaters that run when the unit is de-energised are standard on outdoor and coastal installations, and they are a lightning-adjacent issue: a damp winding is a winding with a lower impulse withstand.
• Arrester insulating covers and weather-protected terminations, to stop a wet, dirty insulator flashing over during the very event it is supposed to survive.

9.2 Altitude above 1 000 m

Air insulation gets weaker as air density falls. IEC 60071-2 gives the altitude correction factor:

Altitude H	Correction factor Ka	Meaning
1 000 m	1.00	No correction
2 000 m	1.13	External clearances must be ~13 % larger
3 000 m	1.28	~28 % larger
4 000 m	1.45	~45 % larger

Two clarifications before you write "high-altitude arrester" into a spec:

• The arrester's internal varistor blocks are sealed; altitude does not change them. What altitude affects is the arrester's external clearance and creepage, its housing, and the transformer's own clearances. "高原 arrester" usually means a unit with increased external creepage and a larger housing — [Confirm] the manufacturer's altitude rating and the pollution class together, because they interact.
• The transformer also needs the cooling altitude correction (IEC 60076-2 for oil-immersed; for dry-type, the manufacturer's derating table), which is a separate calculation from the insulation one. Both apply at the same time.

9.3 High-keraunic (high lightning) areas

Where the isokeraunic level is genuinely high, the cost-effective additions are:

• More arresters, more often along the MV line, not just at the transformer; this reduces the steepness S that reaches the equipment, which is the term that actually breaks windings.
• Externally gapped line arresters (EGLA / series-gap arresters) on the line — the series gap keeps the arrester out of circuit in normal service and gives a very long service life under repeated strikes. [Confirm] suitability for the line and the earthing arrangement.
• Lightning counters (雷电计数器) on the arrester earth leads. Cheap, and they turn "we think we have a lightning problem" into "we had 14 events on this feeder last year" — which is the evidence you need to justify the next budget.
• Shielding wire on the final span(s) approaching the substation, plus a low-impedance down conductor at the terminal tower.

10. Standards Map

Function	IEC / international	North America
Overall lightning risk and LPS	IEC 62305-1 (risk), -2 (physical), -3 (LPS), -4 (internal systems)	NFPA 780; UL 96 / UL 96A
Insulation coordination	IEC 60071-1 (definitions), IEC 60071-2 (application, margins, altitude)	IEEE C62.22 (arrester application), IEEE 1313 series [Confirm]
Surge arresters	IEC 60099-4 (metal-oxide surge arresters for AC systems)	IEEE C62.11 (standard for MOV surge arresters)
LV surge protective devices	IEC 61643-11 (SPDs connected to LV power systems)	UL 1449 (4th edition); NEC Article 285 [Confirm]
Equipment impulse withstand categories	IEC 60664-1	UL 1449 VPR; NEC
Transformer insulation levels	IEC 60076-3 (insulation levels, dielectric tests, clearances)	IEEE C57.12.00 / IEEE C57.12.01 (dry-type); NEMA
Transformer impulse test	IEC 60076-3 — lightning impulse is typically a type test at distribution voltages, not a routine test [Confirm]	IEEE C57.12.90
Grounding / earthing	IEC 62305-3 (LPS earthing), HD 60364 / IEC 60364-5-54	NEC Article 250; IEEE 80 (substation), IEEE 81 (measurement), IEEE 142 (industrial)
Dry-type transformer construction	IEC 60076-11 (cast resin; F1 fire, C2 climate, E2 environmental classes)	UL 1561; IEEE C57.12.01
Cables	IEC 60502-2 (MV power cables)	UL 1072 / ICEA S-93-639 (MV shielded power cable) [Confirm]

11. A Short Commissioning and Records Checklist

Insulation coordination is only as good as the evidence that it was built as designed. Ask for, and keep:

1. The BIL and power-frequency withstand from the transformer type test report (IEC 60076-3), not from the brochure. On a distribution-class unit the lightning impulse test is normally a type test on the design, not a routine test on your unit — so request the report explicitly.
2. The arrester type test certificate (IEC 60099-4 or IEEE C62.11) showing the residual voltage at the stated discharge current and waveshape.
3. The separation distance calculation — arrester-to-transformer distance, assumed steepness, resulting voltage at the transformer terminals, and the margin against BIL.
4. The soil resistivity survey and the grounding design basis (IEEE 80 step/touch, or the local equivalent), plus the measured resistance and the test method (IEEE 81 fall-of-potential).
5. Photographs of the arrester lead routing before the terminal box is closed — the shortest, straightest lead is the one that will actually perform, and it is the detail most often lost on site.
6. The bonding schedule: what is bonded, to what, with what conductor size.
7. SPD inventory and coordination: type, location, In, VPR / Up, and the coordination between stages.
8. Routine test records for the unit itself. Every unit should have passed its factory routine tests before it shipped — the routine testing guide lists what those are and which of them do not cover impulse withstand.
9. Lightning event log: counter readings, plus any operation of the arrester disconnectors. Review annually, and after any visible strike.

12. Where the Money Actually Goes

For a distribution-class cast resin transformer, the surge protection package — arresters, SVLs, SPDs, bonding conductors, insulating covers and earthing upgrades — is a small fraction of the transformer's own price. What is expensive is the failure: a punctured cast resin winding is not repairable on site, a replacement unit is a manufacturing lead time away, and the outage cost is usually one to two orders of magnitude above the cost of the protection that would have prevented it.

The inverse is also true, and worth stating plainly because it is where overspecification happens: adding more arresters beyond the ones at the cable terminations and the transformer terminals has rapidly diminishing returns. If the separation distance is short and the margin is above 20 %, the next increment of risk reduction comes from bonding discipline and cable routing, not from buying a bigger arrester.

If you are still at specification stage, work from the products range and the resource library to lock the insulation levels and the arrester ratings together — those two decisions have to be made in the same document, by the same person, on the same day. The project references show how the same coordination was resolved on sites that are already energised.

---

13. Frequently Asked Questions

Does SCB18 need different lightning protection from SCB13 or SCB14? No. The loss codes (10, 11, 13, 14, 18) describe no-load and load loss under GB 20052 [Confirm] and have no relationship to impulse withstand. What matters is the declared BIL, the system earthing arrangement, the separation distance between the arrester and the terminals, and the grounding. Two units with identical BILs need identical protection regardless of the series code.

What BIL should I specify for a 10 kV / 12 kV class cast resin transformer? IEC 60076-3 gives 75 kV lightning impulse withstand for Um = 12 kV, and that is what most cast resin units in this class carry. In North America, a 15 kV class dry-type is frequently specified at 95 kV or 110 kV BIL [Confirm per IEEE C57.12.01 and the project single-line]. Do not assume an IEC-rated unit is automatically acceptable on a North American circuit — confirm the BIL before quoting.

How close to the transformer does the arrester have to be? At the terminals, if you can. The voltage at the transformer is approximately U_residual + 2 × S × d / v, so at a steepness of 500 kV/µs and a 10 m cable run, roughly 67 kV is added on top of the residual — which blows straight through a 75 kV BIL. At 1.5 m, the addition is about 10 kV and the coordination holds with margin. Keep the arrester lead itself under about 1.5–2 m as well.

Is a 17/45 kV arrester correct for a 10 kV system? On a Chinese 10 kV network that is unearthed or resonant-earthed, a 17 kV rated arrester is the conventional choice, because the temporary overvoltage during an earth fault approaches line-to-line and can persist [Confirm]. On a solidly earthed North American 12.47Y/7.2 or 13.8Y/7.97 kV circuit, the TOV is far lower and a much lower MCOV is selected to IEEE C62.11. Specify the system, the earthing and the required protective level — let the arrester manufacturer pick the rating.

Do I need surge protection on the low-voltage side as well? Yes, and it is usually the cheapest part of the package. Transferred overvoltage through the inter-winding capacitance is typically in the low kilovolt range [Confirm per IEC 62305-4] — not enough to threaten the transformer, but more than enough to destroy VFD inputs, metering, PLCs and UPS control boards. Specify IEC 61643-11 Type 1/Type 2 SPDs for IEC markets, or UL 1449-listed SPDs with the correct type for the NEC location, coordinated in stages.

Is 4 Ω (or 1 Ω) the right grounding target? In North America, no — that is a Chinese code number. The NEC trigger is 25 Ω for a single rod or plate electrode, with a second electrode required if it is exceeded, and substation grounding is designed against IEEE 80 step and touch voltage limits using measured soil resistivity (IEEE 81). Give the soil data and the safety criteria as the design basis and treat any fixed ohm figure as [Confirm: AHJ].

Why does a 1 Ω grounding electrode still fail under lightning? Because lightning is not DC. At fast rise times the electrode's inductance dominates, and beyond the impulse effective length the far end of a long electrode simply does not participate. A grid of many shorter bonded electrodes, with short straight leads to the equipment, performs far better than a long radial that reads beautifully on a fall-of-potential tester.

Should the MV cable sheath be earthed at both ends? For three-core cables, generally yes. For single-core MV cable, both-ends bonding creates circulating sheath current and forces a cable de-rating; single-point bonding with a sheath voltage limiter at the free end, or cross-bonding on long runs, is standard instead. North American MV distribution cable often uses a concentric neutral rather than a metallic sheath — write the requirement for the cable you actually have. [Confirm] with a bonding study.

Do I need a "high-altitude" arrester above 1 000 m? The sealed varistor blocks are not affected by altitude. What changes is the external clearance and creepage, so above 1 000 m apply the IEC 60071-2 altitude correction (Ka = 1.13 at 2 000 m, 1.28 at 3 000 m) to external clearances, and confirm the arrester's creepage and pollution class together with the transformer's own clearance correction and its cooling derating.

Is a cast resin transformer more or less vulnerable to lightning than an oil-filled one? Different failure mode, not simply better or worse. Cast resin insulation is non-self-restoring, so a surge puncture is permanent, and the lower capacitance to earth makes the initial voltage distribution more non-linear and stresses the line-end turns more. Oil-paper insulation recovers better from a partial discharge event but brings fire and environmental consequences. In practice the deciding factor is not the medium — it is whether the arrester is close, correctly rated and well bonded.

What should I send a supplier to get a surge protection package quoted? System voltage and earthing arrangement; transformer BIL and winding configuration; single-line showing the arrester location and the distance to the transformer terminals; cable type, length and whether it is single-core or three-core; soil resistivity or a Wenner survey; the applicable standard set (IEC 62305 or NFPA 780 / IEEE C62 / NEC); and the local isokeraunic level if you have it. Without the single-line and the distance, any quote is a guess.

---

14. Quick Answers (AI-Friendly Summary)

Lightning protection for a cast resin dry-type transformer has three jobs: intercept the strike (IEC 62305-3 or NFPA 780), clamp the overvoltage (IEC 60099-4 or IEEE C62.11 arresters), and dissipate the current without a dangerous ground rise (IEC 62305-4 bonding, IEEE 80 / NEC 250). The most common failure is not a missing arrester — it is an arrester too far from the transformer, because the voltage at the terminals rises by roughly 2 × S × d / v above the arrester's residual voltage. The second most common is a long arrester earth lead, which adds roughly 1 kV of inductive drop per metre. The "18" in SCB18 is a loss code under GB 20052 and has nothing to do with impulse withstand; the number that matters is the BIL on the nameplate.