Railway Traction Transformer: How to Size a 25 kV Traction Substation, Why V/v and Scott Connections Decide Grid Unbalance, and What IEC 62695 / EN 50329 Requires

The answer first: a railway traction transformer is specified by the traction system — voltage, frequency, connection and duty cycle — long before kVA enters the conversation

A traction transformer is the unit that turns the electricity a railway buys from the public grid into the electricity a train can actually use: a three-phase 110 kV, 132 kV, 220 kV or 33 kV supply stepped down — or converted three-phase to two-phase — into a 25 kV 50 Hz, 25 kV 60 Hz or 15 kV 16.7 Hz contact line, or into the 600–1,900 V DC bus of a metro or light-rail system through a rectifier. It is not a distribution transformer with a railway nameplate: traction load is single-phase, intermittent, harmonic-rich and frequently reversed by regenerative braking, and the standards that govern the unit — IEC 62695 / EN 50329 for fixed installations, IEC 60310 for on-board equipment, IEC 61378-1 and IEEE 1653.1 for rectifier transformers — exist precisely because those duties break ordinary distribution designs. Four decisions shape the unit: the connection (single-phase, V/v, Scott), the voltage and frequency of the traction system, the duty cycle including overload and regeneration, and the harmonic content the windings must survive. Get those four right and the kVA follows almost automatically; get the traction system wrong and no kVA value will rescue the project.

Key takeaways

• The traction system decides the transformer, not the reverse. 25 kV 50 Hz, 25 kV 60 Hz, 15 kV 16.7 Hz and the DC families (1.5 kV, 750 V) each imply a different core, insulation level and connection — see 50 Hz vs 60 Hz transformer design and how to choose a voltage class.
• Single-phase traction load unbalances the three-phase grid. Connection choice (single-phase, V/v, Scott), staggered phasing between adjacent substations, and where necessary compensating equipment are the tools that keep negative-sequence voltage unbalance inside the grid code limit — an engineering constraint, not a formality.
• Traction duty is cyclic, not steady. Trains accelerate, coast, brake and regenerate, so the unit must be specified on a load cycle and an overload capability, not on one steady-state figure.
• Rectifier and converter duty adds harmonics, and sometimes DC magnetisation. Where the contact line is DC, or where static converters sit between transformer and train, harmonic loss and derating rules apply — the same mechanism as harmonic overheating in transformers, and the same family as rectifier transformers for electrolysers.
• Under-specifying impedance or cooling is the classic mistake. Impedance sets both fault current and voltage regulation; cooling and temperature rise set insulation life. Both belong in the technical specification before a price is quoted — see short-circuit current calculation and cooling methods and temperature rise.

Where does the traction transformer sit in the electrification chain?

A railway electrification scheme is a chain: public grid → traction substation → feeder and contact line → train. The transformer usually appears twice in that chain (once in the substation, once on the train) and sometimes a third time, as an autotransformer along the track. IEC 62695 lists the fixed-installation family explicitly, and the list is a good map of what a transformer factory is actually asked to build.

Customer-facing projects usually include two or three of these roles in one tender. An African or Asian corridor electrification package typically contains substation traction transformers, trackside autotransformers or booster transformers, and auxiliary units — and the tender documents will name IEC 62695, IEC 60076 and IEC 60850 together. Reading the chain before reading the rating is what keeps a quotation comparable: a supplier who quotes a substation unit when the tender asked for a trackside autotransformer has answered a different question.

On the train itself the equipment changes character. On-board traction and auxiliary transformers are governed by IEC 60310 — narrower temperature windows, forced cooling, severe vibration and shock requirements, weight targets, and a duty cycle built from accelerating, cruising, braking and regenerating phases. On-board units are outside the scope of IEC 62695, and most export factory buyers are not procuring them; the useful takeaway is that the same physics (duty cycle, harmonics, thermal cycling) that shapes an on-board unit also shapes the substation unit, only with more room and less weight pressure.

Which traction system is yours? Nominal voltage and frequency decide everything downstream

Six standardised systems cover most of the world's electrified route length, and they split into three AC families and three DC families. The differences matter to the transformer because they change the core size, the insulation design, the connection and the way the unit responds to load.

For AC systems, the transformer's rated voltage and tap range must be built around the supply voltage window rather than the nominal figure alone. EN 50163 / IEC 60850 defines the traction supply voltage as a set of limits — a lowest permanent voltage, a lowest non-permanent voltage, a highest permanent voltage, a highest non-permanent voltage, and a long-term overvoltage level for a short duration — expressed as per-unit values of the nominal voltage. For the 25 kV system the widely quoted permanent band is roughly −24% to +10% of nominal with a short-duration excursion to about +16%; for the 15 kV 16.7 Hz system roughly −27% to +15%, with further excursion allowed for limited periods. Design to the standard's own table, not to a remembered percentage: the number decides the tap range, the insulation level and how far the train-end voltage is allowed to sag under a heavy simultaneous start. The Chinese equivalent is GB/T 1402 (*Railway applications — Supply voltages of traction systems*), aligned with IEC 60850.

Why can't a distribution transformer do the job? Four ways traction duty differs

Every year some project tries to save money by fitting a standard distribution transformer into a traction duty. Sometimes it works for a while, and then one of four mechanisms catches up. Understanding them is the difference between a competitive quotation and a warranty claim.

1. The load is single-phase, so the grid sees unbalance

A three-phase grid carrying a three-phase load has balanced currents and almost no negative-sequence component. A traction load is, at its simplest, a single-phase load: one phase pair carries the whole traction current and the other phase current is missing. The result is a negative-sequence current that produces a negative-sequence voltage unbalance at the point of common coupling, roughly in proportion to the ratio of the traction load to the three-phase short-circuit capacity at that point (a useful first-order engineering approximation — the exact figure depends on impedance angles and on how the traction load is connected). Grid codes cap this: China's GB/T 15543 limits negative-sequence voltage unbalance at the point of common coupling to 2% under normal operation and 4% for short periods, and comparable limits exist in most European and Asian grid codes — the international standards comparison guide shows how the GB, IEC and IEEE families divide these requirements in practice. The transformer, its connection and the phase rotation between adjacent substations are how a designer meets that limit.

2. The load cycle is violent and repetitive

A distribution transformer sees a slowly varying daily load curve. A traction transformer sees a train, then two trains, then five, then none: tens of MVA appearing and disappearing in tens of seconds, hundreds of times a day, with an overcurrent every time a train starts. Thermal cycling is therefore severe by design, and the specification must be written as a load cycle with an overload class rather than a single continuous rating. This is exactly the territory covered by load factor and overload capacity, and it is the reason traction units are often specified with a higher short-time overload capability than a distribution unit of the same nominal rating.

3. Regenerative braking reverses the power flow

Modern trains brake electrically and push energy back into the contact line, where another train may absorb it or — if the substation is reversible — where it may return to the grid. Regenerative braking can represent a substantial share of traction energy on commuter routes (industry experience; the share depends heavily on timetable density and gradients). For the transformer that means reverse power flow at the same order of magnitude as forward flow, and the transformer must be equally capable in both directions. It also raises the voltage at the receiving end, which interacts with the tap range discussed below. On DC systems, regeneration is the reason reversible substations with inverters exist at all — where a diode rectifier cannot accept power back, braking energy is burned in resistors on the train.

4. The current is distorted, and the fault duty is close and harsh

AC traction fleets have used phase-angle-controlled thyristor converters and, more recently, IGBT and SiC converters. Both produce harmonic currents; DC traction rectifiers produce large characteristic harmonics (5th, 7th, 11th, 13th and their families). Harmonic currents raise winding eddy losses roughly with the square of harmonic order, overheating the unit from the inside, exactly as described in why harmonic currents overheat transformers. At the same time the traction transformer sits electrically close to its load: contact line and feeder faults are frequent compared with distribution faults, so the unit must be designed for repeated short-circuit withstand and for the inrush that follows every re-energisation of a line section. Protection is specified accordingly — the same family of devices and schemes as transformer protection devices, and where the unit is oil-immersed, periodic dissolved gas analysis is the standard condition check.

V/v, Scott, Woodbridge or single-phase: how do you choose the connection?

The connection is the first technical decision on a traction transformer, and it is driven by a grid constraint rather than by a preference. There are four practical answers.

Three practical rules follow from that table. First, the connection is a grid-code argument, not a price argument. If the short-circuit capacity at the point of common coupling is comfortably large relative to the traction load, a V/v connection is normally acceptable and the cheaper answer. If the grid is weak or the unbalance limit is tight, the Scott connection earns its premium. Second, staggered phasing between substations is a system-level tool: connecting adjacent traction substations to different phase pairs so that their single-phase loads partially cancel in the upstream network is standard practice, and it must be coordinated with the connection choice rather than left to the substation designer. Third, the vector group still matters for the three-phase units and for the earthing arrangement. A three-phase unit with a delta primary traps triplen harmonics and decouples earthing systems, which is the same reasoning as in Dyn11 vs Yyn0 vector groups, and the grounding philosophy on the secondary side follows grounding system design — with the railway-specific insistence that the return current path and touch voltages are a safety matter in their own right, governed by the EN 50122 family.

Where a single-phase traction load is large relative to the grid and no connection can fix it, compensation is added — most often a Steinmetz arrangement of reactive elements, or a static compensator. As a transformer supplier this is outside the battery limit of your unit, but it changes what you must quote: the transformer may then be required to carry a load that is deliberately not balanced, or to accept a specified harmonic and reactive duty.

What changes in a 2×25 kV autotransformer system?

On longer corridors, the classic 1×25 kV arrangement — contact line fed at 25 kV, rail and earth as return — runs into a voltage-drop wall: the line impedance and the return impedance limit how far a substation can reach and how much power one section can carry. The standard answer is 2×25 kV autotransformer (AT) feeding: the contact line and a dedicated feeder are held at opposite polarities relative to the earthed rail, traction substations feed the pair at a higher line-to-line voltage, and single-phase autotransformers spaced along the track — commonly in the 8–15 km region, as an industry-experience figure — halve the effective impedance seen by the load.

Four consequences for the specification:

• The autotransformer is not a substation transformer. It is a single-phase, oil-immersed unit, in practice a 1:1 autotransformer between catenary and feeder with the midpoint connected to rail, so its line-to-line rated voltage is twice the system's line-to-rail voltage. Rating bands are typically much smaller than substation units, but it must be designed for outdoor, unattended trackside service and for cyclic load.
• Every unit in the chain sees a higher insulation duty. Because the two conductors are at opposite polarities, conductor-to-conductor voltages are higher than the phase-to-earth values that EN 50163 limits — a point that standards themselves flag for autotransformer systems. Insulation coordination therefore follows the railway insulation-coordination rules rather than a copy of the phase-to-earth design.
• Substation transformer sizing changes. An AT-fed substation delivers roughly the same traction power at half the contact-line current and half the voltage drop, which changes how much of the corridor one substation can serve and therefore how many substations — and how many transformers — the tender contains.
• The return path is a design deliverable. AT feeding exists partly to reduce current in the rails and earth, which reduces interference and touch voltages. The transformer's winding arrangement, its midpoint earthing and the protection that detects a faulted AT are all specified with that in mind.

How do you size a traction transformer? Six steps from timetable to nameplate

Sizing a traction transformer is not a matter of picking a kVA from a table. It is a small power-system study compressed into a specification, and the sequence below is the one that survives an independent review.

1. Fix the traction system and the feeding arrangement. Single-phase or 2×25 kV AT; number of tracks; substation spacing; whether adjacent substations are staggered across phase pairs; whether the substation is reversible. Every later decision inherits these choices.
2. Build the coincident traction load from the timetable, not from the locomotive rating. The traction transformer is sized on the trains that can be in the section at the same time — simultaneous starting is the governing case on commuter and metro networks, and the heavy-haul case can be worse. A corridor with two long freight trains climbing the same grade is not the same installation as one with a single EMU, even if the annual energy is identical.
3. Convert to kVA with the right power factor for the fleet. Modern converter-fed AC locomotives run at high power factor over much of their operating range, whereas older phase-angle-controlled fleets and DC rectifier loads pull reactive power and distort current. The nameplate is in kVA, so an optimistic power factor is an under-sized transformer.
4. Apply the duty cycle and the overload class instead of a steady rating. Express the load as a cycle — traction, dwell, regeneration, no-load — and check the thermal response over that cycle, including the peak hot-spot temperature. This is the step where cooling class and temperature rise and the insulation class are chosen, and where the classic 10 °C-halving-life relationship does the commercial arguing for you.
5. Choose impedance for the actual fight you are having. For directly fed AC traction, impedance is usually kept modest so that voltage regulation along the corridor stays acceptable; for rectifier duty, commutation reactance is a specified design parameter that shapes DC ripple and fault current. In both cases the value interacts with voltage regulation and with parallel operation if two units share a bus. Traction units commonly sit at a higher impedance than distribution units of the same rating (industry experience; typical values are set per project by the utility and the rectifier supplier, not by a rule of thumb).
6. Add environment, redundancy and taps, then select the nearest standard rating. Outdoor trackside units need weatherproofing and a design ambient suited to the site; high altitude or a hot climate requires the altitude and temperature derating treatment; N-1 redundancy decides whether the nameplate must carry the full section load alone; and the tap changer range must cover the EN 50163 supply window plus the corridor's own voltage drop. Only after that do you pick the standard rating — for substation traction transformers this is frequently a bespoke rating rather than a catalogue one.

Rectifier transformers for DC traction: why the pulse number decides the design

Where the contact line is DC, the transformer does not feed the train directly — it feeds a rectifier, and the rectifier's pulse number becomes the transformer's design brief. For a 750 V DC or 1,500 V DC substation, diode or thyristor rectifiers are built as 6-pulse, 12-pulse or 24-pulse groups. Increasing the pulse number does two things at once: it cancels the lowest-order characteristic harmonics at the DC bus and on the AC side, and it splits the current across more thyristors or diodes, which is how very high currents are handled at low voltage.

The consequences for the transformer are concrete:

• Two or four secondaries with a defined phase displacement. A 12-pulse group is normally achieved with two secondaries displaced by 30° (star and delta, or extended-delta windings); a 24-pulse group needs four displaced sets or two cascaded 12-pulse groups. The winding arrangement, not the kVA, is what the designer spends time on.
• Commutating reactance is a specified value. It sets the DC ripple and the fault current on the rectifier side, and it is chosen jointly with the rectifier supplier. This is squarely the territory of IEC 61378-1 for converter transformers in industrial applications, and of IEEE 1653.1 for traction power rectifier transformers for substation applications.
• Harmonic loss and eddy-current loss dominate the thermal design. A rectifier transformer carries a current waveform that a distribution unit was never designed for; derating, a K-factor equivalent, or a bespoke design is required. The mechanism is the same one described in K-factor transformer selection.
• DC-side disturbances can reach the core. Asymmetrical faults and converter behaviour can produce DC components that saturate a core designed only for sinusoidal flux; rectifier transformers are generally designed with more flux headroom and monitored for gas-in-oil anomalies during service.
• Where the rectifier is reversible, regeneration changes the requirement. A reversible substation means power flows back through the rectifier group and the transformer, so both must be specified for bidirectional duty rather than forward-only.

Related duty shows up outside railways too: EV charging hubs, solar PV inverters, battery storage and electrolyser rectifiers all impose the same converter-plus-transformer question, and the design logic transfers between them — only the pulse number and the duty cycle change.

What belongs in the technical specification and on the nameplate?

Traction tenders are lost at the specification stage. The list below is the minimum a supplier needs in order to quote a comparable number, and the minimum a buyer should demand back on the nameplate and the test report.

Two habits pay for themselves. First, ask for the nameplate fields to be listed in the quotation, not after the order: reading a nameplate back is exactly the skill described in how to read a transformer nameplate, and it catches wound-voltage and tap-range mismatches in minutes. Second, treat customisation as a scope list rather than an adjective. Most of what a railway buyer wants is standard options in a different combination; the items a factory genuinely cannot change are the same everywhere, as set out in OEM vs ODM transformer customisation.

Which tests prove a traction transformer is right?

Traction units are tested to the same structure as power transformers — routine tests on every unit, type tests on a representative design, special tests by agreement — with railway-specific content added by IEC 62695 / EN 50329 and, in Europe, by the interoperability framework. What a project should insist on seeing is:

• Routine tests: ratio and connection verification, winding resistance, insulation resistance and dielectric tests, no-load and load loss measurement with impedance, and oil dielectric tests for oil-immersed units. On a single-phase or multi-secondary traction unit, ratio measurement on every winding is not optional.
• Type tests: temperature-rise and impulse tests on a representative design, plus the railway-specific loading and short-circuit-withstand evidence. For rectifier units this is where the converter duty is demonstrated, not assumed.
• Harmonic and loss verification: loss measurement in the presence of harmonic current, or a documented harmonic-loss calculation, is what separates a compliant rectifier transformer from a distribution unit in disguise.
• Factory acceptance testing witnessed by the customer or its consultant, with the pre-shipment checklist closed out before the unit is packed — see pre-shipment inspection and the wider factory audit checklist.
• Site-side verification at commissioning: insulation resistance, ratio, vector-group check, protection settings and the earthing and return-circuit measurements required before a traction section is energised, as set out in commissioning tests.

Transport of large traction transformers deserves its own line in the contract. A substation transformer can be several tens of tonnes and travel by road to a rail site where a crane lift is limited; nitrogen-padded oil transport and the tilt limits described in transportation and installation apply in full, and the site access survey belongs to whoever is buying the unit.

What does a traction transformer cost, and what drives the price?

Traction transformers are not catalogue items and public price lists barely exist, so a range is only useful as a sanity check on a quotation. Three drivers dominate:

• Connection and winding complexity. A single-phase unit is the simplest case; a V/v pair uses two standard units; a Scott or Woodbridge unit needs a teaser or an interconnected winding with a tap at a non-standard position, which raises both material and manufacturing cost. A 12- or 24-pulse rectifier transformer with displaced secondaries is likewise a bespoke winding job.
• Impedance, tap range and cooling. A high-impedance design, a wide OLTC range with a diverter switch, or a forced-cooled design with oil pumps and fans all add cost that a distribution-transformer comparison will miss entirely.
• Evidence and environment. Type-tested designs with harmonic loss verification, C5-grade paint systems, marine or tunnel installations, seismic requirements and witnessed factory testing add cost that is invisible in a kVA figure but decisive in a tender evaluation.

For an order-of-magnitude anchor on the material side, the distribution transformer price breakdown shows how copper, steel and oil scale — but a traction unit should never be priced by scaling a distribution unit. For the commercial side, the right frame is what an FOB quotation actually contains and, over the asset's life, total cost of ownership, where the cost of a single unplanned traction outage dwarfs the difference between two quotations. Warranty and delivery terms are worth negotiating in writing before payment — the clauses that matter are the same ones set out in transformer warranty terms and delivery lead time. Where the tender is public, verified third-party certification such as CE marking or UL/CSA listing may be a pass/fail condition, so confirm it before bidding rather than after.

Five specification mistakes that cost traction projects money

1. Sizing on the locomotive rating instead of the coincident section load. The result is a unit that is both expensive and, on the governing simultaneous-start case, marginally rated. Start from the timetable.
2. Forgetting that regeneration reverses the flow. A transformer designed only for forward power, and a substation without a reversible path, wastes braking energy and, in some designs, exposes the unit to thermal cycling it was not specified for.
3. Quoting a distribution transformer with a railway nameplate. Harmonic loss, inrush duty and short-circuit-withstand requirements differ; the saving appears at quotation and disappears at the first DGA sample.
4. Leaving the supply-voltage window and tap range to the supplier. If the tap range does not cover the EN 50163 / IEC 60850 window plus the corridor's own voltage drop, the unit cannot hold the train-end voltage, and no amount of testing will fix a design that was scoped too narrowly.
5. Choosing the connection on price instead of on the grid code. A V/v connection is often the right and cheaper answer — but only after the negative-sequence limit at the point of common coupling has been checked against the traction load and the grid's short-circuit capacity.

If your project is at the stage where the traction system is fixed and the substation scheme is being drafted, the fastest way to sanity-check a first-pass rating is to run the load and loss arithmetic through before the tender goes out — the same calculation the engineering toolbox provides for power system design work. Bring the traction system, the coincident load and the duty cycle, and the rest of the specification follows.

Sources / 资料来源

• Source: IEC 62695:2014 — Railway applications – Fixed installations – Traction transformers (derived from EN 50329:2003 and its amendment A1:2010)
• Source: EN 50329:2003/A1:2010 — Railway applications – Fixed installations – Traction transformers
• Source: IEC 60310 — Railway applications – Traction transformers and inductors on board rolling stock (2016 edition; 2026 edition)
• Source: IEC 60850 — Railway applications – Supply voltages of traction systems
• Source: EN 50163:2004 and its amendments (A1, A2:2020, A3:2022) — Railway applications – Supply voltages of traction systems
• Source: GB/T 1402 — 轨道交通牵引供电系统电压 / Railway applications – Supply voltages of traction systems
• Source: IEC 60076 series — Power transformers (including IEC 60076-1 oil-immersed, IEC 60076-11 dry-type, IEC 60076-15 gas-filled)
• Source: IEC 61378-1:2011 — Converter transformers – Part 1: Transformers for industrial applications
• Source: IEEE 1653.1 — IEEE Standard for Traction Power Rectifier Transformers for Substation Applications up to 1500 V DC Nominal Output
• Source: IEEE 1653.2 — IEEE Standard for Uncontrolled Traction Power Rectifiers for Substation Applications up to 1500 V DC Nominal Output
• Source: IEC 62590 (EN 50328) — Railway applications – Fixed installations – Electronic power converters for substations
• Source: EN 50122 series — Railway applications – Fixed installations – Electrical safety, earthing and the return circuit (including Part 1 protective provisions and Part 3 mutual interaction of AC and DC traction systems)
• Source: IEC 62497-1 — Railway applications – Insulation coordination – Part 1: Basic requirements
• Source: IEC 62236-5 — Railway applications – Electromagnetic compatibility – Part 5: Emission and immunity of fixed power supply installations and apparatus
• Source: IEC 60850 / GB/T 15543-2008 电能质量 三相电压不平衡 — limit on negative-sequence voltage unbalance at the point of common coupling (2% normal operation, 4% short duration)
• Source: IEEE 519 — Recommended practice and requirements for harmonic control in electric power systems
• Source: IEEE C57.110 — Recommended practice for establishing liquid-immersed and dry-type power and distribution transformer capability when supplying nonsinusoidal load currents
• Source: EN 50388 — Railway applications – Power supply and rolling stock – Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability
• Source: Directive (EU) 2016/797 on the interoperability of the rail system within the European Union — framework under which the EN 50163 / EN 50329 family is applied
• Source: EU Regulation (EU) No 548/2014 on ecodesign requirements for small, medium and large power transformers — applies to grid-side supply transformers where relevant
• Source: QDTB Engineering Team — industry experience / 行业经验值 for typical traction substation and trackside autotransformer rating bands, autotransformer spacing practice, rectifier transformer configurations, relative cost drivers, design ambient and derating assumptions, and the relative share of regenerative braking energy on timetable-dense routes