How a Grid-Scale Battery Makes Money

Revenue stacking: why one market is never enough

The economics of a grid-scale battery rest on a simple fact: no single market pays enough to underwrite the asset alone. Arbitrage is variable and price-driven — it pays only when spreads are wide. Ancillary services pay for availability, so the battery earns by standing ready whether or not it is ever dispatched. The capacity market pays a fixed annual sum for firming, regardless of activation. Each stream carries a different risk and payment basis, and layering them is what turns a battery into a bankable project.

The catch is that all three streams share one physical battery, one day, and one state-of-charge curve. Energy committed to a capacity obligation cannot also be sold into the peak; power reserved for frequency response cannot simultaneously chase an arbitrage spread. Revenue stacking is therefore an optimisation problem, not simple addition — the operator co-schedules the streams against the battery's real power (MW) and energy (MWh) limits.

The Revenue Stacking visual makes this concrete: it traces state of charge across 24 hours for a 50 MW / 200 MWh battery and splits the daily total across arbitrage, Dynamic Containment, and a fixed Capacity Market payment on one shared curve.

Energy arbitrage: buy low, sell high

Arbitrage is the most intuitive stream: the battery charges when wholesale power is cheap and discharges when it is expensive, banking the price spread. The size of that spread is set by the shape of net demand — total demand minus variable renewable output. As solar has grown, midday prices have collapsed while the post-sunset evening peak stays tall, widening the spread a battery is built to capture.

Two physical realities shape the arbitrage math. First, round-trip efficiency (roughly 85–90% for lithium-ion, approximately charge efficiency times discharge efficiency) means the battery must buy more energy than it sells — some of every megawatt-hour is lost as heat. Second, a battery has a fixed duration: a 50 MW / 200 MWh unit is a four-hour battery (0.25C), so it can only shift a bounded quantity of energy per cycle no matter how attractive the spread.

The Duck Curve visual is the clearest single picture of why arbitrage exists: it shows net demand sinking into a midday belly and climbing a steep evening ramp, then a battery charging on the cheap midday surplus and discharging into the expensive evening peak.

Ancillary services: paid to stand ready

Ancillary services are the products a system operator buys to keep the grid stable second to second — chiefly frequency response. Crucially, most of these pay for availability across a contracted window rather than for energy delivered, so a battery can earn while sitting mostly idle, holding headroom to inject or absorb power the instant frequency deviates. That availability model is why frequency response often out-earns arbitrage per MW: the battery is paid for its speed and readiness, not its throughput.

Batteries dominate these markets because inverters can move from zero to full power in tens of milliseconds. The binding constraint is not the hardware but the measurement-and-command chain — detecting the deviation, routing the command through the plant controller to the inverter, and delivering inside the market's response window. In ERCOT, Fast Frequency Response (FFR) requires full delivery within 0.25 seconds; in Great Britain, Dynamic Containment plays the equivalent post-fault role.

The ERCOT FFR Dispatch visual walks this latency budget: an under-frequency trigger travels the POI meter → plant controller → energy storage controller → inverter chain, and the example lands full power well inside the 0.25 s window with margin to spare.

The capacity market: a fixed payment for firm availability

The capacity market pays a battery a fixed sum — typically an annual rate per de-rated MW — for guaranteeing it can deliver during system stress events, independent of whether it is ever called. This is the most predictable stream and the one lenders like most, because it is contracted years ahead and does not depend on volatile spreads or activation counts.

The obligation has a direct cost, though. A capacity agreement requires the battery to keep a minimum quantity of energy available to discharge during a stress event — a reserve floor on state of charge. When that floor sits above the trough the operator would otherwise discharge to, the held-back energy cannot be sold into the peak. So a larger capacity commitment mechanically reduces arbitrage volume while the capacity payment itself stays fixed. That trade-off between the reserve floor and arbitrage is exactly what the Revenue Stacking visual's gold reserve line exposes.

The constraint that ties the streams together

Everything above is bounded by two independent ratings that are easy to conflate. Power (MW) is set by the PCS and grid connection and caps how fast the battery can charge or discharge. Energy (MWh) is set by the battery containers and caps how much it can store. Their ratio is duration — 200 MWh at 50 MW is four hours — and it decides how many streams can physically co-exist. A short-duration battery is better suited to fast frequency response; a longer-duration one can hold energy back for a capacity obligation and still arbitrage the peak.

State of charge is the shared ledger. Every megawatt-hour reserved for a capacity floor, consumed by round-trip losses, or committed to a discharge window is unavailable to the others. Optimising revenue is really about scheduling these claims against one SOC curve without violating power limits, energy limits, or the degradation budget that governs how hard the asset can be cycled over its life.

The markets differ by grid — and so does the stack

Revenue stacking is grid-specific because the products and their rules are set by each system operator. In Great Britain (50 Hz), the stack typically blends day-ahead and intraday arbitrage, the Dynamic frequency-response suite (Dynamic Containment, Moderation, and Regulation), and the Capacity Market. In ERCOT (60 Hz), it blends real-time energy arbitrage with Fast Frequency Response, Responsive Reserve Service, and ERCOT Contingency Reserve Service — and ERCOT famously has no forward capacity market, so that stream is absent and arbitrage plus reserves carry more weight.

The practical consequence is that a battery's business case cannot be lifted from one market to another. The same 50 MW / 200 MWh hardware earns through different products, on different payment bases, under different response windows depending on where it connects. Understanding which streams a given grid offers — and how they compete for the same state of charge — is the first step in sizing and scheduling a project.