The answer first: a 20,000 TCD sugar-ethanol plant is a 33.5 MW load sitting on a 100 MW power station, and three transformer decisions decide whether it exports 62 MW
A mill crushing 20,000 tonnes of cane per day — 833 t/h for 200 days — burns its own bagasse in two 250 t/h, 85 bar boilers, drives a 60 MW extraction-condensing turbine and a 40 MW back-pressure turbine onto a 13.8 kV bus, and sells 62 MW to the grid at 138 kV. Its own demand is only 33.5 MW at 0.93 power factor, so the transformers on the export path, not the load, set the design. Three decisions dominate the package: a 138 kV bank of 2 × 50 MVA that still carries 96% of peak export on one unit, an 8% impedance on the 13.8/4.16 kV mill transformers that holds a 1,600 kW shredder start to a 7% dip, and 12-pulse groups plus 8 MVAr of detuned filtering that keep current TDD at the 138 kV point of common coupling below the 5% the utility writes into the power purchase agreement. Get those three right and the transformer package costs USD 2.4–2.9 million on a plant that costs a hundred times that; get them wrong and you either cannot deliver the contracted 62 MW or you pay a penalty tariff for twenty seasons. The load, loss and dip arithmetic can be reproduced with the power system calculators.
Key takeaways
- 833 t/h of cane at about 40 kWh/t of milling, process and distillery power is 33.5 MW — but it is the 62 MW export contract, not the 33.5 MW load, that sizes the 138 kV transformers and their tap changers.
- Reverse power flow rewrites the specification: HV-side on-load tap changer, direction-sensitive 32/67 protection, anti-islanding on the 13.8 kV bus, and a no-load loss that is paid 8,760 hours a year instead of 4,800.
- 8% impedance on a 12.5 MVA 13.8/4.16 kV mill transformer gives 21.7 kA of fault current and a 7% shredder-start dip; 6% cuts the dip to 5.2% but forces 31.5 kA switchgear, and 10% is cheap but dips 8.7% and trips the mill's own contactors.
- 20 MW of six-pulse VFDs produces 28% current THD at the drive but only about 12% at the 13.8 kV bus; the 5% limit at 138 kV is bought with paired 12-pulse windings and filters, never with a larger transformer.
- In a plant that runs 4,800 hours and idles 3,960, one kilowatt of no-load loss capitalises at USD 5,600 and one kilowatt of load loss at USD 2,270 — which is why loss evaluation, not purchase price, should decide the award.
What electrical load does a 20,000 TCD mill with an ethanol distillery actually present?
The load is unusually flat. Cane arrives at a constant rate because it must be crushed within 24 hours of cutting, the tandem runs at fixed speed, and the distillery is a continuous process. Between the start of the season and the last bagasse truck, demand moves by less than 8%. That is the opposite of an automotive plant, and it means the transformer duty is a long, steady thermal soak rather than a cyclic one — but it also means the load is present for only 200 days.
| Load group | Connected (MW) | Maximum demand (MW) | What drives the number |
|---|---|---|---|
| Cane yard, unloading and preparation (2 × 1,600 kW shredders, leveller, conveyors) | 3.8 | 3.2 | Shredder tip speed and cane blanket density; both shredders run together |
| Milling tandem, 6 mills × 1,400 kW, VFD | 8.4 | 7.6 | About 12 kWh/t of cane for a 6-mill tandem with turbine drives replaced by electric motors |
| Boiler house: FD/ID fans, feed pumps, bagasse and ash handling | 4.5 | 4.0 | Two 250 t/h units, one on load at a time outside peak crush |
| Juice heating, clarification, evaporators, vacuum pans, centrifugals | 6.4 | 5.2 | Pump and centrifugal motors, largely VFD or soft-start |
| Distillery: fermentation, distillation columns, molecular sieve dehydration, cooling water | 7.6 | 6.8 | 1,100 m³/d of hydrous ethanol at about 6.2 kWh/m³ |
| Water intake, cooling towers, compressed air, effluent treatment | 2.4 | 2.1 | Continuous, weather-dependent in the cooling tower |
| Sugar drying, cooling, bagging, warehouse | 2.0 | 1.6 | Two dryers and four bagging lines |
| Lighting, HVAC, laboratory, workshop, admin | 1.8 | 1.4 | Night shift at 80% of day shift |
| Distribution transformer losses inside the plant | — | 1.6 | Sum of no-load and load losses at the demand above |
| Total | 38.5 | 33.5 MW / 36 MVA | At 0.93 plant power factor |
Two conclusions follow. First, 33.5 MW is small compared with the 100 MW installed behind it, which is why a mill is a net exporter for eight months and a 6.5 MW importer for four. Second, the specific consumption of about 40 kWh/t of cane is the number that gets compared between projects, and it is decided long before the transformers are bought — by whether the tandem is electrified or steam-turbine driven, and by how much of the juice goes to ethanol. The transformer engineer inherits the answer and has to make it work at 4.16 kV with 21.7 kA of fault current.
Why is the 13.8 kV generation bus the spine of the plant, not the 4.16 kV mill bus?
Because 100 MW cannot be moved at 4.16 kV. A 60 MW machine at 4.16 kV would deliver 8,600 A at its terminals, which needs four parallel cables per phase and a generator circuit breaker that does not exist at that rating. At 13.8 kV the same machine delivers 2,600 A and a standard 3,150 A breaker, and the 40 MW back-pressure unit another 1,750 A. Every large load then steps down from the generation bus, and the grid transformer steps up from it.
| Voltage | What lives there | Transformers | Fault level | Earthing |
|---|---|---|---|---|
| 138 kV | Utility interconnection, revenue metering, line protection | 2 × 50 MVA, 138/13.8 kV, YNd11, Z = 12%, HV-side OLTC | Set by the utility, typically 4,000–6,000 MVA | Solidly earthed at the utility substation |
| 13.8 kV | Two generators, GSU bank, six plant feeders, 8 MVAr filter | Generator step-up and the 13.8/4.16 kV and 13.8/0.48 kV feeders | 26 kA with both generators and both GSUs in service | 400 A neutral earthing resistor per generator, 10 s |
| 4.16 kV | Mill tandem, shredders, boiler fans, large pumps | 2 × 12.5 MVA, 13.8/4.16 kV, Dyn11, Z = 8% | 21.7 kA per transformer | Low-resistance earthed, 200 A, 10 s |
| 480 V | Distillery, sugar house, water treatment, MCCs | K-13 dry-type, 2,500 kVA and 2,000 kVA, 13.8/0.48 kV | 42 kA at the secondary terminals | Solidly earthed at each transformer |
| 480/277 V | Lighting, HVAC, instrument supplies | 1,000 kVA dry-type | 25 kA | Solidly earthed |
The 13.8 kV bus is also where the plant's electrical identity is decided. It is the only bus that sees both generation and load, so it is the bus that must survive the loss of the grid at full crush: the generators must pick up 33.5 MW within about two seconds or the whole plant trips, the shredders and mills coast down, and the season's maintenance schedule is rewritten. That is a control and protection problem, but it is also a transformer problem, because the GSU bank is the impedance between that bus and the only other source of power the plant has.
How do you size the 138 kV transformer bank for 62 MW of reverse power flow?
On the numbers, 62 MW at 0.97 power factor is 64 MVA, and a single 50 MVA unit would be overloaded at 128% for the whole season. The bank is therefore two units, and the question becomes what one unit must carry when the other is out. With ONAF cooling a 50 MVA unit is rated 66.7 MVA, so N-1 export is 96% of rating at 40 °C ambient — acceptable for a few weeks of the season, not acceptable as a permanent condition, which is why the export contract also contains a 55 MW firm cap.
| Case | Export or import | Bank loading | What it means |
|---|---|---|---|
| Peak crush, both units in service | 62 MW / 64 MVA | 48% of 133 MVA ONAF | Normal operating point, ONAN cooling is sufficient below 50 MVA total |
| Peak crush, one unit out | 62 MW / 64 MVA | 96% of 66.7 MVA ONAF | Permitted with a 55 MW cap and a 6-hour daily review of top-oil temperature |
| Off-season, maintenance | 6.5 MW import | 7% of one unit | Reverse direction; the OLTC now regulates from the grid side |
| Black start of one generator | 4 MW import through one unit | 6% of one unit | Unit auxiliary transformer path, GSU energised from the grid |
| Reactive support at night | 0 MW, 20 MVAr export | 20 MVA at leading power factor | Tap position and AVR limit decide whether the bank is at risk of over-fluxing |
Reverse flow is where most specifications go wrong, because the transformer itself does not care which way the power goes but everything around it does. Four items change. The on-load tap changer moves to the HV winding, so that a ±10% grid voltage swing is absorbed on the 138 kV side and the 13.8 kV bus stays within 1% — putting the OLTC on the LV side of a GSU is a common import-template error and it also puts the tap changer in the 26 kA fault path. The protection scheme needs 32 (reverse power) and 67 (directional overcurrent) elements with the direction set for export, plus 87T differential, 51/51N backup, 63 Buchholz, 26/49 thermal and 49/38 winding and oil temperature trips; the plant side of that scheme is described in transformer protection devices. The utility's anti-islanding requirement, usually IEEE 1547 or the local grid code, applies at the point of common coupling and is implemented on the 13.8 kV generator breakers, not on the GSU. And the vector group must be YNd11 with the neutral solidly earthed on the 138 kV side, because the utility's earth-fault protection depends on it — the reasoning behind Dyn11 versus Yyn0 choices is set out in vector group Dyn11 and Yyn0.
Parallel operation of the two units deserves its own line in the specification. Two 50 MVA transformers with 12% impedance will share 64 MVA almost exactly if their impedance differs by less than 5% and their ratios match at the same tap, but a 12% unit in parallel with an 11.4% unit takes 51% of the load, and the mismatch grows with export. Ask for impedance tolerance of ±5% of the declared value, matching tap positions, and a factory measurement of both units so that the load sharing can be calculated before they leave the works — the method is in transformer parallel operation. The same 12% impedance that makes the units share well also limits the 13.8 kV fault contribution of each GSU to 17.4 kA, which is what keeps the generator breakers inside their 26 kA rating.
Why does a 1,600 kW shredder start decide the mill transformer impedance?
Because the shredder is the largest motor on the plant that is still started direct-on-line, and the voltage dip it produces is a direct function of transformer impedance. A 1,600 kW, 4.16 kV motor at 0.95 efficiency and 0.93 power factor draws 251 A full load. At a locked-rotor current of 6.0 times full load it draws 1,506 A, which is 10.85 MVA of starting load. A 12.5 MVA transformer with 8% impedance has a short-circuit capacity of 156 MVA, so the dip at its terminals is 10.85 divided by 156, or 6.9%, and about 7.4% once the 13.8 kV source impedance is included.
| Transformer impedance | Short-circuit capacity | 4.16 kV fault current | Shredder-start dip | Cost consequence |
|---|---|---|---|---|
| 6% | 208 MVA | 28.9 kA | 5.2% | 31.5 kA switchgear, larger cable, heavier bracing |
| 8% (specified) | 156 MVA | 21.7 kA | 6.9% | 25 kA switchgear, standard mill motor starters |
| 10% | 125 MVA | 17.4 kA | 8.7% | 20 kA switchgear, but two simultaneous starts dip 17% |
7% matters because of what else is on the bus. At 7% dip the mill motor contactors hold, the VFDs ride through on their DC links, and the plant's own 480 V loads see 6% after the 13.8 kV feeders, which is inside the 10% that IEC 61000-2-4 allows for a few seconds. Above 10% the contactors chatter, the centrifugals trip, and the utility's flicker meter at the 138 kV point records a Pst above 1.0 during the crush, which is a contractual event. Starting both shredders together doubles the starting MVA to 21.7 MVA and the dip to about 14%, so the interlock that permits only one large DOL start at a time belongs in the electrical specification, not in the operating procedure.
The alternative is a VFD on the shredders, which caps starting current at 110% and removes the dip entirely, at the cost of about USD 180,000 per drive plus the harmonic burden described below. Many new mills take that route; the ones that do not must buy the impedance margin, and the cheapest place to buy it is a lower-impedance transformer rather than a bigger one. Choosing between the two is the same exercise as sizing any motor-starting transformer, and it is worth checking the short-circuit withstand of the downstream equipment at the same time — short-circuit current calculation covers the arithmetic.
What harmonics does a sugar mill inject, and where do you stop them?
A mill is a harmonic source with an unusually wide spread of drive sizes, and its point of common coupling is the utility's 138 kV bus, where the limit is written into the interconnection agreement. IEEE 519 applies at that bus with a total demand distortion of 5% in most Brazilian, Thai and Indian interconnection studies for a plant of this size, and 8% only where the short-circuit ratio is above 50.
| Drive group | Rating | Front end | Current THD at the drive | Contribution at 13.8 kV |
|---|---|---|---|---|
| Milling tandem, 6 × 1,400 kW | 8.4 MW | Paired 12-pulse (one Δ, one Y secondary per pair) | 28% per 6-pulse group, 9% per pair | 3.1% |
| Cane preparation, 2 × 1,600 kW shredders | 3.2 MW | Active front end | 4% | 0.6% |
| Distillery pumps, centrifugals, column reboiler pumps | 2.8 MW | 6-pulse, detuned reactor | 30% | 2.4% |
| Boiler FD/ID fans and feed pumps | 1.5 MW | 6-pulse | 25% | 1.0% |
| Filter and capacitor bank, 8 MVAr detuned at 4.08 (5th) and 2.04 (7th) | — | Passive | — | −4.2% (cancellation) |
| Result at 13.8 kV | 15.9 MW of drives | — | — | 2.9% TDD |
| Result at 138 kV PCC | — | — | — | 2.1% TDD against a 5% limit |
Without the 12-pulse pairing and the filter the total at 13.8 kV is about 12%, which attenuates to roughly 4.8% at 138 kV — just inside the limit on paper and certain to breach it at the next expansion. The reason the attenuation matters is that the GSU's 12% impedance and the utility's source impedance are in series with the harmonic currents, so the voltage distortion at the PCC is the current distortion multiplied by the impedance at each harmonic order. That is also why a bigger transformer is the wrong answer: a lower-impedance GSU passes more harmonic current to the grid, not less.
Two side effects of harmonics land on the transformers themselves. The 4.16 kV mill transformers carry 8.4 MW of 12-pulse load, which adds about 4% to their eddy-current loss and 3 K to their winding hot spot; that is absorbed by specifying a 65 K average winding rise and verifying it at the factory with a harmonic-loaded heat run, not by oversizing. The 480 V distillery transformers carry 6-pulse load at K-13, which is where the K-factor actually bites. The general method is in transformer harmonics and K-factor transformer selection.
Why does the distillery need its own 13.8/0.48 kV substation and K-rated transformers?
Because the distillery is a different electrical environment from the sugar house, and mixing them is how plants end up with overloaded neutrals and hot 480 V busbars. The distillery has 2.8 MW of 6-pulse VFD load concentrated in pumps and centrifugals, a high density of electronic instrumentation, and a load that never stops during the season — three shifts, 200 days, at 6.8 MW. Its own 13.8/0.48 kV substation with K-13 dry-type transformers gives it separate fault levels, separate harmonic filtering and a single place to hang the earthing system.
| Transformer | Rating | Type | Why this choice |
|---|---|---|---|
| Distillery main 1 and 2 | 2 × 2,500 kVA, 13.8/0.48 kV | Cast-resin, K-13, IP31, AN/AF | Non-harmonic load is 4 MW; the K-13 covers the 2.8 MW of 6-pulse VFDs without derating |
| Sugar house and bagging | 2,000 kVA, 13.8/0.48 kV | Cast-resin, K-4, IP31 | Mostly motors on soft starters, low harmonic content |
| Water and effluent | 1,600 kVA, 13.8/0.48 kV | Cast-resin, K-4, IP54 | Outdoor installation beside the cooling towers, condensation risk |
| Emergency and black start | 1,250 kVA, 13.8/0.48 kV | Oil-immersed, ONAN, IP54 | Fed from the diesel generator, must carry the boiler feed pumps and instrumentation |
The K-factor is not a marketing number. A standard distribution transformer designed for sinusoidal load has its eddy-current loss rise with the square of the harmonic frequency, so a K-4 unit carrying K-13 load overheats its windings even at 70% nameplate current. The K-13 units above are specified with the harmonic spectrum in the purchase order and their temperature rise verified at K-13, which is the only way to know the hot spot is inside 150 °C at the end of the season. In practice the K-13 premium is 12–18% of the transformer price, and it is cheaper than replacing a failed 2,500 kVA unit in the middle of a crush.
There is also a voltage-quality reason for the separate substation. The mills and shredders produce dips and the distillery produces harmonics; putting them on the same 480 V bus means the distillery's instruments see the mill's dips and the mill's contactors see the distillery's distortion. Separating them at 13.8 kV costs two extra feeders and buys two independent power quality domains, which is worth more than the cable.
How do you keep ethanol vapour and bagasse dust out of the transformer?
By treating the distillery as a classified area and the boiler house as a dirty one, because the two hazards attack the transformer in completely different ways. Ethanol vapour forms a Zone 2 envelope around pump seals, flange joints and the distillation column, extending typically 3 m for Zone 1 and a further 4.5 m to the Zone 2 boundary under IEC 60079-10-1 for a IIA T3 gas. An oil-immersed transformer with a breather, an explosion vent and a pressure relief device cannot be installed inside that envelope; it goes outside the Zone 2 boundary with its radiators clear of the plume, or the substation is pressurised and the transformer is cast-resin, per IEC 60079-13.
| Hazard | Where | Failure mechanism | Specification response |
|---|---|---|---|
| Ethanol vapour, IIA T3 | Fermentation, distillation, dehydration | Vapour ingress through silica-gel breather acidifies oil; LV busbar corrosion | Sealed tank or cast-resin, no breather in the classified area, C5-M paint, IP55 |
| Bagasse fly ash and cane fibre | Boiler house, bagasse yard, mill floor | Radiator fins blind in 2–3 seasons; top-oil rise climbs 8–12 K | Radiator fin pitch 50 mm minimum, vertical fins, removable screens, IP54, top-oil alarm at 85 °C |
| High humidity, 90% RH in the rainy season | Whole plant | Condensation on windings during shutdown; low insulation resistance at restart | Anti-condensation heaters interlocked with the breaker, IP54, dew-point-controlled space heating |
| Molasses and juice spillage | Sugar house, molasses storage | Conductive, corrosive film on bushings and cable boxes | Bushed-phase barriers, cable boxes with drains, epoxy-coated tank top |
| Continuous full load for 200 days | All | Thermal ageing of paper; 6 K over design halves insulation life | 65 K winding rise, ONAN/ONAF, fibre-optic hot-spot probes in the LV winding |
Ambient temperature rarely helps. A mill in São Paulo state sees 38 °C in the crush season with 90% humidity; one in Uttar Pradesh sees 46 °C and a monsoon; one in Chaiyaphum sees 40 °C. A transformer rated at 40 °C maximum ambient and installed between two buildings with 1.5 m clearance is effectively a 45 °C unit, and the derating is 1% per degree above rating for oil-immersed units under IEC 60076-7 loading guidance. That is why the specification should state the ambient and the clearances, and why the heat run at the factory should be performed at the design ambient rather than at 25 °C.
Which losses should you capitalise in a plant that runs 4,800 hours and idles 3,960?
Both, and the split is unusual: in an industrial plant that runs 8,000 hours, no-load loss and load loss capitalise at roughly the same rate. In a mill, no-load loss is paid for 8,760 hours a year while load loss is paid only during the crush, weighted by the square of the load factor. That inverts the usual weighting and makes the no-load figure the one to argue about.
| Loss component | Hours paid per year | Energy price | Annual cost per kW | Capitalised at 8%, 20 years |
|---|---|---|---|---|
| No-load (core) loss | 8,760 | USD 0.065/kWh | USD 569 | USD 5,590 |
| Load loss at 0.86 average load factor | 4,800 × 0.86² = 3,550 equivalent hours | USD 0.065/kWh | USD 231 | USD 2,270 |
| Cooling power to remove 1 kW of loss | As above, at 0.02 kW/kW for ONAN fans | — | USD 11 | USD 112 |
Put those rates on a 50 MVA GSU with 35 kW of no-load loss and 320 kW of load loss and the transformer's real cost appears: USD 196,000 of capitalised no-load loss plus USD 726,000 of capitalised load loss, which together are about 60% of the purchase price of the unit. A competing bid with 45 kW of no-load loss and 380 kW of load loss looks USD 60,000 cheaper on the quotation and costs USD 192,000 more over twenty years — a difference that dwarfs any commercial negotiation over price. This is the argument in total cost of ownership, and it is the reason a loss evaluation formula should be in the purchase order with a penalty of USD 5,600 per kW of no-load loss and USD 2,270 per kW of load loss, measured at the factory test and applied as a price adjustment rather than as a rejection criterion.
There is a second, subtler loss item that matters in mills: the off-season. For 3,960 hours the plant imports 6.5 MW for maintenance and draws it through one GSU at 7% loading, where the transformer operates at its worst efficiency — all core loss, almost no load loss. A mill that installs a 2,500 kVA off-season transformer fed from the 13.8 kV bus and takes the GSU out of service saves the core loss of a 50 MVA unit (35 kW) for four months, which is USD 13,000 a year, and pays back the cost of the extra transformer in about eight years. It also keeps the GSU available for maintenance during the only window in which maintenance is possible.
What does the transformer package cost, and what is the lead time?
The package for a 20,000 TCD mill with a distillery is 10 transformers and about 110 MVA of installed capacity. Prices below are ex-works China, 2026, for IEC 60076 designs with type-tested short-circuit withstand, witnessed routine tests and a loss penalty clause.
| Item | Rating and type | Unit price (USD) | Lead time (weeks) |
|---|---|---|---|
| 138 kV GSU | 50 MVA ONAN/66.7 MVA ONAF, 138/13.8 kV, YNd11, Z = 12%, HV OLTC ±8 × 1.25% | 720,000–880,000 | 26–34 |
| Mill transformers | 12.5 MVA ONAN, 13.8/4.16 kV, Dyn11, Z = 8% | 220,000–270,000 | 18–24 |
| Distillery transformers | 2,500 kVA cast-resin, 13.8/0.48 kV, K-13 | 95,000–120,000 | 14–18 |
| Sugar house and water | 2,000 / 1,600 kVA cast-resin, K-4 | 70,000–90,000 | 12–16 |
| Emergency | 1,250 kVA oil-immersed, 13.8/0.48 kV | 48,000–60,000 | 10–14 |
| Package total | 2 GSU + 2 mill + 6 distribution | 2,400,000–2,900,000 | Driven by the GSUs |
The GSUs are on the critical path and the reason is not the winding: it is the 138 kV bushings, the OLTC and the impulse test. A 138 kV class unit needs lightning impulse testing at 650 kV, switching impulse at 550 kV where applicable, and a tap changer from a supplier with a 20-year service record in the local grid — if the mill buys on price and takes a 22-week delivery from a factory with no 138 kV type test report, the utility's interconnection study will reject the unit and the season is lost. Two practical rules follow: award the GSUs first, before the distillery transformers, because they set the commissioning date; and require a factory acceptance test with the buyer present, including a full loss measurement at rated voltage, a temperature rise run and a noise measurement, which is what pre-shipment inspection is for.
What should the purchase specification say?
The specification should be short, numeric and testable. Anything that cannot be measured at the factory or on site does not belong in it.
| Item | GSU (138 kV) | Mill transformer (4.16 kV) | Why |
|---|---|---|---|
| Rated power and cooling | 50 MVA ONAN / 66.7 MVA ONAF | 12.5 MVA ONAN | N-1 export at 96% and a 7% dip both depend on it |
| Voltage ratio and taps | 138 ± 8 × 1.25% / 13.8 kV, OLTC on HV | 13.8 ± 2 × 2.5% / 4.16 kV, off-circuit | Grid swings absorbed on the HV side; mill bus set once per season |
| Impedance | 12% ± 5% at 75 °C | 8% ± 5% at 75 °C | Load sharing between GSUs; dip and fault level at 4.16 kV |
| Vector group | YNd11 | Dyn11 | Utility earth-fault protection; harmonic cancellation between pairs |
| Insulation level | LI 650 / AC 275 kV | LI 95 / AC 38 kV | IEC 60076-3, standard for the class |
| Temperature rise | 60 K oil / 65 K winding at 40 °C ambient | 60 K oil / 65 K winding | Harmonic and fouling margin for a 200-day season |
| Losses | 35 kW no-load, 320 kW load, guaranteed with penalty | 12 kW no-load, 78 kW load, guaranteed with penalty | USD 5,600 and USD 2,270 per kW capitalised |
| Short-circuit withstand | 2 s at 21 kA symmetrically | 2 s at 21.7 kA | Mill bus faults clear in 0.4 s but the thermal duty is specified at 2 s |
| Noise | ≤ 78 dB(A) at 1 m, ONAN | ≤ 68 dB(A) at 1 m | Residential boundary 150 m from the plant; see transformer noise levels |
| Enclosure and finish | IP55, C5-M, 50 mm radiator pitch | IP54, C4, removable screens | Bagasse fly ash and ethanol vapour |
| Monitoring | Two fibre-optic hot-spot probes, DGA valve, oil and winding temperature trips, OLTC tap position | Winding temperature trip, DGA valve | Condition data for the 4,000 idle hours, when a fault has nowhere to hide |
| Tests | Routine per IEC 60076-1, impulse, temperature rise, noise, zero-sequence impedance, OLTC sequence | Routine, impulse, temperature rise, short-circuit type test report | Factory evidence, not promises |
Frequency needs one line too. Brazil, and therefore most of Latin America's sugar industry, is 60 Hz; Thailand, India, Vietnam and Indonesia are 50 Hz. A transformer designed for 60 Hz and installed on a 50 Hz system sees 20% more flux, saturates, overheats and produces fifth-harmonic current from no load at all. If the mill is a repeat order from a Brazilian group building in Thailand, the design must be re-rated rather than re-plated, and 50 Hz versus 60 Hz sets out the derating. Cooling design also differs: ONAN ratings at 40 °C in Thailand with 90% humidity cannot be copied from a 30 °C Brazilian design, and the difference is worth 8% of rating.
What actually fails in sugar mill transformers?
Five failure modes account for most of the field problems in this industry, and none of them is a manufacturing defect in the winding.
| Failure mode | Typical time to failure | Root cause | Prevention |
|---|---|---|---|
| Radiator fouling by bagasse fly ash | 3–5 seasons | Fin pitch 20 mm, horizontal fins, no screening | 50 mm vertical fins, annual dry-ice cleaning, top-oil alarm at 85 °C |
| OLTC contact wear from daily cycling | 8–12 years | Weak grid, 30 tap operations a day in export mode | Vacuum OLTC, operation counter with alarm at 80% of rated operations, annual DGA |
| Moisture ingress at the breather | 10–15 years | Silica gel exhausted in the rainy season; oil acidity rises | Sealed tank for the distillery units, breather inspection monthly, oil treatment at 0.1 mg KOH/g |
| Hot spot at LV busbar joints | 2–4 years under K-13 load | Harmonic current heating in a K-4 transformer | K-13 rating, torque-checked joints, infrared survey twice a season |
| Generator trip on transformer inrush | Immediate | Re-energising two GSUs and six plant transformers from a 60 MW machine after a mill trip | Staggered energisation, 2nd-harmonic restraint settings verified, inrush current calculated — see inrush current |
The inrush case is the one that ends seasons. A 50 MVA GSU draws 6–8 times rated current for up to 0.15 s when energised, and a 60 MW generator with 12% subtransient reactance sees that as a 300 MVA load for a tenth of a second; the voltage collapses to 60%, the generator's loss-of-excitation or undervoltage protection operates, and the plant goes dark. The fix is a switching sequence programmed into the PLC — GSUs first, one at a time, 5 s apart, then plant transformers in groups of two — and a protection study that confirms the 2nd-harmonic blocking settings on each transformer differential relay. Both are free at the design stage and expensive afterwards.
Loadability during the crush is the other half of the same story. A mill that pushes the tandem to 110% of rated cane rate for a week does not overload the transformer thermally, because the load is flat and the ambient is stable — but it does raise winding temperature by 8–10 K, and paper ageing doubles for every 6 K. That is why the specification asks for fibre-optic hot-spot probes rather than a calculated winding temperature from a thermal image, and why load factor and overload guidance should be read before, not after, the first season. Mills that also sell steam to a neighbouring distillery or a district heating network should read the same numbers through the lens of a plant that never fully shuts down: district heating and CHP transformers covers the 8,760-hour case, and biomass plant transformers the fuel-handling side.
Sources / 资料来源
Source: IEC 60076-1 — Power transformers: general.
Source: IEC 60076-2 — Power transformers: temperature rise for liquid-immersed transformers.
Source: IEC 60076-3 — Power transformers: insulation levels, dielectric tests and external clearances in air.
Source: IEC 60076-5 — Power transformers: ability to withstand short circuit.
Source: IEC 60076-7 — Power transformers: loading guide for mineral-oil-immersed power transformers.
Source: IEC 60076-11 — Power transformers: dry-type transformers.
Source: IEEE 519 — Recommended practice and requirements for harmonic control in electric power systems.
Source: IEEE C57.110 — Recommended practice for establishing transformer capability when supplying non-sinusoidal load currents.
Source: IEEE C57.91 — Guide for loading mineral-oil-immersed transformers and step-voltage regulators.
Source: IEEE 1547 — Interconnection and interoperability of distributed energy resources with associated electric power system interfaces.
Source: IEC 60079-10-1 — Explosive atmospheres: classification of areas, gas atmospheres.
Source: IEC 60079-13 — Explosive atmospheres: equipment protection by pressurized room and artificially ventilated room.
Source: IEC 61936-1 — Power installations exceeding 1 kV AC: common rules.
Source: IEC 61000-2-4 — Electromagnetic compatibility: compatibility levels in industrial plants for low-frequency conducted disturbances.
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 sugar, ethanol, biomass and cogeneration projects in Brazil, Thailand, Indonesia and Vietnam, 2015–2025 (cane rates, specific power consumption, load factors, impedance spreads, loss capitalisation rates, price bands and lead times quoted here are industry experience values, not published standards).