Transformer

What it is (precise)

A BESS step-up transformer is typically an oil-filled or cast-resin dry-type three-phase unit, packaged as a skid-mounted or container-integrated MV power block serving one or more PCS units. The low-voltage winding ties to the inverter terminals; the MV winding feeds the collection feeder. The LV winding is usually delta; MV practice varies by market.

Many plants use a grounded-wye MV winding, vector group YNd11 (capitals denote the higher-voltage winding), which grounds the collection system directly, while others use Dyn11 units (delta on the MV winding) plus a separate collector grounding transformer. Either way, the delta winding blocks zero-sequence (ground-return) currents and triplen harmonics (3rd, 9th, 15th... harmonics, which add up in the neutral instead of cancelling) from passing between windings.

Many modern units are dual-secondary, with two galvanically separated LV windings so a single MV transformer serves two inverter blocks. Larger sites add one or more MV-to-high-voltage main power transformers at the project substation, for example 34.5 kV to 115, 138, or 230 kV, aggregating many collection feeders up to transmission voltage at the POI.

Vocabulary matters here: the small dedicated unit at each PCS is the inverter, step-up, or pad-mount transformer; the large station unit is the GSU or main power transformer; and a separate small auxiliary transformer feeds HVAC, controls, and battery management loads. The MV / LV / HV boundaries these units span define the whole electrical architecture of the plant.

Why it matters in a real grid-scale project

The transformer sets the impedance, fault behavior, and losses between the inverter and the grid, so it is central to both the interconnection study and project economics. Its short-circuit impedance shapes the fault current the grid sees, the voltage drop under load, and the harmonic filtering between the PWM-switching inverter and the network. Utilities scrutinize all of these in the system impact study, and the transformer model (impedance, X/R ratio, winding configuration, grounding) is a mandatory input to every short-circuit and load-flow case the interconnection process runs.

Commercially, transformers are long-lead, high-cost items that frequently sit on the critical path. Lead times on the order of 12 to 18 months are common for MV units, and HV station transformers have run 24 to 36 months or longer. Their losses erode round-trip efficiency and revenue in two distinct ways: no-load (core) loss runs all 8,760 hours a year whether or not the plant dispatches, while load (copper) loss scales with the square of throughput. A battery plant that cycles once a day spends most of its life energized but idle, so no-load loss is disproportionately expensive.

Sizing follows apparent power because S squared equals P squared plus Q squared; the misconception below walks through the arithmetic. The practical consequence is that the transformer must be selected from the interconnection agreement's reactive-power requirements at the POI, plus margin for the reactive power the transformer and collection system themselves consume, not from the plant MW nameplate. Undersizing silently clips the plant's P-Q capability at exactly the operating corners the grid code cares about, while gross oversizing wastes capital and adds permanent no-load losses.

Typical values and standards

Typical unit ratings track the PCS blocks they serve: roughly 1 to 5 MVA per inverter block, with dual-winding skid transformers commonly in the 4 to 8+ MVA range. Short-circuit impedance is typically 5.5 to 8 percent on the unit's own base.

No-load losses run on the order of 0.1 to 0.2 percent of rated power and load losses roughly 0.5 to 1 percent at full load, giving full-load efficiencies typically around 99.0 to 99.4 percent; the 99.5-percent-plus figures often quoted are DOE-style efficiencies referenced to 50 percent load. Standard thermal design is a 65 °C average winding rise over a 30 °C average ambient for IEEE-market liquid units, with cooling classes such as ONAN/ONAF (oil) or AN/AF (dry-type).

Anchor the design to recognized standards in their correct roles. In North America, IEEE C57.12.00 covers general requirements for liquid-immersed transformers, IEEE C57.12.90 is the test code, IEEE C57.91 the loading guide for thermal limits and insulation aging, and IEEE C57.110 the method for de-rating under nonsinusoidal (harmonic-rich) load current. The IEC 60076 series is the international equivalent.

US DOE efficiency rules (10 CFR Part 431) set minimum efficiencies for many distribution-class units. For siting, the BESS installation as a whole follows NFPA 855, while oil-filled-transformer separation distances are governed more directly by the NEC, IEEE 979, NFPA 850, and the local fire code; many developers specify ester fluids or dry-type units to ease placement near battery enclosures.

Because the LV winding sees PWM inverter current rather than a clean sine wave, harmonics are a design input, not an afterthought: specifications either cite the PCS harmonic spectrum against IEEE C57.110 or call for a K-factor rated design (a transformer built with extra thermal capacity to carry a specified harmonic-rich current spectrum without overheating). Other standard datasheet values include off-load tap changers at plus/minus 2 times 2.5 percent, basic insulation levels (BIL) of roughly 95 to 200 kV for common MV classes, and guaranteed sound power levels for permitting.

How it shows up in specs, studies and contracts

On a datasheet, check the rated MVA at each cooling stage, impedance and its standard tolerance (plus/minus 7.5 percent per IEEE for two-winding units, which matters because fault current and voltage regulation move with it), vector group, tap range, BIL, temperature rise, guaranteed no-load and load losses, and the fluid type. Ask explicitly whether the rating is valid for the site ambient and altitude, and whether the unit is specified for inverter duty with the actual PCS harmonic spectrum, cycling profile, and any short-time overload the energy management strategy assumes.

In interconnection work, the transformer appears as impedance and X/R data feeding short-circuit, load-flow, and stability models, and as part of the reactive-capability demonstration at the POI. Energization is its own study topic: inrush current of a large transformer can cause voltage dips and sympathetic inrush (a neighboring, already-energized transformer briefly saturating and drawing its own inrush when a new unit is switched in), and some utilities set explicit dip limits.

Grid-code ride-through performance is also verified with the transformer in the model, since its impedance sits between the inverter terminals, where the PCS controls act, and the POI, where compliance is measured.

Contractually, losses are usually capitalized in bid evaluation, with no-load losses often valued several times higher per kW than load losses because they run continuously; suppliers pay liquidated damages for exceeding guaranteed figures measured in the factory acceptance test per IEEE C57.12.90 or IEC 60076-1.

Round-trip efficiency guarantees measured at the POI include transformer losses, so the sizing engineer, the PCS supplier, and the performance-guarantee owner must use the same loss model. Delivery dates carry schedule liquidated damages precisely because the transformer so often gates energization.

Common pitfalls

The classic errors are unit and reference-point confusions. The MW-versus-MVA sizing trap (see the misconception below) typically stays hidden until commissioning power-factor tests fail. Quoting datasheet efficiency at rated load as the in-service figure is equally misleading: a BESS spends long periods at partial output or idle, where no-load loss dominates and effective efficiency is lower. And the auxiliary transformer is a separate device with its own losses that count against round-trip efficiency at the POI.

Watch also for off-load taps left at the factory setting rather than the value the load-flow study assumed, which shifts collection-bus voltage and can push PCS units toward their voltage limits; for harmonic heating on units bought as generic distribution transformers without inverter-duty review; and for spare-part strategy, since a failed step-up transformer with an 18-month replacement lead time can strand an entire multi-MW block. All of these are recurring findings in real project reviews.