MVA

What it is (precise)

Apparent power in MVA is what the hardware physically feels: it is proportional to RMS current at a given voltage, S = √3 × V(L-L) × I for a three-phase system. Real power (MW) does useful work; Reactive power (MVAR) supports voltage but circulates without net energy transfer. The two add as vectors, and MVA is the hypotenuse of that right triangle. The ratio MW/MVA is the Power factor, so a PCS delivering its full MVA at 0.95 power factor produces only 95% of that figure as MW. Memorize the triangle and most sizing questions answer themselves.

Heating in windings, IGBTs, busbars and cables depends on current, not on whether that current carries watts or VARs, so the MVA value, not the MW value, sets the thermal envelope. A 4.0 MVA PCS pushing 4.0 MW at unity power factor and the same unit pushing 3.2 MW plus 2.4 MVAR draw essentially the same current and dissipate essentially the same I-squared-R losses. This is precisely why PCS units, MV skid transformers and the main power transformer are all rated and purchased in MVA or kVA, never in MW alone. Where you meet it first is the datasheet apparent-power line.

The rating is voltage-referenced. A converter is fundamentally a current-limited device, so its deliverable MVA scales with the AC voltage it operates at: hardware listed at 4.4 MVA at 690 V delivers proportionally less apparent power if configured for a lower AC output voltage.

Because MVA is an AC-side quantity while battery power is quoted on the DC side, the AC / DC conversion sits between the two ratings. A Nameplate MVA is therefore meaningless without its reference voltage, frequency, ambient temperature and altitude, which is exactly why vendors publish it as a matrix rather than one headline number.

Why it matters in a real grid-scale project

Grid codes increasingly require BESS plants to supply reactive power and ride through voltage events, so the PCS rarely runs at unity power factor at the POI. If the interconnection agreement demands 0.95 leading and lagging power factor at full MW output, a common transmission-level requirement, the plant needs roughly 1.05 MVA per MW at the POI, plus margin for transformer and collection-system losses upstream.

Sizing on MW alone is one of the most common ways a project ends up unable to meet its reactive obligation without curtailing real power, which is the moment the system operator most needs voltage support.

Commercially, MVA drives capital cost and the interconnection study. Transformers and switchgear are priced per MVA; protection settings, cable ampacity, and continuous and short-time overload capability are all expressed against the MVA limit.

Underrating causes nuisance trips, hot-weather derating, or an inability to deliver contracted ancillary services; overrating wastes capital on copper and iron the revenue model never uses. MVA also feeds short-circuit and stability studies, because fault-current contribution from inverter-based resources is capped near 1.0 to 1.2 per unit of the converter's rated current, far below a synchronous machine of the same MVA.

Notice what MVA does not tell you: nothing about energy or Duration. A 100 MW / 400 MWh project is a 4-hour, 0.25 C-rate plant whether its converters total 105 MVA or 120 MVA; the extra apparent-power headroom buys reactive capability and thermal margin, not one additional megawatt-hour. Apparent-power sizing and Energy sizing are separate engineering exercises answering separate contract clauses. A change to one does not automatically justify a change to the other, and confusing the two is a fast route to a mis-specified plant.

Typical values and standards

Container- and skid-class PCS units commonly land around 1 to 5 MVA each, aggregated through medium-voltage transformers of typically 2 to 8 MVA per skid, then up to main power transformers rated from tens of MVA to 100 to 400-plus MVA for large plants.

Large oil-filled transformers carry multiple MVA ratings on one nameplate for the same unit at different cooling stages (ONAN/ONAF/OFAF style), so one transformer might read 75/100/125 MVA. PCS ratings are reference-condition figures: expect derating above roughly 25 to 50 C ambient (vendor-specific) and above about 1000 m altitude, so always confirm continuous MVA at the actual site conditions and AC voltage.

The requirements that drive MVA sizing come from the interconnection framework: IEEE 2800 for inverter-based resources on transmission systems and IEEE 1547 for distribution-connected systems both specify reactive-capability envelopes, typically on the order of 0.95 power factor or a defined Q/P envelope at full output. Regional grid codes (ERCOT, CAISO, ENTSO-E members) add their own reactive and ride-through demands.

For the equipment, the IEEE C57 family and IEC 60076 govern transformer MVA and loading, while IEC 62933 covers the grid-integrated storage system. Safety standards constrain deployment but not the electrical math: UL 9540 is the ESS system listing, UL 9540A is the separate fire-propagation test method that feeds NFPA 855, the US installation standard.

How it shows up in specs, studies and contracts

On a PCS datasheet, read the apparent-power rating as a matrix, not a single number: continuous kVA or MVA versus AC voltage, ambient temperature and altitude, usually with a separate short-time overload figure such as 110% for 10 minutes. Confirm that the quoted MW at unity power factor and the MVA at rated power factor refer to the same operating temperature. On a transformer nameplate, check which cooling stage each MVA figure corresponds to and what the impedance is at the base MVA, because per-unit impedance quoted on the wrong base silently corrupts fault studies.

In interconnection studies, the plant is characterized by its MW and MVAR capability at the POI, net of all losses; the reactive-capability (D) curve submitted to the operator is drawn in the P-Q plane, and its outer boundary is the aggregate converter MVA minus collection-system consumption.

Ask on every project: is the required power factor defined at the POI or at the inverter terminals? Is reactive capability required at zero MW output, since many codes now demand STATCOM-like behavior from an idle BESS? What is the continuous MVA at the site's worst-case ambient? Each of these has caught real projects out.

In supply contracts and warranties, MVA appears as the guaranteed continuous rating at stated reference conditions, and the remedy language should tie to that exact figure. Never accept a nameplate MW number as a proxy: nameplate MW at unity power factor, deliverable MVA at site conditions, and contracted capability at the POI are three different numbers, and the gaps between them are where disputes live. This mirrors the familiar Nameplate versus Usable energy distinction on the energy side, now applied to the power path, so demand the reference voltage, temperature and altitude in writing.

Common pitfalls

The classic unit errors: confusing MVA with MW, which coincide only at unity power factor; confusing MVA with MVAR, the whole hypotenuse versus one leg; and confusing any of them with MWh, which is energy, not Power.

A subtler trap is mixing AC and DC reference points, because converter MVA is an AC-side quantity while battery power is quoted on the DC side, so the AC / DC conversion losses sit between them. Finally, remember that aggregate installed converter MVA is not automatically deliverable MVA at the POI: transformer impedance, collection-system losses and auxiliary loads each take a cut before the meter.