Automotive Assembly Plant Transformer: How to Size a 150,000-Unit/Year Car Plant, Why Spot Welding and Press Starts Decide the Substation

The answer first: a 150,000-unit/year car plant needs about 23,200 kVA, and the body shop and press line — not the total — decide the substation

A passenger-car assembly plant with 42,580 kW of connected load does not need 42,580 kW of transformer capacity. Apply area demand factors — 0.35 on the press line, 0.42 on the body shop, 0.78 on the paint shop — and the peak lands at 21,330 kW, which at an effective power factor of 0.918 is 23,236 kVA. That is two 25 MVA 33/11 kV main transformers running split-bus at 46% each and 93% on loss of one, feeding twelve unit substations with 38,400 kVA installed so that every process area survives a single transformer outage. Two loads then override the arithmetic. The body shop is 31% of peak demand and swings 3.5 MVA in under two seconds on every weld cycle, so it gets four dedicated 3,150 kVA cast-resin units, not a share of a general board. And the 2,500-tonne press produces a 6.5% voltage dip on every stroke, which is enough to drop robot controllers and contactors on the same board — so it gets its own 11/0.69 kV pair. Get those two right and the transformer package costs about 6% of plant capex and runs for 25 years; get them wrong and you spend the first year chasing nuisance trips instead of building cars.

Key takeaways

• Size on demand, not nameplate: 42,580 kW connected → demand factor 0.50 → 21,330 kW → 23,236 kVA. Summing the connected load would have specified 49,900 kVA and bought 115% too much iron.
• Body shop load is 6,636 kW but its cyclic behaviour is what matters: 420 medium-frequency welding guns pull the section from 1.2 to 3.5 MVA within two seconds, roughly 60 times a minute. Specify four 3,150 kVA units with K-factor and thermal-cycling data, not one large unit.
• A 400 kW press motor at 6× inrush produces a 6.5% dip on a 2,500 kVA 6% transformer — and only 1.2% at the 11 kV bus. The dedicated transformer is not about kVA, it is about confining a dip to the one board that causes it.
• Medium-frequency DC welding runs at power factor 0.92–0.96 with THDi under 5%; a legacy 50 Hz AC welding shop runs at 0.40–0.55 with 25–40% THD. Specifying K-13 cast resin everywhere "because it is a welding plant" adds 10–15% to the package for no electrical benefit.
• At 4,600 productive hours a year, S13 over S11 saves about USD 5,900 of capitalised loss per 3,150 kVA unit against an USD 8,000–9,000 price premium — it does not pay back. For the 25 MVA mains, keep both energised: two units at 46% lose 79 kW where one at 93% loses 108 kW, i.e. USD 22,400 a year.

What electrical load does a 150,000-unit/year assembly plant actually present?

An assembly plant is five different electrical businesses on one site: a metal-forming shop, a welding shop, a chemical process (paint), a light assembly operation, and a utilities plant that is larger than most factories. Each has its own demand factor, its own power factor and its own power-quality signature. Sizing starts by tabulating them separately, because averaging them into a single number destroys exactly the information the transformer spec depends on.

Three observations come out of that table. First, the overall demand factor is 0.50, and it is 0.50 for a structural reason: no single area is ever at its own peak at the same time as the others. The press line peaks during the shift start and again before body-in-white buffer replenishment; the paint shop runs continuously at 78% of connected load because ovens and abatement cannot be cycled; the battery test bays peak when a batch of packs is released. Second, the two largest contributors by peak kW are the body shop (31%) and the paint shop (18%), but the paint shop is a steady load and the body shop is a pulsed one — they need different transformers even at the same kVA. Third, effective power factor is 0.918, not 0.85: modern MFDC welding and AFE drives are far better than the 1980s AC welding plant that most transformer sizing rules of thumb were written for.

The kVA total is simply the arithmetic: 21,330 kW ÷ 0.918 = 23,236 kVA, plus 5% for the inevitable press-shop upgrade gives 24,400 kVA. If you want the same arithmetic done for a different plant size, the power system calculators on this site run the demand-factor build-up and the transformer kVA in one step; the general method is set out in transformer sizing and budget in one formula.

Why does the body shop, not the paint shop, decide the transformer specification?

Because kVA is the wrong unit for a welding shop. A body shop with 420 guns does not draw 6.6 MW steadily; it draws it in 0.15-second bursts, and the transformer sees a load that steps between roughly 35% and 100% of its rating dozens of times a minute. Three things follow: the winding sees thermal cycling that a steady-load transformer never sees, the voltage regulation is dominated by the burst rather than the average, and the harmonic spectrum depends entirely on whether the guns are AC or MFDC.

How much load does one welding gun actually represent?

A medium-frequency DC gun is rated around 130–160 kVA intermittent on a 50% duty cycle, with a weld time of 0.12–0.25 s and 40–70 welds per minute per gun in a high-rate line. The nameplate kVA cannot be summed: 420 guns × 160 kVA would be 67 MVA. The realistic calculation uses simultaneity — typically 18–25% of guns welding at any instant on a balanced line — and then applies the line's own diversity. That is how 13,400 kVA of installed welding capacity becomes 6,636 kW of demand. The ratio matters at RFQ stage: ask the body shop integrator for simultaneity and duty cycle, not installed kVA, or you will be quoted a transformer three times too large.

MFDC versus 50 Hz AC welding: why the transformer spec changes

The practical consequence is money. A K-13 cast-resin unit carries a 10–15% premium over a K-4 unit of the same rating because the winding has to be designed for extra eddy-current loss. On a body shop of four 3,150 kVA units, that premium is USD 55,000–70,000 spent for a harmonic spectrum the plant does not have. On the other hand, a greenfield plant that inherits a used AC welding line, or a brownfield plant adding guns to an existing 1990s installation, absolutely does need K-13. The selection logic is in K-factor transformer selection, and the underlying loss mechanism is in why harmonic currents overheat transformers.

Why four 3,150 kVA units instead of one 12,600 kVA unit?

The body shop demand is 6,636 kW, which at pf 0.95 is 6,985 kVA. One 8,000 kVA unit would run at 87% and lose the whole body shop on a single fault — in a plant where a body shop stop costs a full shift to restart the line buffer. Four 3,150 kVA units on two bus sections, tie open, give 55% normal loading per unit and 111% on loss of one unit in a section. That 111% is acceptable because welding load is cyclic, not continuous: the ten-minute emergency overload permitted by IEEE C57.91 covers it, and the line can drop to 80% throughput until the unit is swapped. The same logic — many smaller units, split bus, open tie — is why data centres size unit substations this way rather than buying one large transformer per hall. Redundancy in a car plant is a throughput decision, not an electrical one.

What does cyclic loading do to winding life?

A welding transformer's hot-spot temperature tracks the weld cycle with a time constant of minutes, not seconds, so the daily peak is what accumulates insulation ageing. At 55% average loading with bursts to 100%, the hot spot sits 15–20 K below its continuous rating for most of the shift and touches the limit only during high-rate production. That is healthy. What is not healthy is specifying a unit at 85% average loading with bursts to 140%, which is what happens when a body shop is added to a general-purpose board sized on total plant kVA. Overload capability and hot-spot limits are covered in how much you can load a transformer; for the body shop, design to 60% average so that the bursts fit inside the cyclic rating rather than above it.

How much harmonic distortion does a modern car plant inject — and what does it cost in kVA?

About 37% of the peak load is converter-fed: 3,150 kW of chillers, 1,872 kW of compressors, roughly 1,900 kW of paint-shop AHUs and oven fans, and 900 kW of conveyors. If all of those are six-pulse drives, each contributes 30–40% current THD, and the plant-level distortion at the 11 kV point of common coupling lands at 12–14% — roughly three times the IEEE 519 limit for a plant of this stiffness.

The limit depends on the ratio of available short-circuit current to load current. With one 25 MVA, 10.5% transformer in service the 11 kV fault level is 238 MVA, or 12.5 kA, and the load current at 23,236 kVA is 1,220 A, giving I_SC/I_L = 10.2. That places the plant in the 10–20 band:

The cheap fix is at the drive, not at the transformer. Eighteen-pulse front ends on the two largest chillers and active front ends on the four air compressors take the plant from 13% to under 5% for about 3–4% of the electrical package cost. Adding 7% detuned reactors on the remaining six-pulse drives handles the resonance risk, and two 600 A active filters at the paint-shop LV board trim the residual. Buying a bigger transformer instead does not fix distortion — it lowers the impedance and therefore raises harmonic currents, which is the opposite of what the specification needs.

How much nameplate kVA do harmonics actually remove?

Under IEEE C57.110, a transformer loaded with a distorted current must be derated so that the extra eddy-current loss in the winding does not push the hot spot past its limit. For a typical six-pulse spectrum at 35% THDi, the permissible load current is about 0.88–0.90 of rated, so a 3,150 kVA unit delivers roughly 2,800 kVA of usable capacity. That is a 350 kVA loss — more than the entire battery-pack substation's spare capacity. Two ways to recover it: reduce the distortion at source, or specify a K-factor unit whose winding is designed for the harmonic loss. The second is more expensive and only makes sense when the drives cannot be changed, which is typical of a brownfield line extension.

Where does the K-factor actually belong in a car plant?

Does the press line need its own transformer? The voltage-sag arithmetic

Yes, and the reason is measurable in about four lines of arithmetic. Take the 400 kW main drive of the 2,500-tonne press: at 690 V, 0.95 efficiency and 0.90 power factor it draws 391 A full load. Direct-on-line starting at six times rated current with a starting power factor of 0.35 gives a starting apparent power of 2,804 kVA. On a 2,500 kVA transformer with 6% impedance, the dip is:

ΔU ≈ U_k × (S_start / S_N) × cos(φ_SC − φ_start) = 6% × (2,804 / 2,500) × 0.96 ≈ 6.5%

A 6.5% dip lasting 1.5–2 s is inside the ride-through of a well-specified robot controller but outside that of most contactors, which drop out at 70–80% residual voltage, and outside the tolerance of many CNC and vision systems. In practice a press start on a shared board produces exactly the failure mode that car plants are famous for: three robots fault out, the line stops, and 40 minutes are lost restarting and re-homing. On its own transformer the same start produces 6.5% at the press board and only 1.2% at the 11 kV bus, because the dip divides by the ratio of transformer impedances — 2,804 kVA against a 25 MVA, 10.5% source is a small disturbance to the rest of the plant.

What about flicker?

Press lines cycle at 10–20 strokes per minute, which puts the disturbance in the 0.17–0.33 Hz band where the human eye is most sensitive to lamp flicker. Using the IEC 61000-3-7 assessment method, a 6.5% step at 12 strokes per minute gives a short-term flicker severity P_st of roughly 0.8–1.1 against a planning level of 1.0 — borderline. Two engineering answers exist: a flywheel motor-generator set, which is the traditional solution and still the most robust, or a regenerative drive with an active front end, which holds the DC link and lets the press return braking energy to the 690 V board. Both confine the disturbance; neither is needed if the press shares a transformer with the body shop, which is the mistake to avoid.

Why 690 V rather than 400 V for the press line?

At 690 V the press motors draw 42% less current than at 400 V for the same power, so cable sizes drop by roughly half and the fault level at the press board falls from 60 kA to 35 kA — 50 kA switchgear instead of 65 kA. The trade-off is that 690 V is not a standard utilisation voltage everywhere: it is normal in Europe and in plants built to IEC practice, less common in North America where 575 V or 600 V applies. If the plant is destined for a 480 V market, keep the press at 480 V and accept the larger cables; do not import a 690 V press board into a 480 V plant and add a transforming section.

Where must you use dry-type instead of oil-immersed?

Three places in a car plant, for three different reasons. The paint shop needs dry-type because it is a hazardous area with a flammable solvent load and an oil-filled transformer inside the building is a fire-load and insurance problem. The body shop and final assembly need dry-type because the substations sit inside the production building, where an oil-filled unit requires a fire-rated compartment, a bunded oil pit and blast venting — civil works that cost more than the transformer. The rest of the site — stamping, utilities, the 33/11 kV yard — should be oil-immersed outdoors, because oil cools better, costs less per kVA and is the right answer when you have the space.

Inside the paint shop the classification itself drives the layout. Solvent handling, gun cleaning and the flash-off zone are normally Zone 1 or Zone 2 to IEC 60079-10-1, with the extent of the zone determined by ventilation rate and solvent vapour pressure. The transformer does not have to sit in the zone — put the substation in a pressurised electrical room at the end of the building, with the room maintained at positive pressure from a clean air intake and the classification reduced to non-hazardous. What you must not do is install an oil-immersed unit in a corridor next to the booth because it was cheaper. Cast resin at IP54 with a 100 K temperature rise, forced-air cooled to 125% for the oven start-up surge, is the standard answer; the fire-rated separation and the insurance position both improve, and the comparison is set out in oil-immersed versus dry-type transformers.

Does the paint shop's steady load change the sizing rule?

Yes, in the opposite direction to the body shop. Paint-shop load is 78% of connected and essentially flat for 6,000 hours a year, because ovens and RTO abatement cannot be switched off without losing the shift. That means the transformer runs near its thermal limit continuously, so the loss evaluation matters much more than the price, and the cooling class matters more than the K-factor. Two 4,000 kVA units at 53% each, AN/AF, with the AF stage triggered at 60% winding temperature, give the paint shop N-1 capability at 106% on one unit — achievable only by deferring one curing oven, which is a production decision the plant manager should sign off at design stage rather than discover at commissioning.

How do you size the 33/11 kV mains for N-1 without oversizing?

Two transformers, split bus, tie normally open, each carrying half the plant. At 23,236 kVA demand and 25 MVA per unit, normal loading is 46.5% and single-unit contingency is 93% — inside the 1.0 pu continuous rating with no load shedding. That is the whole design. It is worth contrasting with the common alternative of one 31,500 kVA unit: it would run at 74%, lose the entire plant on a fault or on OLTC maintenance, and cost more in losses than the two-unit arrangement.

The numbers that matter at the main-transformer level are:

• Impedance 10.5%. High enough to hold the 11 kV fault level at 12.5 kA per unit so that 25 kA switchgear is adequate, low enough that voltage regulation at full load stays under 4%. A 8% unit would push the fault level to 15.6 kA and force 31.5 kA switchgear across every 11 kV board — a much larger cost than the transformer saving.
• Vector group YNd11. Star on the 33 kV side gives a neutral for earthing and a convenient point for zig-zag earthing transformers on the 11 kV side; delta on 11 kV blocks zero-sequence from the plant. The choice and its consequences are in Dyn11 versus Yyn0.
• OLTC ±8 × 1.25% on the 33 kV side. Car-plant voltage is set by welding and robot controllers that tolerate ±10%, but the utility feeder may vary ±6% and the plant load swings 40% between shifts. On-load tap changing on the HV side keeps the 11 kV bus inside ±2% without touching the unit substations.
• ONAN/ONAF cooling. ONAF gives 125% of the ONAN rating for the shift-start surge and for N-1 operation in summer, without buying a larger core.
• Fault level at 33 kV: 4.2 kA per unit. 25 kA switchgear is more than adequate and leaves room for a third transformer bay in the future.

Should the two mains run in parallel or split?

Split, with the 11 kV tie open and an automatic transfer on loss of one incomer. Parallel operation halves the source impedance, doubles the fault level to 25 kA, requires matched tap positions and impedances, and adds circulating current for no operational benefit in a plant that has no need to share load between the two halves. The conditions for safe paralleling — same ratio, same vector group, same impedance within 10%, same tap position — are set out in transformer parallel operation; the car-plant answer is simply not to rely on them. Transfer time matters: robots and PLCs ride through a 0.3–0.5 s outage, chiller compressors do not restart instantly, so the transfer should be fast (under 0.5 s) and the chillers should be on a staged restart sequence.

What does the EV transition change in the substation design?

Three things, and only one of them is about kVA. First, the battery-pack line adds a dry room: 200–300 kW of dehumidification running 8,760 hours a year, which is a continuous base load that behaves like the paint shop, not like the body shop. Second, the end-of-line DC test bays add 24 × 120 kW of connected charger load at 250–800 V DC, with a demand factor near 0.2 because packs are tested in batches — but each bay's charger front end is a harmonic source, and an 800 V pack tester with a six-pulse front end injects 30% THDi into the assembly board. Third, the pack area wants its own 1,000 kVA unit so that a charger fault does not take out final assembly.

The kVA numbers are modest: 750 kW peak, 789 kVA at pf 0.95, one 1,000 kVA cast-resin unit at 79% loading. The design effort goes into three details that are easy to miss. Specify active front ends on the test-bay chargers so the assembly board meets IEEE 519 without filters. Provide a 400 V DC busbar for the test bays rather than 24 individual long cable runs, because at 120 kW the volt drop and the cable cost both favour a local busbar. And size the dry-room dehumidifier as a continuous load with its own ATS-backed circuit, because a humidity excursion scraps packs — the cost of one scrapped batch exceeds the cost of the redundant feeder.

Which losses should you capitalise at 4,600 hours a year?

Loss evaluation is where car plants differ most from utility practice. A distribution transformer on a utility network runs 8,760 hours near its rating; a body-shop transformer runs 4,600 productive hours at 55% average loading and sits at no-load for the rest. Capitalising no-load loss at utility rates therefore overvalues it, and capitalising load loss at full rating overvalues it badly.

Using capitalisation rates of USD 6,300 per kW of no-load loss and USD 2,700 per kW of load loss, and weighting load loss by the square of the average load factor (0.55² = 0.30), a 3,150 kVA unit compares as follows:

The S13 saves USD 5,857 of evaluated loss and costs about USD 9,000 more, so on a two-shift plant it does not pay back — the opposite of the conclusion you would reach by applying utility load factors. That conclusion holds only where minimum efficiency performance standards do not apply: inside the EU, Regulation 548/2014 as amended by 2019/1783 sets a Tier 2 floor that effectively rules out S11-class losses for a 3,150 kVA unit, so the economics drop out of the decision. In markets without mandatory MEPS the honest answer is that S11 is the right buy for a two-shift body shop, and the money is better spent on AFE drives that recover 350 kVA of harmonic derating. The full method, including how to set the capitalisation rates from your own tariff, is in transformer total cost of ownership.

Why keep both 25 MVA mains energised?

Because splitting the load between two units beats concentrating it in one, even counting the second unit's no-load loss. At 46.5% loading each unit loses 17 kW no-load plus 105 × 0.216 = 22.7 kW load, or 39.7 kW; two units total 79.4 kW. One unit at 93% loses 17 + 105 × 0.865 = 107.8 kW. The saving from running both is 28.4 kW, which at 8,760 hours and USD 0.09/kWh is USD 22,400 a year — and it also halves the voltage regulation and keeps the plant at N-1 without a transfer for any single unit-substation fault. The only reason to run one is during maintenance of the other.

How do you keep robots and PLCs alive through dips?

Ride-through is a system property, not a transformer property. Four measures, in order of cost-effectiveness: put the disturbing load on its own transformer (the press line, done above); keep welding and press boards off the same LV section as robot controllers; specify controllers with at least 200 ms ride-through at 70% residual voltage and prove it with a dip generator at commissioning rather than taking the vendor's word; and add a 10–30 s UPS or a 400 V DC-bus ride-through module only for the PLC racks and safety controllers that cannot tolerate a controlled stop. A dynamic voltage restorer at the 11 kV bus is technically effective and rarely justified in a car plant, because the dips are generated internally and can be removed at source for a fraction of the cost.

One measurement is worth more than any of this: a 7-day power quality survey at the 11 kV PCC and at each unit-substation LV board, before the line is commissioned. It establishes the baseline harmonic spectrum, the actual P_st, and the real dip profile from press starts and weld bursts. Without it, every later nuisance trip is diagnosed by guesswork, and the transformer specification is written from a template rather than from the plant.

What should the transformer specification sheet contain?

The RFQ should be a table, not a paragraph. This is what a car plant's transformer package looks like when it is written correctly:

Two fields are routinely omitted and both cost money. Guaranteed loss figures with the IEC 60076-1 tolerance applied — a 15% tolerance on load loss is worth USD 4,000 of evaluated cost on a 3,150 kVA unit, so ask for the guarantee and enforce it at FAT. And the short-circuit withstanding current, which for a body-shop unit with 60 kA available must be stated explicitly rather than assumed from the rating; the calculation is in transformer short-circuit current.

What are the five specification mistakes that cost the most?

One: sizing on connected load. 42,580 kW becomes a 49,900 kVA order instead of 23,236 kVA. The plant then runs at 47% loading, the customer pays interest on iron they do not use, and no-load loss runs 8,760 hours a year for nothing.

Two: K-13 everywhere because "it is a welding plant". On an MFDC body shop this is USD 55,000–70,000 of winding premium for a spectrum that does not exist. Conversely, omitting K-factor on a legacy AC line costs a transformer in year four.

Three: press line and robots on the same board. The 6.5% dip becomes a 40-minute line stop, several times a week. The fix at design stage is one 2,500 kVA transformer; the fix after commissioning is a DVR or a re-cabling project during a shutdown window.

Four: oil-immersed transformers inside the paint shop or the assembly hall. Fire compartment, bund, blast relief and insurance loading together exceed the transformer price difference, and some insurers will not accept the risk at all.

Five: buying on price with no loss evaluation. The mirror image of mistake one. A 25 MVA main transformer at 105 kW of load loss and 17 kW of no-load loss, run 24/7, burns 1,070 MWh a year; a 10% worse unit burns 107 MWh more, or USD 9,600 a year, which over 20 years is three times the price difference between a cheap and a good unit. The evaluation arithmetic is the same one used in total cost of ownership, and it takes an afternoon.

How do you phase a plant that must never stop?

Build the 33 kV yard for three bays and populate two. Build the 11 kV switchboard with 20% spare bays, because the second body-shop line and the battery expansion both arrive within five years of start of production. Order unit substations as skid-mounted modules — transformer, LV board, protection relay and cooling in one factory-tested enclosure — so that a line expansion is a weekend crane lift rather than a three-week shutdown. Keep one spare 3,150 kVA cast-resin unit on site: the lead time for a replacement is 10–14 weeks, and a body shop cannot wait that long.

Lead times drive the programme more than engineering does. A 25 MVA 33/11 kV unit with OLTC is 16–22 weeks from order to FAT; a 3,150 kVA cast-resin unit is 10–14 weeks; 11 kV switchgear is 12–16 weeks. On a 30-month plant build, that means the transformer order must be placed within four months of design freeze, before the building steel is erected. Everything in this article is cheap to change at design stage and expensive to change after the FAT.

What does commissioning and handover require?

The transformer tests are routine; the system tests are what catch car-plant problems. On the units: insulation resistance, winding resistance, turns ratio on every tap, vector group check, no-load and load loss measurement at FAT, partial discharge for cast resin, SFRA baseline for the mains, and a full OLTC cycle test through all 17 positions. On the system: primary injection of every protection relay, CT ratio and polarity, busbar torque check, thermographic survey at full load, phase-sequence check at each unit substation, and a 72-hour soak at 60% load before the line is released.

Three system-level tests are specific to this plant type and are usually skipped. A staged energisation of the body shop with 20% of the guns active, to confirm the cyclic voltage regulation before full production. A press-start dip measurement with a power quality analyser on both the press board and the 11 kV bus, to prove the 6.5%/1.2% split the design assumed. And a harmonic measurement at the PCC at full production, to confirm the 5% TDD the utility connection agreement requires. If any of the three fails, the fix is still cheap — a filter, a reactor, a tap change — but only if it is found before the plant is handed over.

Sources / 资料来源

Source: IEC 60076-1 — Power transformers: General.
Source: IEC 60076-2 — Temperature rise for liquid-immersed transformers.
Source: IEC 60076-3 — Insulation levels, dielectric tests and external clearances in air.
Source: IEC 60076-5 — Ability to withstand short circuit.
Source: IEC 60076-11 — Dry-type transformers.
Source: IEC 60076-20 — Energy efficiency for transformers.
Source: IEEE C57.12.00 — General requirements for liquid-immersed distribution, power and regulating transformers.
Source: IEEE C57.12.01 — General requirements for dry-type distribution and power transformers.
Source: IEEE C57.110 — Establishing transformer capability when supplying non-sinusoidal load currents.
Source: IEEE C57.91 — Loading guide for mineral-oil-immersed transformers.
Source: IEEE 519 — Harmonic control in electric power systems (current distortion limits at the point of common coupling).
Source: IEEE 1453 — Flicker measurement and limits.
Source: IEC 61000-3-6 — Assessment of emission limits for distorting loads in MV and HV power systems.
Source: IEC 61000-3-7 — Assessment of emission limits for fluctuating loads (flicker planning levels).
Source: IEC 61000-2-4 — Compatibility levels in industrial plants for low-frequency conducted disturbances.
Source: IEC 60079-10-1 — Classification of areas: explosive gas atmospheres.
Source: IEC 62271-200 — AC metal-enclosed switchgear and controlgear for rated voltages above 1 kV and up to and including 52 kV.
Source: IEC 61439 — Low-voltage switchgear and controlgear assemblies.
Source: ISO 669 — Resistance welding: resistance welding equipment ratings.
Source: IEC 62135-1 — Resistance welding equipment: safety requirements.
Source: ISO 12944 — Corrosion protection of steel structures by protective paint systems (C5-M category).
Source: EN 50588-1 — Medium power transformers 50 Hz with highest voltage for equipment not exceeding 36 kV.
Source: Commission Regulation (EU) 548/2014 as amended by 2019/1783 — Ecodesign requirements for transformers.
Source: GB/T 6451 — Technical parameters and requirements for three-phase oil-immersed power transformers.
Source: QDTB Engineering Team — oil-immersed and cast-resin transformer supply to automotive, industrial and infrastructure projects, 2015–2025 (demand factors, simultaneity, dip percentages, price bands, loss capitalisation rates and lead times quoted here are industry experience values, not published standards).