How to size a traction power transformer for railway load demand

A traction power transformer should be sized from the electrical demand seen at the substation bus, not from the rated power of a single train. The practical starting point is to build a time-based load model for the route section supplied by the substation, then convert the resulting maximum demand, overload duty, power factor, harmonic content, voltage-drop limits, and redundancy requirements into transformer ratings and impedance requirements.

This becomes critical when a railway line appears adequately supplied on paper but experiences low catenary voltage during simultaneous acceleration, transformer overheating during dense service periods, or nuisance protection operations after a timetable change. In these situations, increasing the nameplate MVA alone may not solve the real issue. The selected traction power transformer must be evaluated as part of a system that includes the incoming grid, feeder arrangement, return circuit, rolling-stock behavior, and operating plan.

Start with the service pattern, not train nameplate power

A train may draw a very high power level while accelerating, but that demand is neither continuous nor necessarily coincident with the demand of other trains. Conversely, a line with moderate-power vehicles can produce a severe substation load where short headways, station spacing, gradients, and constrained dispatching cause several trains to accelerate within the same electrical section.

Begin by defining the electrical territory of the proposed substation. This should include the feeding arrangement, sectioning locations, adjacent substations, normal feed direction, and any contingency feed scenarios. A transformer that is acceptable with parallel feeding in normal operation may be insufficient when one neighboring supply point is unavailable and its section must be back-fed.

The load model should reflect actual operating states rather than an assumed average train load. At minimum, distinguish between:

• Acceleration after station stops, junction holds, or speed restrictions;
• Cruising demand at line speed;
• Auxiliary consumption for HVAC, compressors, lighting, controls, and onboard services;
• Coasting periods;
• Regenerative braking periods;
• Standing trains with auxiliaries energized;
• Degraded operating conditions, such as reduced headway recovery or single-track working.

For AC railways, the relevant values are commonly expressed as active power in MW, reactive power in MVAr, and apparent power in MVA. Transformer thermal sizing is tied principally to current and apparent power, while voltage regulation and network impact depend strongly on both active and reactive components.

Build a demand envelope from train movement

The most reliable approach is a load-flow or traction power simulation using route gradients, speed profiles, rolling-stock traction curves, station dwell times, timetable headways, feeder impedance, and substation topology. Where a full simulation is not yet available, a preliminary study can use a conservative diversity-based demand estimate, but the assumptions should be visible and traceable.

For each time interval, determine the combined power at the substation supply point:

P_total = ΣP_traction + ΣP_auxiliary + P_{network losses}

Reactive demand must then be included:

S = √(P² + Q²)

where S is apparent power, P is active power, and Q is reactive power. For a three-phase AC traction supply, line current can be estimated as:

I = S / (√3 × V)

where V is the relevant line-to-line voltage at the transformer secondary or supply bus.

These calculations should be made across the operating profile, not only at one selected peak. The evaluator needs at least three demand values: the highest short-duration peak, the highest sustained demand, and the expected demand under a credible contingency arrangement. A transformer may tolerate a short peak within its overload capability but still be unsuitable if that peak occurs repeatedly enough to raise winding hot-spot temperature beyond the intended duty.

Do not use average load as the primary rating basis

Average energy consumption is useful for utility billing and broad capacity planning, but it can conceal the demand events that determine transformer performance. A substation serving trains on a steep departure gradient may have a modest daily energy total while repeatedly seeing high current at the same points in the timetable. The thermal effect depends on load magnitude, duration, ambient conditions, cooling method, and the time available for the transformer to cool between peaks.

Conversely, applying the sum of every train’s maximum traction power without diversity can produce an unnecessarily oversized design. The right question is not “What is the total installed train power?” but “What combinations of train states can realistically occur inside this feeding zone, and how long do they persist?”

Separate normal capacity from contingency capacity

A major source of undersizing is treating normal operation as the sole design condition. Railway supply systems often need to maintain restricted service when a transformer, feeder, or adjacent substation is unavailable. The required level of continuity varies by route importance and operating philosophy, but the contingency case must be evaluated explicitly.

Where two transformers operate in parallel, their ratings, impedance values, vector groups, and tap positions must be compatible. A pair with substantially different impedance may not share load in proportion to nameplate rating. One unit can become overloaded while the other remains below its capacity, producing a misleading impression that total installed MVA is sufficient.

For a single-transformer installation, the decision is not simply whether the unit can carry the peak. It is whether the railway can accept the service consequence of losing that supply point, whether adjacent feeding can maintain voltage within vehicle limits, and whether scheduled operational restrictions are realistic during a fault or maintenance period.

Check voltage regulation before finalizing MVA

Transformer capacity and voltage quality are related but different constraints. A transformer can have enough thermal capacity while still allowing excessive voltage depression at the catenary or third-rail supply point during acceleration. The voltage seen by the train is influenced by source voltage, transformer tap setting, transformer impedance, busbar arrangement, feeder resistance and reactance, return-path impedance, and simultaneous train loading.

Transformer impedance deserves particular attention. Lower impedance generally reduces voltage drop through the transformer and can improve traction voltage during high demand. However, it also increases prospective short-circuit current, affecting switchgear duty, protection coordination, and earthing design. Higher impedance can limit fault current but may worsen voltage regulation. There is no universal “best” impedance percentage; it must be selected against the network study results.

Tap-changing capability should be considered as a voltage-management tool, not as a substitute for proper capacity or feeder design. Fixed taps may be suitable where the utility source is stable and the operating window is narrow. On-load tap changing may be justified where supply voltage varies materially and voltage must be maintained under changing demand. Its control logic should avoid excessive tap operations caused by fast traction-load fluctuations.

Use the correct voltage reference point

Checking only the transformer secondary bus can hide a problem at the far end of a feeder. The relevant limit is usually the voltage available to the rolling stock at its pantograph or collector interface, considering the worst credible train position. A demanding train may be electrically distant from the substation, with multiple trains between it and the source. That scenario can govern feeder reinforcement, substation spacing, transformer impedance, or the need for a revised sectioning strategy.

Account for power factor, harmonics, and unbalance

Modern traction converters can operate with improved power factor compared with older equipment, yet reactive demand should not be assumed away. The actual power factor during acceleration, low-speed operation, auxiliary loading, and regeneration should be obtained from the rolling-stock electrical data or validated simulation assumptions. A lower power factor raises current for the same MW demand and therefore consumes more transformer MVA capacity.

Harmonic currents require a separate assessment because they create additional losses in windings, structural parts, and connected equipment. The transformer should be evaluated for the expected harmonic spectrum of the traction fleet, including the possibility of mixed rolling stock during transition periods. A standard utility transformer design may not automatically suit a traction duty with significant converter-generated harmonics, rapid load variation, and frequent thermal cycling.

For single-phase railway electrification supplied from a three-phase utility system, phase balance is also a central issue. The selected transformer connection and network arrangement must address negative-sequence effects and the utility’s allowable unbalance conditions. A transformer rated adequately in MVA can still be unacceptable if the supply configuration creates excessive phase imbalance upstream.

Regenerative braking reduces energy, not always peak transformer duty

Regenerative braking is often credited as a direct reduction in substation capacity, but that conclusion is only valid when another load can absorb the returned energy at the same time. A braking train can supply accelerating trains in the same electrical section, reducing net demand at the transformer. If no receptive load is available, the energy may be rejected, limited by vehicle controls, absorbed by onboard resistors, or directed to an energy-storage or grid-export arrangement where such equipment exists.

Therefore, include regeneration as a time-coincident behavior in the load model rather than deducting a fixed percentage from traction demand. Also check reverse-power conditions where the network permits export toward the upstream system. Transformer, protection, metering, and utility-interface requirements may differ when bidirectional power flow is possible.

Translate the study into a transformer specification

Once the load and network calculations are complete, the transformer specification should state more than primary and secondary voltage plus MVA rating. Technical evaluation should normally capture the following items:

• Continuous rating at the defined ambient and cooling condition;
• Permitted cyclic and emergency overload duty, with the assumed load profile clearly stated;
• Primary and secondary voltage ranges, frequency, vector group, insulation levels, and neutral arrangement;
• Specified impedance and its permitted tolerance;
• Tap range, tap changer type where applicable, and voltage-control requirements;
• Harmonic loading assumptions and any required derating or enhanced thermal design;
• Short-circuit withstand capability consistent with calculated fault levels and clearing times;
• Parallel-operation requirements, including impedance matching and control interlocking;
• Cooling class, auxiliary supply needs, monitoring points, and alarm interfaces;
• Space, foundation, transport, fire-protection, and maintenance constraints at the substation.

The overload requirement should be written carefully. A generic request for “overload capability” is not enough because permissible loading depends on duration, starting oil and winding temperatures, ambient conditions, cooling availability, and expected repetition. Define the duty cycle that the transformer must survive rather than assuming every vendor interprets the requirement identically.

Leave room for expansion without buying unusable capacity

Future proofing should be based on identified changes: shorter headways, longer trains, new rolling stock, route extensions, additional branches, or anticipated changes in feeder configuration. It is often more practical to provide civil space, protection provisions, and switchgear bays for a future parallel transformer than to select one very large initial unit that operates lightly for years.

Light loading is not automatically harmless. It can affect efficiency, loss economics, cooling behavior, and the value of installed redundancy. The preferred arrangement depends on the expected growth path and on whether a second unit can be added without prolonged outages or major changes to the bus arrangement.

Before approving the final traction power transformer rating, reconcile the transformer thermal study with the traction load-flow results. The selected unit should pass the normal and contingency demand profiles, maintain acceptable rolling-stock voltage at the critical locations, remain within fault-duty constraints, and have a defined response to regeneration and harmonic loading. Treating those checks as one coordinated decision prevents the common mistake of selecting an MVA rating that looks sufficient on a schedule but fails under the operating conditions the railway will actually impose.