Four-quadrant operation

What it is (precise)

Put real power on one axis of the P-Q plane and reactive power on the other. The four quadrants are: discharge while injecting VARs, discharge while absorbing VARs, charge while injecting VARs, and charge while absorbing VARs. A BESS PCS that can reach all four is four-quadrant. It gets there by controlling the magnitude and phase of its AC current relative to grid voltage, placing the operating point anywhere inside the capability boundary, including pure reactive support at P = 0. Both Grid-following and Grid-forming controls can do this; the quadrants describe capability, not control mode.

The boundary is the apparent-power rating: S = √(P² + Q²), so P and Q share one MVA number and pushing more VARs leaves fewer megawatts of headroom. State of charge constrains only the P axis. Reactive power circulates between the grid and the inverter's switching stage each half-cycle without depleting stored energy, so a four-quadrant PCS can hold voltage up at P = 0 with nearly empty cells. The DC side sees only ripple, which is why SOC and reactive capability are tracked separately.

Why it matters in a real grid-scale project

Interconnection agreements and grid codes increasingly require BESS plants to supply and absorb reactive power on demand, not just shift energy. Four-quadrant capability lets one asset do voltage regulation, power-factor correction, and dynamic VAR support alongside its arbitrage or Frequency response duty, often replacing a separate STATCOM or capacitor bank. Most US ISOs treat reactive supply inside the required band as an Interconnection obligation, not a paid service, so budget it as a compliance cost, not revenue.

The trade-off is sizing. Because VARs consume MVA headroom, you either oversize the PCS to meet its reactive range at full real power, or accept that maximum P and maximum Q cannot occur together. Meeting full rated MW at 0.95 power factor takes roughly 5% extra MVA (1/0.95 = 1.053); at 0.90 it is about 11%. That single decision flows into converter block count, the Transformer rating, capex, and the P-Q capability curve you submit in the interconnection study.

There is an idle-hours dimension too. A four-quadrant plant sitting at 0 MW overnight can still follow a voltage schedule, and operators increasingly write that into agreements: reactive capability that holds at zero real-power output. For the owner this turns dead hours into grid support; for the designer it means the PCS, its controls, and its auxiliary supply must be specified to run continuously in Q-only mode. That continuous VAR duty belongs in the auxiliary-load budget and round-trip-efficiency accounting, not just the energy-dispatch case.

Typical values and standards

In the US, FERC Order 827 requires new non-synchronous generators, including BESS, to provide reactive power across a 0.95 leading to 0.95 lagging power-factor range, referenced to the high side of the generator substation.

For transmission-connected plants, IEEE 2800-2022 defines the performance envelope for inverter-based resources: continuous reactive capability, voltage-control response, and Ride-through behavior. Many international grid codes demand more, often down to about 0.90 power factor, and several specify a rectangular P-Q envelope in which full Q must stay available across most of the P range.

On distribution circuits, IEEE 1547-2018 governs. It defines the reactive control modes a DER must offer (constant power factor, volt-VAR, constant reactive power, watt-var) and sets minimum capability: injection up to 44% of nameplate apparent rating for both performance categories, absorption of 25% for Category A and 44% for Category B. A utility-scale BESS interconnecting at medium voltage under 1547 is typically Category B. These are floors; the physical converter circle usually offers more, so read the datasheet.

Physically, the capability curve of a typical 1-5 MVA utility PCS block is a circle (or slightly clipped circle) of radius equal to its MVA rating, but only at nominal voltage and moderate ambient. Because the converter is current-limited, available apparent power scales with AC voltage: at 0.90 pu grid voltage the same current delivers about 10% less MVA, which is exactly when the grid wants maximum VAR injection. High ambient temperature derates it further, and reactive current heats the switching stage like real current, adding small but real losses.

How it shows up in specs, studies and contracts

On a PCS datasheet, look for the P-Q capability chart as a family of curves versus AC voltage and ambient temperature, not a single circle. Check that the advertised power-factor range holds at maximum ambient and at 0.90 pu voltage, and confirm reactive capability at zero real power: some grid-following firmware needs the DC bus energized and a minimum enable state to make VARs at night, so Q-at-zero-P may be a configurable option rather than a default. Ask how it behaves at the curve edge: current-limit, derate, or trip.

In interconnection studies, the reactive capability you submit must be net at the Point of Interconnection, not at converter terminals. The step-up Transformer absorbs reactive power roughly as current squared times its reactance; with typical 6-8% impedance on the transformer's own base (larger units up to ~10%) that shift is material at full load, while MV collection cables generate VARs at light load across the MV / LV / HV chain.

Power-flow and dynamic studies then verify the plant holds its voltage schedule across the committed envelope. Ask early which reference point the grid code uses and whether the envelope holds across the full ambient and voltage range.

In contracts and market rules, the LGIA or its local equivalent fixes the reactive obligation, and commissioning tests demonstrate operation at agreed points of the capability curve, typically full lead and full lag at several real-power levels, witnessed and logged in the test report. Availability guarantees may count Q-only hours as operating hours, which changes warranty duty-cycle assumptions. If the plant will run heavy continuous VAR duty, tell the PCS vendor up front: thermal cycling of the switching stage is driven by total current, and Q current counts the same as P current.

Common pitfalls

The classic trip-wire is specifying power factor where reactive power is meant. Power factor is undefined at P = 0, so a requirement like "0.95 PF at all outputs" collapses at low power; well-written codes specify MVAr (or a percent of rated MVA) at zero and partial output instead. Related: a power-factor band quoted only at rated power says nothing about the rest of the envelope, and the gap between a circular and a rectangular P-Q requirement can be several percent of converter rating.

Second, do not conflate four-quadrant with grid-forming: grid-following inverters are equally four-quadrant, and GFM versus GFL describes how the converter behaves as a voltage or current source, not where its operating point can sit. Third, do not confuse reactive headroom with real-power Clipping / curtailment; a plant clipped on P may still have full Q available, and vice versa. Finally, remember auxiliary load: holding VARs at P = 0 still runs controls, cooling, and switching losses — cheap, but not free, and it belongs in the auxiliary-power and RTE accounting.