There’s a product your battery plant sells that never passes through the battery. It doesn’t consume fuel, doesn’t drain the cells, and yet grids pay for it and interconnection agreements demand it. It’s reactive power — the least intuitive concept in AC engineering, and worth fifteen minutes of anyone’s life to actually understand.
For a visual companion, explore the interactive Reactive Power Triangle diagram on BESS.Engineer.
Beer, foam, and the power triangle
In an AC system, voltage and current are both waves. When they rise and fall together, every watt delivered does useful work — that’s active power (P), measured in watts, the thing that spins motors and charges batteries.
But real grids are full of inductive equipment — motors, transformers — whose magnetic fields make current lag behind voltage (capacitive equipment makes it lead). When the waves are out of step, part of the flow just sloshes back and forth each cycle, building and collapsing fields without doing net work. That’s reactive power (Q), measured in VARs.
The classic image: a mug of beer. The liquid is active power — what you actually wanted. The foam is reactive power — takes up space in the mug, isn’t drinkable, yet the mug (your cables, transformers, and inverters) must be sized for liquid plus foam. The whole mug is apparent power (S), in VA, and the three obey a right triangle: S² = P² + Q². The ratio P/S is the power factor, cos φ — the fraction of your capacity doing real work.
Why the grid needs the foam
Here’s the twist that makes reactive power respectable: it isn’t waste — it’s what holds voltage up. Transmitting power across an inductive network consumes VARs along the way; without reactive support injected in the right places, voltage sags, equipment strains, and in the worst cascades, collapses. Reactive power barely travels — it’s needed locally — which is why grids can’t just make it somewhere central and ship it.
Historically, synchronous generators provided VARs as a by-product. As they retire, the job transfers — and inverter-based plants inherited it.
The part everyone gets wrong about batteries
Batteries do not store reactive power. Inverters manufacture it. A cell is a DC device; φ means nothing to it. But the PCS — which constructs its AC output waveform from scratch — can shape the phase relationship between its current and the grid’s voltage at will. Command it to lead or lag, and it injects or absorbs VARs with electronic precision, at almost no cost in stored energy (only the small resistive losses of pushing extra current).
Three consequences worth internalizing:
- VARs at zero state of charge. A “flat” battery can still provide full reactive support — voltage service without touching the MWh.
- The capability curve is a circle, roughly. The inverter’s total current is the hard limit, so at full active power there’s little headroom for Q, and vice versa. Grid codes specify the P–Q envelope you must deliver; oversizing inverters slightly buys reactive headroom.
- It’s a revenue and compliance item. Interconnection agreements set power-factor or voltage-control obligations, and several markets pay for reactive/voltage service — one of the quieter lines in the storage revenue stack, and a headline one for grid-forming plants whose voltage-source behavior makes them natural stabilizers.
The one-paragraph physics
For the mathematically inclined: with phase angle φ between voltage and current, P = V·I·cos φ and Q = V·I·sin φ. An inductive load takes current lagging voltage (positive Q consumed); a capacitor supplies it. An inverter simply chooses φ. That’s the entire mechanism — everything else is consequences. It’s also a purely AC phenomenon; DC circuits, with no alternation, have no phase and no Q.
FAQ
Does providing reactive power drain the battery? Essentially no — only marginal conversion losses. The energy sloshes between the inverter and the grid’s fields each cycle rather than being consumed.
What’s a good power factor? Unity (1.0) means all apparent power is active. Grid codes typically require plants to operate across a band around unity — commonly somewhere in the 0.9–0.95 leading/lagging territory, per your interconnection agreement.
Can a BESS fix low voltage in an area? Often, yes — injecting VARs locally raises voltage, which is precisely why storage gets sited and paid for voltage support in weak parts of networks.
P, Q, S and the rest of the AC toolkit get the full intuitive treatment in my Grid-Scale BESS: Complete Guide.