BESS Revenue Stacking Explained

Four streams dominate discussions of a merchant, hybrid, or customer-sited battery: energy arbitrage, ancillary services, capacity payments, and behind-the-meter bill savings. Which of them are actually payable depends on the market rules and on where the project interconnects.

Energy arbitrage

Energy arbitrage is charging when electricity is inexpensive and discharging when it is expensive. NREL defines it that way for the operator, and notes a related use: charging on surplus renewable generation that would otherwise be curtailed, then discharging when demand is higher. [2] The U.S. Energy Information Administration uses the same definition: buying electricity when prices are relatively low and selling it when prices are high. [3]

In EIA's survey of 2024 operations, operators reported that 66% of U.S. utility-scale battery capacity had arbitrage among its uses, and that 41% of total capacity was primarily used for arbitrage. Frequency regulation was the next most common primary use, at 24% of capacity. In earlier surveys, frequency regulation had been the most common use. At the end of 2024, the California Independent System Operator (CAISO) reported 11.7 GW of battery capacity, 43% of it primarily used for arbitrage, and the Electric Reliability Council of Texas (ERCOT) reported 8.1 GW, half of it primarily used for arbitrage. [3]

A spread on a price chart is not the spread the project keeps. Round-trip efficiency is the ratio of energy discharged to energy charged, including self-discharge and other electrical losses. NREL notes that the AC-to-AC figure at the point of interconnection is the one the power system sees, which is often lower than the DC-to-DC figure quoted by a cell vendor. [2] Charging cost, those losses, and any limit on lifetime throughput all sit inside the arbitrage decision. Day-ahead and real-time spreads are covered in How battery arbitrage works.

Ancillary services: frequency regulation and reserves

Ancillary services pay the battery for grid support. Some of that payment is for energy actually moved. Much of it is for being available.

Frequency regulation keeps system frequency near its target. EIA describes it as maintaining the grid at 60 cycles per second, and notes that batteries suit the service because they need no startup time and can absorb or inject power quickly. [3] DOE's valuation review separates that continuous product from contingency reserves. Regulation is exercised through the hour as the control signal moves. Contingency, or operating, reserves are called after a sudden loss of generation or transmission. Spinning reserve comes from resources already synchronized that can increase output immediately and reach full output quickly. Non-spinning reserve is the slower contingency product. Spinning reserve is generally not deployed unless a contingency occurs. [1]

NREL's service table puts regulation on a timescale of about 15 minutes to an hour, and contingency spinning reserve on a timescale of about 30 minutes to two hours. Primary frequency response is faster still, on the order of seconds, and is paid in only a limited number of markets. [2] Speed and accuracy change the value. FERC Order 755 requires frequency-regulation pay to reflect what the order calls "the inherently greater amount of frequency regulation service being provided by faster-ramping resources." Order 841 later directed wholesale market operators to write rules so storage can participate in energy, capacity, and ancillary-service markets, taking account of how storage actually operates. [2]

Holding a megawatt for regulation or reserves uses headroom. Upward service needs room to increase discharge or reduce charging. Downward service needs room to do the opposite. That headroom cannot also be sold as energy in the same direction at the same time. Because regulation deployment is uncertain, the state of charge also has to cover the energy the signal might take before the battery can recharge.

Capacity payments

Capacity is the ability to deliver electricity when the system is short, not the megawatt-hour itself. DOE describes the related benefit as a capacity payment, a reduction in a capacity charge, or an avoided peaker, depending on whether the region uses a wholesale market, a bilateral contract, or integrated planning. [1]

NREL groups this with firm or peaking capacity: reliable capacity to meet peak demand. In its service table, firm capacity is associated with a duration of four hours or more, and with value that can come through scarcity pricing, a capacity market, or resource-adequacy payments. [2] A shorter battery can still earn energy and reserve revenue. It should not be modeled as if it automatically qualified for the same capacity payment as a resource that can cover the peak window.

A capacity obligation also changes dispatch. If the contract requires discharge across a defined peak, state of charge has to be built before that window, which forecloses some earlier arbitrage. How those obligations differ from an energy payment is the subject of the capacity market primer.

Demand charges and behind-the-meter savings

In front of the meter, the ISO settles energy, reserves, and capacity. Behind the meter, the customer bill is the settlement. DOE's review of storage valuation tools describes behind-the-meter dispatch as minimizing that bill, including energy charges and demand charges. [1]

A demand charge is assessed on the customer's highest demand during the billing period. The useful discharge is the one that cuts that billed peak, which is not necessarily the highest wholesale price of the day. The savings are cash, but they are a reduction in a retail bill. They do not sit on top of a wholesale stack unless the same battery is allowed to do both. The retail peak and the wholesale peak often fall in the same hours. When they do, the battery cannot discharge the same megawatt-hour twice.

Co-optimization chooses a mix of services that respects power, energy, and state of charge. Adding each product's standalone value does the opposite: it double-counts megawatts that can be used only once. DOE puts the difficulty on charging control. Services compete for limited power and energy inside a single interval and across later intervals, so a valuation that ignores that competition overstates the stack. [1]

NREL states the operating consequence. Some services are mutually exclusive for a given design. A short-duration device used for regulating reserves has limited value for a multi-hour transmission deferral. Even when the hardware can offer several products, the battery must prioritize, because it runs out of charge. The California Public Utilities Commission rules NREL summarizes put reliability services ahead of other uses, and they bar a contract for an extra service that would interfere with a reliability obligation. [2]

A workable priority order, before any price forecast, follows those constraints:

1. Cover obligations that carry a performance penalty or a must-offer rule. Resource-adequacy awards and ancillary awards often require enough state of charge to sustain the award for its duration.
2. Reserve headroom for products that pay for availability in that interval.
3. Use the remaining power and energy for the charge or discharge with the best spread after losses.
4. Leave state of charge where the next binding interval needs it, instead of emptying the battery for a smaller spread now.

The grid operator may co-optimize as well. ERCOT's Real-Time Co-optimization plus Batteries design can award energy and more than one ancillary service to a storage resource in the same security-constrained economic dispatch run. Awards are limited by qualification, telemetered capability, state of charge, ramp rate, and each product's duration requirement. ERCOT describes that as a change from leaving state-of-charge coverage entirely to the scheduling entity. [4] Under the duration parameters in that ERCOT overview, regulation service and responsive reserve (excluding fast frequency response) are 30 minutes, and ERCOT contingency reserve service is one hour. Those are protocol settings, not a price forecast, and protocol revisions can change them.

Dispatch priority is therefore two decisions at once: what the market will clear given the offer, and what state of charge the owner is willing to promise. Offering the full nameplate into reserves can clear an availability payment and still block the energy discharge that justified the project.

The figures in this section are round numbers chosen to show the trade-off. They are not ERCOT, CAISO, Chile, or Mexico prices. They are not a settlement, a forecast, or an Optimus result.

Take a 10 MW, 40 MWh battery. NREL defines storage duration as usable energy divided by power, and its example is a 1 MW, 4 MWh battery with a four-hour duration. [2] The 10 MW, 40 MWh unit is that same four-hour shape, scaled up so the arithmetic is easy to read.

The efficiency assumption is also illustrative. NREL defines round-trip efficiency and does not publish a single AC-to-AC value for every project. This example uses 85% only so the losses are visible in the arithmetic.

The battery charges 40 MWh at an illustrative $20/MWh, so the charging cost of a full charge is $800. At 85% round-trip efficiency, energy that can later be discharged is 34 MWh. Consider one later hour in which the energy price is an illustrative $100/MWh:

• If all 10 MW discharge, 10 MWh are sold and the rest stays in the battery. Delivering 10 MWh at 85% efficiency required about 11.8 MWh of purchases (10 divided by 0.85), or about $235 at $20/MWh. Energy revenue is $1,000. The hour's energy margin is about $765, before degradation, station service, and market charges.
• If the same hour clears 4 MW of upward reserve and only 6 MW of energy, energy sales fall to 6 MWh, or $600. Purchases for those 6 MWh are about 7.1 MWh, or about $141. At an illustrative reserve price of $15 per MW for that hour, the reserve payment is $60. The combined margin is about $519.

In this made-up price set, the energy-only hour earns more. A higher reserve price, or a lower energy price, flips the choice. The point of the example is the flip. The megawatt used for reserve is not also available for energy, so a model that adds the maximum arbitrage revenue to the maximum reserve revenue overstates the day.

The energy left after that hour can serve a later hour. A reserve award does not delete it. If the award also requires state of charge to cover 30 or 60 minutes of possible deployment, the energy that is free to sell is smaller than the nameplate 40 MWh. That is the same constraint ERCOT applies when it limits awards by state of charge and by each product's duration. [4]

Degradation

NREL states the limit directly. Even when a battery can technically offer several services, the additional cycling may degrade it and shorten its lifetime and economic viability. [2] A stack that looks best on day-one merchant revenue can be the stack that uses the warranty fastest. Shallow, frequent regulation moves state of charge differently from a deep arbitrage cycle. The question for the operator is whether the extra service payment is larger than the life it uses. This note does not assign a dollar cost per cycle. That cost depends on the cell, the warranty, and the depth of discharge.

Market saturation

Ancillary markets are thin relative to the energy market. In 2019, NREL wrote that most U.S. storage was providing operating reserves and ancillary services, that the total market for those services is limited, and that utility-scale storage was likely to provide more firm and peak capacity as more of it was built. [2] Lawrence Berkeley National Laboratory later described ancillary-service markets as small and quickly saturating with the first wave of storage deployments, and pointed to ERCOT as an example of that trend. [5]

ERCOT's own August 2025 discussion of storage in unit commitment says most storage revenue in the region still comes from ancillary services, while it is gradually shifting to include energy arbitrage as those ancillary markets saturate. The same note puts average storage duration in ERCOT at 1.5 hours, with some two-hour and four-hour projects entering operation, and says the energy price shape has not by itself justified capital investment in duration beyond about two to four hours. [6] EIA's 2024 survey is the companion fact on use, not on dollars: arbitrage is now the most common primary use reported by U.S. operators, and frequency regulation is second. [3] A revenue case that freezes an early ancillary price for the full project life is the case most exposed to saturation.

ERCOT and CAISO do not pay reliability the same way

ERCOT and CAISO are where most U.S. utility-scale battery capacity sits, and they do not stack the same products.

ERCOT does not have a capacity construct. The August 2025 ERCOT note makes that contrast explicit. Regions with a capacity market or a resource-adequacy mandate, CAISO among them, see longer storage durations because duration changes the capacity contribution. ERCOT does not pay that capacity product, so duration is driven by ancillary-service rules and by how many hours of price spread the energy market actually offers. [6] Reliability revenue is carried by scarcity pricing in energy and by ancillary services such as regulation, responsive reserve, and contingency reserve. The co-optimization rules above decide how those products share one state of charge.

CAISO batteries that participate in resource adequacy are paid when they contract with load-serving entities, and they take on must-offer obligations. CAISO's 2024 battery report says batteries in the program must bid their upward resource-adequacy capacity into the market, and that about 65% to 94% of available battery net qualifying capacity was counted toward resource-adequacy requirements, depending on the month. [7] The CPUC qualifying-capacity method, as CAISO restated it in an October 2024 white paper, still requires a resource to be able to deliver its qualifying capacity for four or more consecutive hours. The monthly value passed to CAISO is the megawatt level the storage resource can sustain for at least those four hours. [8]

A stack copied from one region into the other will mis-state both the capacity payment and the hours in which the battery is not free to chase energy.

Optimus does not publish an ERCOT or CAISO revenue stack, and this article is not a model run. On the markets the platform does run, the dispatch simulation charges and discharges the battery on Chile or Mexico prices, under a contract or as a merchant plant. Wear, round-trip efficiency, and a reserve state of charge are applied in order. Energy, capacity, grid charges, and shortfall costs stay separate types rather than being folded into one annual line. A megawatt held for capacity is a megawatt that did not buy low and sell high. In Mexico, wholesale market revenue and an on-site CFE bill stay in separate results.

1. U.S. Department of Energy, Energy Storage Valuation: A Review of Use Cases and Modeling Tools, June 2022. energy.gov
2. Bowen, Chernyakhovskiy, and Denholm, Grid-Scale Battery Storage: Frequently Asked Questions, National Renewable Energy Laboratory, 2019. nrel.gov
3. U.S. Energy Information Administration, "Utility-scale batteries are more commonly used for price arbitrage," Today in Energy, 22 September 2025. eia.gov
4. ERCOT, Real-Time Co-Optimization + Batteries overview, 15 July 2025. ercot.com
5. Lawrence Berkeley National Laboratory, Renewable-Battery Hybrid Power Plants in Congested Electricity Markets, 2024 preprint. lbl.gov
6. ERCOT, Security Constrained Unit Commitment (SCUC)-Based Tool to Manage the Grid with Large Penetration of Energy Storage Resources, August 2025. ercot.com
7. CAISO, 2024 Special Report on Battery Storage, 29 May 2025. caiso.com
8. CAISO, Updated White Paper: Resource Adequacy Processes and CPUC's Slice of Day, 14 October 2024. caiso.com