P-Q capability

What it is, precisely

Every grid-tied PCS has a continuous MVA rating. Because apparent power S is the vector sum of P and Q, no operating point can exceed it: the locus of allowable (P, Q) points is bounded by a circle of radius S_rated. Discharging is +P, charging is −P; supplying reactive power (capacitive, boosting voltage) is +Q, absorbing it (inductive, lowering voltage) is −Q. Four-quadrant operation — the ability to sit anywhere in that plane — is standard for stationary storage converters and is what separates a BESS from most conventional generation.

The pure circle is the ideal. Real capability curves are clipped: a maximum AC-current limit caps S and makes deliverable MVA proportional to AC voltage, DC-bus and battery voltage windows limit how much Q can be pushed at high P, and converter and transformer thermal limits plus battery state of charge constrain sustained P. The result is a rounded-rectangle or D-shaped envelope, usually published as a family of curves at several AC voltages (for example 0.90, 1.00 and 1.10 pu) and ambient temperatures.

Three envelopes share the name, and confusing them is the most expensive mistake available on this page. The first is the normalized OEM curve: both axes in per unit of rated MVA, one chart serving an entire product family, drawn at a stated reference AC voltage and ambient.

Digitized vendor sets look exactly like that — a pu envelope with one trace per AC voltage at 0.9, 1.0 and 1.1 pu, and no MVA figure anywhere on the sheet. What such a chart transfers is a shape. It reads correctly only with the per-unit discipline that belongs to it: pu of what, defined where, given that 1.0 pu is the nameplate at the reference conditions and not at your site.

The second envelope is a specific frame in engineering units, drawn at the inverter's AC terminals — 690 V is the common LV point for utility blocks — at a stated ambient, altitude and DC-voltage window. Here the reference conditions stop being free. One 5 MVA-class 690 V block published at 40 °C and 1000 m, DC bus between 1100 and 1500 V, reaches about 5.38 MW at zero vars and the same 5.38 Mvar at zero real power; the same frame family at 50 °C ambient reaches roughly 4.84 MW and 4.84 Mvar.

The normalized chart behind both is identical. The third envelope is the plant one, measured where the obligation lives: the substation high side or the Point of Interconnection. Vars consumed by the medium-voltage Transformer, the main power transformer and the collector cables come off the terminal total before capability can be set against requirement, and a plant-level P-Q study is what converts one into the other.

Why it matters on a real project

Interconnection agreements and grid codes rarely pay only for energy throughput. They mandate reactive support — a power-factor range at the reference point, or explicit voltage-control duty — that the plant must meet even while charging or discharging at full P. If the PCS fleet cannot reach the required Q at the contracted P, the project fails its functional and witness tests and cannot energize. P-Q capability is therefore a commissioning gate and an Interconnection deliverable, not a datasheet footnote.

It is also a sizing lever with simple arithmetic behind it. Holding 0.95 power factor at full output requires Q ≈ 0.33 per unit of P (tan of arccos 0.95), so the converter needs S ≈ 1.05× the real-power rating before any margin for transformer var losses and voltage derating. That is why a nominal 100 MW plant commonly carries 110–120 MVA of installed PCS capacity.

That difference is the plant's MVA headroom, and temperature derating and transformer vars both spend it before it reaches the POI. Undersize the MVA and the plant either violates its power-factor band or must back real power off — a self-imposed curtailment of the revenue-earning MW.

Reactive capability at or near zero real power — Q-at-night or STATCOM-mode operation — can be a separate, paid service in some markets, and most modern PCS can supply the large majority of rated Mvar at P = 0 provided the DC bus stays energized and controls support it.

The 690 V block above is a clean instance: its Q corner at P = 0 is the full rated value at 40 °C and still about 4.84 Mvar at 50 °C — the same 10 percent the P corner gives up, more than half a megawatt, because the hot envelope is a smaller circle rather than a differently shaped one. Whether the plant runs Grid-following or Grid-forming control does not change the MVA circle itself, but it changes how the envelope is used dynamically, especially how current headroom is allocated during faults.

Typical values and standards

In the US, FERC Order 827 removed the reactive-power exemption for non-synchronous generation: new plants must provide reactive power across a ±0.95 power-factor range at the high side of the generator substation, available across the full real-power output range. Some transmission operators and international grid codes go further, requiring capability equivalent to ±0.90 power factor, and many define a rectangular P-Q window rather than a power-factor wedge — which is exactly what forces the rectangular clipping of the MVA circle.

IEEE 1547 governs distribution-connected resources: its Category B — the usual assignment for storage — requires reactive injection and absorption of at least 44% of nameplate apparent power, available down to low real-power output, plus volt-var and volt-watt control modes.

IEEE 2800 extends comparable and stricter obligations to transmission-connected inverter-based resources, including continuous reactive capability across the voltage range and defined performance at the plant reference point. These standards specify what must be achievable; the vendor curve proves whether a given fleet achieves it.

Typical hardware anchors: utility-scale PCS blocks run roughly 1–5 MVA each on a DC bus at or below 1500 VDC (2000 V emerging), with full MVA usually available up to an ambient in the 40–50 °C range and derating above it. Capability is commonly flat between about 0.90 and 1.10 pu AC voltage and shrinks outside that window because the current limit binds at low voltage and the modulation ceiling binds Q injection at high voltage.

Flat in per unit is not flat in megavolt-amperes, though: wherever the AC-current limit sets the boundary, the circle radius tracks voltage, and one string-PCS characteristic set publishes the same converter as a 236 kVA circle at and above nominal, 225 kVA at 0.95 Un and 213 kVA at 0.90 Un. Always read the derate tables, not just the headline circle.

How it shows up in specs, studies and contracts

On a PCS datasheet, look for the P-Q chart family: at which AC voltages and temperatures is it drawn, is it continuous or short-time, does it apply at all states of charge, and is Q available at P = 0. Add one question to that list — is the chart normalized or in engineering units, and if normalized, what MVA does 1.0 pu stand for, at what reference voltage and ambient.

A pu envelope with no stated base is a picture. In the interconnection study, the plant-level envelope is checked in power-flow cases at the POI, with transformer impedances and collector reactive losses modeled; the utility will test the corner points, not the easy center. The plant controller (PPC) must then actually dispatch the fleet to those corners.

In contracts and market rules, the term appears as a power-factor band in the interconnection agreement, as a reactive-capability demonstration in commissioning test procedures, and sometimes as a compensated voltage-support product. Practical questions to ask on any project: what is the required Q at maximum charge as well as maximum discharge (charging is where designs most often fall short), what happens to the envelope at the site's design-maximum temperature, and who owns the gap if the delivered curve is smaller than the modeled one.

Common pitfalls

The classic trap is comparing a terminal-referenced vendor curve directly against a POI-referenced requirement. Transformers absorb vars roughly in proportion to loading squared times impedance, and an MV skid transformer carries 5.5–8% on its own base (larger units up to about 10%), so a fully loaded 5 MVA block gives up a few tenths of a Mvar at its own step-up before the collector cables and the main transformer take their share.

A fleet that comfortably makes ±0.95 at its own terminals can therefore miss the band at the substation high side at full output; auxiliary loads and long collector runs move the numbers further. The fix is a translated plant-level P-Q study early in design, not a discovery during witness testing.

The same error also runs the other direction, one frame earlier: a normalized family curve accepted as evidence for the frame actually purchased, because the shape looked familiar. The shape is transferable; the megavolt-amperes are not. Any capability claim in a study or a contract should name its frame, its AC voltage, its ambient and altitude, and the bus it is referenced to — four attributes, or the number cannot be checked.

Two more recurring errors: first, treating the kW rating as the kVA rating — converters and transformers are sized in apparent power, and a unit labeled by its MW at unity power factor has no reactive headroom left at that output. Second, assuming the static envelope holds during disturbances: during Ride-through events the converter hits its current limit and control firmware decides the P-versus-Q priority, so dynamic reactive delivery at depressed voltage is a separate, tested behavior — related to, but not read off, the steady-state P-Q chart.