Powering High Speed: The Auto-Transformer (2x25kV) System Explained

The engineering problem that drove the invention of the railway autotransformer system was not, at its origin, about high speed. It was about mountains. When Japanese National Railways began planning the extension of the Sanyo Shinkansen from Okayama to Hakata in the late 1960s, the route crossed the Chūgoku Mountains through a series of long tunnels and steep grades that would concentrate traction loads in exactly the zones farthest from substations — the worst possible combination for a conventional 25 kV system. Calculations for the Shin-Kanmon Tunnel section, where multiple 16-car Hikari sets would simultaneously accelerate on an ascending grade after emerging from the tunnel, showed that a standard 1×25 kV supply with booster transformers could not maintain the 20 kV minimum pantograph voltage required for full-speed operation without substation spacing of less than 12 km — requiring four or five traction substations in a 50 km mountain section, each requiring a new 110 kV grid connection. The alternative proposed by JNR’s electrical engineering research team in 1970 was a supply arrangement that had been theoretically understood since the 1920s but never applied to a heavy railway: a 2×25 kV autotransformer system. Pilot testing on a 25 km section of the Tokaido Shinkansen between Odawara and Mishima in 1971 confirmed that the system could maintain pantograph voltage above 22 kV under all simulated traffic loadings with autotransformer spacings of 12 km and substations 50 km apart. The Sanyo extension, opened in 1975, became the first mainline railway in the world to operate with a full 2×25 kV AT supply from day one. Within twenty years, every major new high-speed line in France, the UK, South Korea, Taiwan, and China had adopted the same principle. The technology born from a Japanese mountain tunnel is now the global standard for any railway that must deliver more than approximately 20 MW per 20 km section.

What Is the Auto-Transformer (2×25 kV) System?

The Auto-Transformer (AT) system, formally designated the 2×25 kV system in EN 50163 (Railway applications — Supply voltages of traction systems) and IEC 60850, is a railway traction power supply configuration in which a traction substation feeds a 50 kV single-phase transmission circuit comprising a positive feeder conductor (+25 kV relative to the neutral rail) and a negative feeder conductor (−25 kV relative to the rail), with the running rail maintained at approximately neutral potential. Autotransformers — single-winding transformers with a mid-point tap — are installed at intervals of 10–15 km along the route, connecting the +25 kV contact wire, the −25 kV negative feeder, and the neutral rail, and presenting a nominal 25 kV supply voltage to the train’s pantograph at all points between substations.

The governing standard in Europe is EN 50163 for supply voltage quality, EN 50388 (Railway applications — Power supply and rolling stock — Technical criteria for the coordination between power supply and rolling stock) for the interaction between the AT supply and rolling stock converters, and the relevant clauses of EN 50119 for the overhead contact system. In the UK, Network Rail’s NR/SP/ELP/21321 governs AT system design; in France, SNCF Réseau’s IN-4012 technical instruction series; in Japan, JR’s internal IRTS-300 specification.

Circuit Operation: How the Autotransformer Works

The Autotransformer as a Mid-Point Tap Device

An autotransformer is a transformer with a single continuous winding, tapped at an intermediate point that serves as both the secondary terminal and the rail connection. In the AT railway system, the winding has three terminals: the top terminal connected to the contact wire (+25 kV), the bottom terminal connected to the negative feeder (−25 kV), and the mid-tap connected to the running rail (0 V nominal). The voltage across the full winding is 50 kV; the voltage from either end to the mid-tap is 25 kV.

When a train draws current I_train from the contact wire, the autotransformer’s action redistributes this current between the contact wire and the negative feeder so that the net current flowing in the earthed rail is minimised. The key current relationships at an autotransformer under load are:

The factor-of-four reduction in transmission losses is the most important single number in AT system economics. For a high-speed line carrying 20 trains simultaneously each drawing 10 MW (a typical morning peak on the LGV Est), the total traction power demand is 200 MW. In a conventional 25 kV BT system, resistive losses in the feeder conductors might consume 8–12% of transmitted power — 16–24 MW of losses, or approximately €1.5–2.2 million per year in wasted electricity at European grid prices. The AT system reduces this to 2–3% — a saving of €1.1–1.7 million per year in electrical energy alone, against which the capital cost of autotransformer equipment (approximately €200,000–350,000 per AT station, installed) pays back in under 5 years on a heavily trafficked line.

Voltage Profile Along the Feeding Section

The voltage that appears at the train’s pantograph — the pantograph voltage — determines whether the train can maintain full traction power. EN 50163 specifies a minimum permanent pantograph voltage of 17.5 kV AC (rms) for 25 kV nominal systems, with a non-permanent minimum of 15 kV during starting surges. The voltage profile in an AT system is fundamentally different from a BT system: rather than a single monotonically decreasing voltage gradient from substation to mid-point, the AT system creates a series of local voltage maxima at each autotransformer location and local minima at the midpoint between adjacent autotransformers.

The practical advantage becomes decisive when multiple trains are simultaneously in the feeding section. In a conventional BT system with 20 trains in a 40 km section, the cumulative voltage drop from superimposed load currents can reduce pantograph voltage below 20 kV during peak demand periods. The AT system’s lower effective impedance per unit length — achieved by the autotransformer’s current-halving action in the transmission section — maintains pantograph voltage above 22 kV under the same loading. This is why EN 50388 requires AT systems for any new HSR line designed for more than two simultaneously operating trains per 40 km section.

Autotransformer Station Architecture

An autotransformer station (ATS) is a compact, unmanned installation housing one or two autotransformer units, their associated switching equipment, protection relays, monitoring systems, and SCADA remote terminal units. A typical ATS occupies approximately 200–400 m² of land area — considerably smaller than a full traction substation — and requires only an 11 kV or 33 kV auxiliary power connection for station services (no high-voltage grid connection is needed, since the ATS draws its operating power from the 50 kV AT transmission loop itself via a small auxiliary transformer).

Switching Topology: Normal and Degraded Mode

Under normal operating conditions, all AT stations in a feeding section operate simultaneously, sharing the load current approximately equally (in proportion to their proximity to the loaded train). When an AT fails — whether due to transformer winding fault, protection operation, or planned maintenance outage — the feeding section’s control system automatically reconfigures: the sectioning switches adjacent to the failed AT open, electrically removing the unit from the loop, and the two adjacent ATs carry the additional load. The voltage profile degrades somewhat (the effective AT spacing doubles in the faulted zone from 12 km to 24 km), but the system typically remains within EN 50163 pantograph voltage limits if no more than one AT per section is simultaneously unavailable. This N−1 contingency capability is a mandatory design requirement under EN 50388 Clause 7.4 for all AT systems on lines operating more than 100 trains per day.

Power Demands of High-Speed Rail: Why Conventional 25 kV Cannot Scale

The power demand of modern high-speed rolling stock is the fundamental driver that makes AT systems mandatory rather than optional. Understanding the magnitudes involved explains why system designers have no viable alternative.

The N700 Shinkansen’s 17.1 MW peak demand figure is particularly instructive. On the Tokaido Shinkansen, up to 10 trains may be simultaneously in a single AT feeding section of 50 km during the morning peak, representing a peak simultaneous demand of up to 171 MW from a single 50 kV AT transmission circuit. Managing this load — while maintaining pantograph voltage above 20 kV on each train and keeping rail-to-earth leakage within EN 50122-1 limits — is a real-time power systems problem of considerable complexity, solved by the AT system’s distributed voltage support and JR Central’s train control system that regulates departure timing to stagger peak demand.

Regenerative Braking Recovery in AT Systems

The AT system’s distributed transmission architecture makes it the most effective conventional traction power supply for recovering regenerated braking energy. When a high-speed train brakes regeneratively, its traction motors operate as generators, pushing current back into the contact wire. In a conventional 25 kV BT system, this regenerated current can only be absorbed by other trains in the same electrical feeding section — approximately 20–40 km. If no receptive train exists, the regenerated energy raises the contact wire voltage, eventually triggering the rolling stock’s own regeneration inhibit function (which prevents the contact wire voltage exceeding approximately 29 kV AC) and wasting the energy as heat in rheostatic braking resistors.

In an AT system, the 50 kV transmission loop effectively connects all trains within the entire AT zone (50–100 km between substations) as well as the substation itself (which, in a reversible substation configuration, can feed regenerated energy back to the grid). A braking TGV Duplex at Valence generates approximately 5–7 MW for 45–90 seconds during its deceleration approach to the station. In the AT system of the LGV Méditerranée, this energy is available to any of the three to five trains simultaneously in the 80 km feeding section — including trains that may be on the opposite track 50 km away but connected to the same AT transmission circuit through the cross-track bonding at each AT station.

1×25 kV vs. 2×25 kV AT System: Full Technical Comparison

AT System Deployments: Global Specifications

Frequently Asked Questions

1. If the AT system transmits at 50 kV, why does the train still receive 25 kV — and how does this work without the train needing its own 50 kV capability?

The train always sees 25 kV because the autotransformer acts as a continuous voltage regulator between the 50 kV transmission loop and the 25 kV train supply. At every autotransformer station along the route, the contact wire (+25 kV) is connected to the mid-tap of the autotransformer winding, and the negative feeder (−25 kV) is connected to the bottom terminal. The top terminal of the autotransformer winding connects to the contact wire of the adjacent AT zone through the sectioning switch. When a train is between two AT stations and draws current from the contact wire, it draws from a source that is referenced to the neutral rail at +25 kV — the voltage of the mid-tap, not the full 50 kV between feeder and contact wire. The train’s pantograph, the rolling stock’s traction transformer (which steps 25 kV down to the intermediate DC voltage of the traction drive), and all rolling stock electrical clearances are designed for 25 kV maximum, as specified in EN 50163. The 50 kV transmission voltage is entirely confined to the transmission section between the substation output and the AT input — a section where no rolling stock equipment is connected and all electrical clearances are designed for 50 kV. This architectural separation between 50 kV transmission and 25 kV distribution is the autotransformer system’s most elegant design feature: it delivers the efficiency benefits of high-voltage transmission without requiring any change to the rolling stock specification.

2. What happens to the power supply during the period when a train is directly beneath an autotransformer station — does the transition create any electrical discontinuity?

Passing directly beneath an AT station is, from the train’s electrical perspective, entirely transparent. The autotransformer maintains the contact wire voltage at its output terminal continuously, regardless of whether a train is approaching from the upstream side, the downstream side, or is directly beneath it. What changes as the train passes through is the current distribution in the circuit: when the train is upstream of the AT, current flows from the substation through the upstream contact wire, into the train, back through the upstream rail, and returns through the AT to the negative feeder. As the train crosses the AT, the current path transitions: part of the return current switches from the upstream rail to the downstream rail, and the AT begins sourcing current from both its upstream and downstream connections proportionally. At the midpoint of the AT, both sides contribute equally. After the train passes, the downstream AT becomes the primary current source. At no point does the contact wire voltage experience a significant step change — the AT’s response to changing current distribution is essentially instantaneous (the transformer’s electrical time constant is microseconds). Train acceleration and power consumption are completely unaffected by AT station passage. This is in deliberate contrast to phase break passage, where the train must open its VCB and coast — an operationally significant interruption to traction. The transparent AT passage is one of the practical operational advantages of the AT system over designs that use discrete switching points for voltage step-down.

3. How does the AT system handle a contact wire short-circuit fault — and is fault location more difficult than in a conventional 25 kV system?

A contact wire short-circuit in an AT system produces fault behaviour that is more complex than in a conventional 25 kV system but is managed by the protection architecture. When a fault occurs at a contact wire — typically a pantograph strike, a fallen object bridging the wire to earthed metalwork, or an insulator breakdown — it presents a near-zero-impedance path from the contact wire (+25 kV) to earth. In a conventional system, the fault current flows from the substation through the contact wire to the fault, and back through the rail and return conductor to the substation negative: a simple, largely linear current path. In an AT system, fault current flows simultaneously from the upstream substation, from the downstream substation (in the opposite direction through the feeder), and from the nearest AT stations on both sides, all converging on the fault location. The magnitude of the fault current is higher (because more sources are contributing) — typically 3–8 kA at a mid-section fault — but it decreases more rapidly with fault location distance from a substation than in a BT system. The protection relays at both the traction substations and the AT sectioning switches use distance protection algorithms (impedance relay elements, IEC 61850 type 21) that measure the impedance to the fault and compare it with the known impedance profile of the AT feeding section. Modern AT protection systems using travelling-wave fault location technology — originally developed for HV overhead lines and adapted for traction use in Japan in the 1990s — can locate a fault to within ±100 m on a 50 km AT section within 50 ms of fault inception, enabling rapid targeted isolation and minimising the length of OCS taken out of service.

4. Can the AT system accommodate mixed traffic — both 25 kV AC electric locomotives and diesel or battery-electric trains on the same route — without disrupting the AT circuit?

Mixed traffic is well accommodated by the AT system architecture. Diesel trains, battery-electric trains in off-wire mode, and hydrogen fuel-cell trains draw no current from the OCS and are electrically invisible to the AT circuit — they pass beneath the energised overhead equipment without any interaction. The AT feeding section simply supplies zero current to those train positions; the voltage profile is unaffected. Bi-mode trains (such as the UK’s Class 800/802 series operating in electric mode under 25 kV OCS and in diesel mode on non-electrified sections) connect to the AT system normally when their pantograph is raised and the main circuit breaker is closed in electric mode. The only mixed-traffic scenario that requires active management is when a non-electrified section meets an electrified AT section at a changeover point: the sectioning at the changeover must ensure that the 50 kV negative feeder terminates cleanly and that no induced voltage from the live AT feeder reaches the non-electrified section’s metalwork. This is managed by the same insulation and bonding arrangements used at AT system phase break boundaries. On the East Midlands Railway network, where bi-mode Class 810 trains will transition between AT-electrified Midland Main Line sections and non-electrified peak-hour working, the changeover points at Market Harborough and Leicester have been specified with a 300 m transition zone incorporating AT feeder termination insulators and earthed guard wires to prevent induction onto the non-electrified section’s OCS stubs.

5. Why does the AT system produce less grid unbalance than a conventional 25 kV system, and does this matter for railway operators?

Both conventional 25 kV and AT systems are fundamentally single-phase loads on a three-phase grid — and single-phase loads always cause some degree of grid unbalance (negative-sequence voltage) that reduces the efficiency of rotating machinery connected to the same grid bus. The AT system does not eliminate this unbalance; it reduces it relative to the BT system because it draws less current from the grid at each substation connection point for the same power delivered to trains. The current drawn from the grid at a traction substation is proportional to the load current in the feeding section. In an AT system, the same train load is supplied at half the feeder current (due to the 50 kV transmission voltage), so the current drawn from the 132 kV grid connection is approximately half that of a BT system serving the same traction load. The negative-sequence voltage component induced in the grid is proportional to the unbalanced current — so the AT system produces approximately half the grid unbalance of an equivalent BT installation. For a heavily trafficked HSR line connected to a grid node that also supplies industrial processes sensitive to negative-sequence voltage (arc furnaces, large induction motor drives), this matters enormously: grid operators in France, Japan, and the UK routinely require traction load unbalance assessments as a condition of grid connection consent, and AT systems consistently achieve compliance at connection points where BT systems would require phase-balancing compensation equipment (Scott transformers, STATCOM installations) costing €3–8 million per substation.