Transformer Sizing for PV and BESS: What Trips People Up

Sizing off the sum of inverter nameplate ratings overshoots the real requirement, since DC/AC ratios of 1.2 or higher mean the inverter rarely pushes full nameplate outside a handful of clipping hours. Realistic continuous AC output — based on site irradiance data, plus whatever power factor the utility requires — is the number that matters.

BESS is different again, since power flows both directions and, depending on the application (arbitrage, frequency regulation, peak shaving), the unit can sit near rated power far more often than a PV plant ever would.

Inverters and PCS units are switching devices, and harmonic current still flows into the transformer regardless of how clean the filtering looks on paper — driving extra eddy current and stray losses that a plain 60 Hz sine-wave design never accounted for. Actual THD data at partial load, not just rated output, matters more than most vendor sheets suggest.

A K-factor rated unit, or a derating calculated per IEEE C57.110, is the difference between normal insulation life and replacing it ten years early.

Most ground-mount utility-scale sites still run oil-filled, pad-mounted units — cheaper per kVA, better overload handling. BESS sites increasingly push toward dry-type or ester-fluid transformers even at a cost premium, mainly because it simplifies the fire protection conversation with the AHJ when the unit sits next to battery enclosures. Rooftop and carport PV is almost always dry-type given the installation location.

On sites that run hot, or where the load profile sits near rated output often, ONAN/ONAF instead of plain ONAN is usually worth the extra cost — the fan stage buys roughly 15–30% more capacity for a small fraction of total transformer cost. Ambient temperature and altitude correction factors are worth double-checking too; this gets missed constantly on desert and high-elevation sites.

Impedance should be worked out with a protection engineer before finalizing anything — too low and downstream fault currents get nasty, too high and voltage regulation problems show up, which matters more on PV because of cloud transients and on BESS because of fast power swings during cycling. Vector group matters too: Dyn11 (or Dyn1, region dependent) is the usual choice for step-up transformers, since it provides a grounded reference point that most inverters and PCS units need on their side.

Unless there's a specific requirement for active voltage regulation as a grid service, an on-load tap changer usually isn't worth the added cost and maintenance burden. A de-energized tap changer, set once at commissioning based on measured site voltage, is sufficient for the large majority of PV and BESS installations.

Nameplate ratings shouldn't just be added together. The actual coincident power flow needs modeling, since PV generation and BESS charge/discharge rarely peak in the same direction at the same time — and the real ceiling is usually the POI interconnection limit, not the sum of equipment nameplates.

A loss evaluation — no-load losses and load losses, capitalized — is worth requesting before picking a transformer on price alone. Over a 20–30 year project life, losses can outweigh the purchase price difference between two otherwise similar units. Utilities usually have their own loss evaluation formula available on request.