Resource

Transformer Rectifiers for Steel and Aluminum Smelting

Aluminum runs on DC rectifiers (300–500 kA at 1,200–2,000 V across 300+ cells in series); steel runs mostly on AC arc-furnace transformers, and only a DC arc furnace needs a smelting rectifier. Pick the wrong category and no amount of pulse-count engineering saves the order.

If you searched for transformer rectifiers for steel and aluminum smelting because one supplier quote lists both industries under the same product page, this guide is for you. Those two processes sit at opposite ends of the power-conversion spectrum, and treating them as one load is the single most expensive mistake in metallurgical procurement. Here is the honest breakdown: what each process actually draws, how to turn that into a transformer rating, where the harmonic limits really apply, and what to put in an RFQ so that three suppliers come back with comparable numbers.

Start here: steel and aluminum are not the same load

Aluminum is a DC process, steady for years

Primary aluminum is made by the Hall-Héroult process. Current passes through hundreds of reduction cells connected in series, and each cell holds an operating voltage of roughly 3.9 to 4.5 V. That tiny per-cell voltage is why the cells are stacked in series. One modern potline runs 300 to 500 kA continuously, and there is no meaningful load cycle to speak of. It starts once and then holds that current for years.

The supply side has to deliver enormous DC current at a modest DC voltage. ABB’s review of high-power rectifiers reports that 300 pots in series used to mean 1,200 to 1,350 VDC, that current smelters run at 1,500 to 1,850 VDC, that plants with more than 460 pots are being designed near 2,000 VDC, and that ratings up to 3,000 VDC are expected around 2030. The trend in current is just as clear: potlines at 300 to 500 kA today, with super-high-amperage lines beyond 500 kA coming.

Steel is mostly an AC process

A conventional electric arc furnace does not take DC at all. It takes a three-phase AC feed, usually 33 or 34.5 kV on the utility side, stepped down to a few hundred volts at the electrodes. The transformer in that arrangement is a furnace transformer, built to survive electrode shorts and violent thermal cycling. Calling it a rectifier transformer is not just loose language; it points the supplier toward the wrong design rules.

Where a rectifier genuinely appears in a steel plant

Three places, and only three in most mills:

  • DC arc furnaces, where a rectifier feeds the graphite electrode. Published ratings run 50 to 130 kA at 600 to 1,150 VDC.
  • Rolling mill DC drives, where rectifiers feed the main mill motors. These are load-commutated or voltage-source arrangements, not electrolysis supplies.
  • Auxiliary DC buses and electrostatic precipitators, which are small by comparison.

Everything else in steelmaking, including ladle furnaces and submerged arc furnaces, is AC furnace-transformer territory.

Why the mix-up costs real money

A rectifier transformer is designed around converter topology, non-sinusoidal current, a specified secondary winding arrangement, and a continuous duty profile. A furnace transformer is designed around short-circuit withstand, high reactance, and an on-load tap changer rated for hundreds of thousands of operations per melt campaign. Ask a factory for “a rectifier transformer for a steel plant” and you may receive a unit with the wrong impedance, the wrong tap duty, and no water-cooled low-voltage bus. It will pass a routine test and then fail in service.

What a “transformer rectifier” actually is

The transformer does not make DC

This distinction comes up in almost every technical FAQ, and it is worth stating plainly. The transformer changes voltage, provides isolation, and creates the phase displacement that the rectifier needs. The diodes or thyristors perform the AC-to-DC conversion. Specify them as one system, because the winding arrangement and the bridge topology have to match. A phase-shift error between the two halves of a 12-pulse supply does not stay theoretical; it shows up as circulating current and extra heating in parallel units.

For a three-phase bridge, the no-load DC voltage is approximately 1.35 times the secondary line-to-line RMS voltage, before commutation reactance and regulation drops. That relationship is the bridge between a DC requirement and a transformer nameplate.

The five blocks of a smelter DC supply

A large smelter supply is a chain, not a single box:

  1. Grid connection at the point of common coupling, commonly 132, 220, or 380 kV for a primary aluminum plant.
  2. A regulating transformer, often a three-winding unit with a tertiary winding feeding a power-factor and harmonic filter.
  3. One or more rectifier transformers with phase-shifted secondaries.
  4. The rectifier bridges, built from diodes or thyristors in parallel, with cooling and fusing.
  5. The DC bus and the potline or furnace, plus the saturable reactors or firing-angle controls that trim current.

Ignore any one of those five and the harmonic study at the PCC will not reflect reality.

Aluminum potlines: the numbers worth memorizing

Cell voltage sets the series voltage

Start with the cell. At 3.9 to 4.5 V per cell, a 300-cell line lands between 1,170 and 1,350 V before bus drops. Add roughly 5% for busbar loss and regulation margin and you are near 1,300 VDC. A 460-cell line at the same per-cell figure is closer to 2,000 VDC. The practical consequence is simple: the transformer is specified against the full cell count at end of campaign, not against today’s voltage.

Current sets the number of parallel units

Single rectifier-unit ratings have been pushed beyond 100 kA, but a whole potline at 300 to 500 kA is fed by four to six units in parallel, and sometimes more at the highest currents. ABB’s design guidance calls for n-1 redundancy on parallel semiconductors and n-2 redundancy on the parallel rectifier units feeding the line. In plain terms, the plant is expected to keep producing after one unit is taken out of service.

Worked example: a 300 kA potline

Take a 300-cell line at 4.2 V per cell. That gives a stack voltage of 1,260 V. Add a 5% busbar allowance and the rectifier envelope is about 1,323 VDC. At 300 kA, the DC power is 378 MW.

Split that across six parallel units: each carries 50 kA and about 63 MW of DC. At a practical conversion factor of 0.97, each transformer is roughly 65 MVA. That is the order of magnitude that shows up in real project references, and it is why aluminum rectifier transformers are usually oil-immersed with forced cooling. A dry-type unit at that current and duty is not a realistic proposition.

Voltage regulation is where the designs diverge

Two main approaches show up in aluminum:

  • Diode rectifiers with on-load tap changers and saturable reactors. Diodes have dominated aluminum since the 1970s because control is simple and efficiency is high. Coarse voltage comes from the tap changer, and fine current trim comes from saturable reactors, typically over a 60 to 80 VDC window. This is the classic low-harmonic, high-power-factor arrangement.
  • Thyristor rectifiers with firing-angle control. First delivered for aluminum in 1985 and increasingly accepted since. They regulate faster and respond quickly to load-shedding schemes, but they generate more harmonics and run at a lower power factor. Keeping the firing angle inside a narrow band, combined with tap control, recovers much of that power-factor penalty.

Neither is universally better. Diode plus OLTC plus saturable reactor suits a stable base load. Thyristor suits a plant that needs fast response. Say which one you want before you ask for a price.

Steel: furnace transformers, DC arc furnaces, and mill drives

The AC electric arc furnace is the default

A 50 MVA-class ladle furnace transformer might be fed at 33 or 34.5 kV and deliver 180 to 400 V at the secondary, with nominal current around 72 kA. The tap changer commonly offers 32 steps across a wide range, and the low-voltage side uses water cooling rather than oil, because oil is not practical at those currents. Short-circuit duty is measured in multiples of rated current for a defined duration. Electrode contact with scrap is a normal event, not a fault.

The OLTC on that transformer is specified to an operating class such as C4 or C5 under IEC 60214, rated for more than 200,000 operations without maintenance. A standard distribution OLTC will not survive a melt shop.

DC arc furnaces are a different animal

When the furnace is DC, the supply looks much more like an electrolysis rectifier. Published converter ratings for DC arc furnace service sit at 50 to 130 kA and 600 to 1,150 VDC. At 100 kA and 800 V that is 80 MW of DC power. The rectifier has to tolerate rapid load swings as the arc lengthens and shortens, so dynamic response matters more here than in an aluminum potline.

Rolling mill drives bring harmonics of their own

Hot and cold continuous mills use high-power drives that inject 5th, 7th, and higher harmonics into the plant supply. The countermeasure is high K-factor construction, multi-winding phase shift, split windings, or some combination. If your enquiry is for a mill drive and you describe it as “smelting,” a supplier may quote a completely different thermal design.

Side by side: what each process actually needs

ItemAluminum potlineDC arc furnaceAC arc furnace (not a rectifier load)
OutputDC, steady for yearsDC, fast load swingsAC, severe cyclic duty
Typical DC/AC voltage1,200–2,000 VDC, higher planned600–1,150 VDC180–400 V secondary
Typical current300–500 kA per line50–130 kAup to ~72 kA at 50 MVA class
Common topology12-pulse unit, 5 units for 60-pulse6/12-pulse, sometimes more unitsThree-phase AC, no rectifier bridge
RegulationDiode + OLTC + saturable reactor, or thyristor firing angleThyristor firing angle for speedHeavy-duty OLTC, 32 steps
CoolingOil-immersed, ONAN/ONAF/OFAF/OFWFOil or water depending on designWater-cooled low-voltage side
Reliability basisn-1 semiconductors, n-2 parallel unitsAvailability plus arc dynamicsShort-circuit withstand, OLTC duty class
ApplicationCurrent rangeDC voltage rangeSource pattern
Aluminum potline10,000–300,000 Aunder 1,300 V, newer lines higherseries cells, parallel rectifier units
DC arc furnace50,000–130,000 A600–1,150 Vsingle or few bridge units
Graphitizing furnace20,000–120,000 A50–250 Vvery high current, low voltage
Copper refining10,000–50,000 A40–350 Vlower voltage electrolysis
Zinc and lead electrolysis5,000–100,000 A100–1,000 Vcontinuous duty
Chemical electrolysis5,000–150,000 A40–1,000 Voften double-star with interphase transformer

Sizing the transformer from the DC duty

Step 1: fix the DC envelope, not the average

Write down four numbers before anything else: normal DC voltage, maximum DC voltage at full cell count, continuous DC current, and peak current with its duration. The overload profile matters as much as the nameplate. A common electrolysis requirement is 1.2 times rated load for four hours and 1.75 times for one minute. A furnace has a completely different profile. If you only give a supplier “60 MVA,” you have given them nothing they can design against.

Step 2: apply the conversion factor, then the margin

Work from DC power to apparent power with the rectifier conversion factor, then add a continuous-duty margin of 1.25 to 1.35 for a 24/7 electrolysis line. In the worked example above, 63 MW of DC per unit became roughly 65 MVA at a 0.97 factor, before applying the service margin that a real project would add. Do not confuse the two. The conversion factor accounts for the bridge; the margin accounts for the fact that the plant never stops.

Step 3: state impedance honestly, because the sources disagree

Here is the part where published numbers appear to contradict each other. You will find 6 to 12% quoted for furnace and general power transformers, 20 to 50% quoted for arc-furnace transformers on the low-voltage side, and 4 to 8% quoted for rectifier transformers. All of them can be correct, because the figure is chosen by what the designer is trying to limit.

For an electrolysis rectifier, commutation reactance governs current transfer between valves and sets the fault level that the semiconductors must survive. Too low and a commutation fault destroys the thyristors; too high and voltage drop under load becomes excessive. For an AC arc furnace, high reactance limits the current when an electrode shorts to scrap and helps stabilize the arc. That is why furnace figures run much higher.

The right move in an RFQ is to ask for impedance at a named tap and a named reference side, with a tolerance. Parallel rectifier transformers are commonly held within about ±2.5% of each other so that current shares evenly. Never copy a distribution-transformer impedance into a converter specification.

A note on “standard” values

If a supplier quotes one impedance figure without saying which tap, which side, and which tolerance, treat the quote as incomplete. The same transformer can present different values depending on where you measure, and the converter designer needs the value at the operating tap.

Harmonics and pulse number: what IEEE 519 actually asks for

The harmonic ladder

For a p-pulse converter, characteristic harmonics occur at h = kp ± 1, with amplitude falling roughly as 1/h. Ideal total harmonic distortion follows the expression √( (π²/p²)·csc²(π/p) − 1 ). Working through it:

Pulse countIdeal THDPhase displacement between bridges
631.08%60°
1215.22%30°
1810.11%20°
247.57%15°
365.04%10°
483.78%7.5°
603.02%6°

These are ideal values. Real plants measure higher, because phase-shift error, firing angle, and commutation overlap all spoil the cancellation. Five 12-pulse units phased at −12°, −6°, 0°, +6°, and +12° give a 60-pulse system on the DC bus, which is a recognized aluminum configuration.

Why 24 pulse is often the practical answer

More pulses mean lower harmonics but also more windings, more interfaces, more floor space, and more places for phase error to creep in. For many industrial plants, 12 pulse is the baseline and 24 pulse is the upgrade when the utility contract is tight. A 60-pulse arrangement makes sense on a large primary smelter with a dedicated high-voltage connection. It rarely makes sense on a mid-size plant.

THD belongs at the PCC, not on the nameplate

IEEE 519-2022 applies at the point of common coupling, to the whole installation, not to one transformer. A supplier cannot certify IEEE 519 compliance from transformer data alone. They can supply the impedance, phase displacement, and loss inputs that a system study needs. If a datasheet claims “IEEE 519 compliant” without naming the PCC and the rest of the load, ask what study it came from.

Reliability: n-1, n-2, and the cooling decision

Redundancy is a design input

For an aluminum potline, n-1 redundancy on parallel semiconductors and n-2 on parallel rectifier units are usual design requirements. Four to six units in parallel is normal. If a vendor proposes a single large unit to save capital cost, ask what happens during an unplanned outage and what the production loss per hour is worth. The answer usually settles the argument.

Diode or thyristor changes the whole system

Diode rectifiers are simpler, efficient, and produce fewer harmonics, but they need a tap changer and saturable reactors for control. Thyristors respond faster and support load shedding, but they add harmonics and consume reactive power. Your choice affects the transformer, the filter, the PCC study, and the control philosophy. Decide it before the enquiry goes out.

Cooling in real smelter service

Oil-immersed construction dominates high-current converter duty. Cooling classifications run from ONAN and ONAF through OFAF and OFWF, the last being forced oil with water cooling where cooling water is available. Cast-resin dry type earns its place indoors, at lower power, and where fire safety rules out oil. A dry-type unit is not a credible substitute for a 65 MVA aluminum rectifier transformer, but it can be the right answer for an indoor DC bus or a traction substation.

Standards and documents to name in the RFQ

StandardScopeWhy it matters here
IEC 61378-1Converter transformers for industrial semiconductor converter plantsPrimary standard for rectifier-duty construction
IEEE C57.18.10Semiconductor rectifier transformersCommon North American reference
IEC 60214-1 / -2On-load tap changersOperating class and duty, especially for furnace OLTC
IEEE 519-2022Harmonic control at the point of common couplingSets the system-level harmonic target
IEC 60076 seriesGeneral power transformer requirementsInsulation, temperature rise, short-circuit withstand
IEC 60076-5Short-circuit withstandCentral to furnace and high-fault converter duty

Ask for the factory acceptance test list to include load-loss measurement, impedance measurement, partial discharge, temperature rise, and harmonic-loss verification. For parallel units, ask for the current-sharing evidence.

How to write an RFQ that gets comparable quotes

The data pack to send

  • DC voltage, continuous DC current, peak current, and the duration of each peak
  • Rectifier topology, pulse number, and required THD or the utility contract limit
  • Primary grid voltage, frequency, and tap requirements
  • Short-circuit impedance target, with the reference side and tolerance stated
  • Cooling type, indoor or outdoor installation, altitude, ambient range, and any corrosive atmosphere
  • Applicable standards and required test certificates
  • Redundancy philosophy, expansion margin, and the interface point with the rectifier supplier

Three questions that expose a weak offer

  1. “Which winding arrangement and phase displacement does your design assume, and will you confirm it with our rectifier supplier?” A vague answer means the two packages will not integrate.
  2. “At which tap and which side is your impedance figure quoted?” If there is no answer, the short-circuit and harmonic numbers are not usable.
  3. “What is excluded?” Freight, commissioning, documents, spares, and testing are where two apparently similar quotes diverge.

Five mistakes that keep repeating

  • Ordering a standard power transformer and connecting a rectifier to it. Harmonic currents create extra eddy losses in windings, clamps, and structural parts. The unit overheats early and the warranty does not cover it.
  • Quoting DC power only. The transformer sees apparent power with a non-sinusoidal current. Form factor and conversion factor both matter.
  • Choosing pulse number in isolation. A 48-pulse system is not automatically better than 12 pulse. Check the source balance, the PCC objective, the footprint, and the maintenance obligation.
  • Leaving the duty cycle undefined. The factory will assume continuous steady load unless you say otherwise, and a furnace is anything but steady.
  • Splitting transformer and rectifier between two suppliers without a shared interface document. This is how phase-shift mismatches and circulating currents get designed into a plant.

FAQ

What is a transformer rectifier used for in steel and aluminum smelting?

In aluminum it feeds the reduction cells with steady high-current DC. In steel it is used mainly for DC arc furnaces and mill drives. A conventional AC arc furnace uses a furnace transformer, not a rectifier transformer.

Does the transformer convert AC to DC by itself?

No. The transformer changes voltage, provides isolation, and sets phase displacement. The diode or thyristor bridge performs the AC-to-DC conversion. They must be specified as one system.

What DC voltage does a 300-cell aluminum potline need?

At 3.9 to 4.5 V per cell, a 300-cell line is roughly 1,200 to 1,350 V before busbar drops. Newer lines with more than 460 cells are designed near 2,000 VDC.

How much current does a modern aluminum potline draw?

Typically 300 to 500 kA, with a trend toward lines above 500 kA. Four to six rectifier units usually share that current in parallel, with single-unit ratings now beyond 100 kA.

Why are so many rectifier units connected in parallel?

For redundancy and current sharing. Aluminum designs commonly require n-1 redundancy on semiconductors and n-2 on parallel rectifier units, so the line keeps running when one unit is out of service.

Is 24 pulse always better than 12 pulse?

No. Moving from 12 to 24 pulse lowers ideal THD from about 15.22% to 7.57%, but it adds windings, interfaces, and phase-error risk. Choose against the PCC limit, source balance, and footprint.

Can transformer data alone prove IEEE 519 compliance?

No. IEEE 519-2022 applies at the point of common coupling to the whole installation. The supplier provides impedance, phase displacement, and loss inputs for a system study, not a compliance certificate.

Why do published impedance figures range from 4% to 50%?

They describe different design goals. Electrolysis rectifiers are commonly 4 to 8%, general power transformers 6 to 12%, and arc-furnace transformers 20 to 50% on the low-voltage side to limit electrode short-circuit current. Always ask for the tap, the reference side, and the tolerance.

Diode or thyristor rectifier for an aluminum smelter?

Diodes with an OLTC and saturable reactors suit a stable base load and give high efficiency with lower harmonics. Thyristors respond faster and support load shedding, at the cost of more harmonics and lower power factor.

Why does an arc-furnace transformer need a special tap changer?

Electrode shorts and arc instability force hundreds of tap changes per melt. Furnace OLTCs are specified to an operating class such as C4 or C5 under IEC 60214, rated for more than 200,000 operations without maintenance.

Can a dry-type rectifier transformer be used in a smelter?

For high-current primary aluminum duty, oil-immersed construction with forced oil or water cooling is the norm. Dry type makes sense indoors at lower power, for traction substations, and where fire rules out oil.

What should I send to get a real quotation?

DC voltage and current with peak durations, rectifier topology and pulse number, primary voltage and frequency, impedance target, cooling and site conditions, standards and tests, and the interface point with the rectifier supplier. Without those, three quotes will not be comparable.

What is the biggest procurement mistake in this category?

Treating steel and aluminum as the same load. Aluminum is a steady DC electrolysis process; most steelmaking is an AC furnace process with severe short-circuit duty. The transformer designs are not interchangeable.