Ancillary services

What it is (precise)

Ancillary services maintain reliability around the edges of the energy market: holding frequency at its nominal 50 or 60 Hz, keeping voltage inside band, and standing ready to cover the sudden loss of a large generator or transmission line. They are distinct from the wholesale energy a plant delivers and are procured in separate day-ahead or real-time auctions run by the ISO / RTO.

Common products include frequency regulation (continuous fine balancing, e.g. PJM RegD, CAISO regulation up and down), fast frequency response (ERCOT FFR, GB Dynamic Containment), contingency reserves (spinning and non-spinning), and reactive power / voltage support delivered at the point of interconnection (POI).

For a stationary BESS, every one of these products is delivered by commanding the PCS active-power (P) and reactive-power (Q) setpoints inside the plant's apparent-power (S) envelope, where S squared equals P squared plus Q squared.

Fast frequency products are driven either by a local frequency measurement acting through a droop characteristic (typically 2 to 5 percent) or by an operator dispatch signal arriving via the plant controller; regulation follows a centrally computed AGC signal updated every ~2 to 6 seconds in most US markets (PJM ~2 s, CAISO/ERCOT ~4 s, NYISO ~6 s). Reactive support is bounded by the Q-capability curve of the inverters and the grid-code requirement at the POI, not by the battery energy.

Why it matters in a real grid-scale project

Ancillary services are frequently the highest-value, lowest-throughput revenue stream available to a grid-scale battery, which is why merchant projects build their case on revenue stacking: regulation plus reserves plus energy arbitrage, re-optimized every day.

Because fast-frequency products clear on response speed and accuracy rather than energy volume, a battery often earns more per MW reserved here than from arbitrage, while the duty is dominated by shallow cycling rather than deep daily discharge. That has direct engineering consequences: it favors high-power, shorter-duration designs and makes round-trip efficiency, control latency, and state-of-charge (SOC) management the binding constraints rather than raw energy capacity.

The catch is that these markets are shallow and saturate quickly. Total regulation demand in a large ISO / RTO is typically only a few hundred to roughly a thousand MW, so a handful of large batteries can meaningfully move the clearing price.

GB frequency-response and ERCOT ancillary prices both fell steeply once installed battery capacity exceeded product demand; in GB, Dynamic Containment clearing prices dropped by an order of magnitude within roughly two years of launch. A robust merchant vs. contracted analysis therefore treats ancillary revenue as a decaying early-life layer, with arbitrage and capacity market value carrying the later years.

Typical values and standards

Response speed defines the product tiers, and these are the numbers worth memorizing. ERCOT fast frequency response requires full output within about 0.25 seconds (15 cycles) of frequency reaching 59.85 Hz, sustained for 15 minutes; GB Dynamic Containment requires full delivery within 1 second of the deviation; continental-European FCR requires full activation within 30 seconds, symmetric across roughly plus or minus 200 mHz.

Regulation is slower but continuous: the AGC signal updates every 2 to 4 seconds and tracking accuracy is scored, with PJM's performance score directly scaling payment. Contingency reserves sit at 10 minutes (spinning) and 30 minutes (non-spinning). A modern PCS reaches full rated P in tens to a few hundred milliseconds, well inside every window.

Energy requirements follow from sustain duration. Fast-frequency and regulation roles typically need only about 0.25 to 1 hour of usable energy per MW committed, versus the 2 to 4 hours common for arbitrage and capacity accreditation; several markets now impose explicit SOC or duration rules, such as ERCOT's requirement that reserve awards be backed by enough stored energy for the product's sustain period.

Reactive capability is set by interconnection rules rather than markets in most regions: FERC Order 827 requires newly interconnecting inverter-based plants in the US to provide roughly 0.95 leading to 0.95 lagging power factor at the POI, and IEEE 2800 defines the corresponding performance envelope for inverter-based resources.

None of this changes the safety and certification basis of the plant. The same installation still complies with NFPA 855, UL 9540 certifies the ESS product, and UL 9540A large-scale fire-propagation test data informs spacing and fire response. LFP remains the dominant chemistry for stationary storage on thermal-stability grounds, with NMC the higher-energy-density contrast.

Ancillary duty does change the degradation picture: shallow high-rate regulation cycling accumulates throughput and stresses cells differently from deep daily cycling, so the expected regulation mileage must be checked against the warranty's guaranteed throughput before the revenue is banked.

How it shows up in specs, studies and contracts

On the equipment side, ancillary capability appears as PCS response-time and ramp-rate figures on the datasheet: look for the time-to-full-power line (often specified as under 100 ms) and the Q-capability curve across voltage and temperature, then add the plant controller's closed-loop latency, which frequently dominates the inverter itself.

The one question to put to a vendor is measured end-to-end latency from frequency event to POI power change, because summing component numbers understates it. Interconnection studies fix the reactive-power range and voltage ride-through obligations at the POI; the market qualification test then proves the plant can actually follow the regulation signal or the frequency trigger before it is allowed to earn.

Commercially, the term shows up in each ISO / RTO's market qualification rules (product definitions, telemetry, and metering requirements), in market-consultant revenue forecasts that deserve hard scrutiny on their saturation assumptions, and in offtake structures: a tolling or capacity contract may reserve ancillary rights to the offtaker, while a merchant vs. contracted hybrid splits them.

The binding contract check is the battery warranty's throughput and cycle-count terms against expected regulation duty, plus whether the augmentation plan assumes deep arbitrage cycling or shallow ancillary cycling. The two age the cells on different trajectories, and the wrong assumption quietly breaks both the warranty compliance and the pro forma.

Common pitfalls

Two recurring traps. First, sizing on power alone: a product pays per MW, but the operator still requires the SOC headroom to actually sustain it, in both directions for symmetric products, so a battery sitting at 95 percent SOC cannot sell symmetric regulation at full capability and must hold energy in reserve to stay dispatchable.

Second, extrapolating today's clearing prices: ancillary demand is roughly fixed while storage supply grows, so a pro forma that holds current $/MW-h flat for ten years is almost always wrong. Model the stack with frequency regulation revenue decaying toward energy arbitrage and capacity market value as the market saturates, not as a flat annuity.