Pricing

29 Oct 2024
Avery Dekshenieks

Part Two: Building a battery energy storage revenue forecast in ERCOT

Modo Energy has developed a forecast for battery energy storage revenues in ERCOT. Following our explanation of the production-cost model, let's examine the second core component of the Modo Energy forecast: the dispatch model.

This model simulates battery operations within certain market conditions, and its custom configurations allow battery owners, operators, developers, and financiers to tailor their revenue forecasts out to 2050, to the specific characteristics of the project they would like to explore.

A forecast user can customize their inputs to see revenue projections for batteries of different sizes, capacities, durations, and locations. This way, they can easily build and compare business cases based on different scenarios. To find out more, schedule a call now.

So, how are these projected battery revenues calculated?

The dispatch model simulates the operations of a single battery energy storage system. In doing so, it calculates the revenues and cycling rates of the battery.

What is the dispatch model?

The dispatch model simulates battery operations and ultimately revenue performance within a set of market conditions. The model performs this in 15-minute intervals to the end of 2050.

The production-cost model produces a set of market conditions. It calculates the generation output by technology type, Ancillary Service procurement, and energy prices in each region for every period. This represents a simulation of the optimal economic dispatch of generation to meet projected demand.

Using the production-cost model outcomes, the dispatch model calculates projected revenues and operations for a battery with custom specifications.

But how does the dispatch model work?

Let’s start with the configuration of the battery you wish to forecast.

What aspects of a battery are customizable in the forecast?

The Modo Energy ERCOT forecast allows users to build revenue forecasts tailored to their individual battery energy storage systems. This begins with the customization of specifications such as:

  • rated power (MW),
  • energy capacity (MWh),
  • location (i.e. ERCOT West Load zone),
  • or round-trip efficiency (%).

In addition to the battery’s physical characteristics, the dispatch model considers how the battery will realistically operate over its lifetime, as shown in the table above.

To do this, the forecast is also configurable to:

  • daily and yearly cycling limits (i.e. maximum of two cycles per day),
  • lifetime of a battery - according to its warranty (i.e. 8000 cycles),
  • cost of a cycle ($/MWh) - this factors in the cost of degradation over time due to battery charge and discharge,
  • state-of-charge limits (as a % of available energy capacity),
  • degradation curve assumptions (as a % of installed energy capacity).

How does this translate to day-to-day operations in the forecast?

Implementing these operational constraints also results in a more sophisticated model.

In particular, the cost of cycling and cycling limits make simulated operations more representative of the real-world. A simulated battery with perfect foresight could, theoretically, trade every small arbitrage spread throughout the day.

This would result in unrealistically high cycling rates - and revenues.

However, with cycling constraints, the dispatch model only pursues a limited number of larger arbitrage spreads - which tend to be more predictable in reality.

Additionally, the dispatch model also accounts for the Ancillary Service qualification limits imposed by NPRR 1186. Depending on the battery’s duration, this limits the amount of responsibility that a battery can be awarded in some Ancillary Services.

This applies, specifically, to ECRS and Non-Spinning Reserve:

  • ECRS: a one-hour battery can be awarded a maximum of 50% of its rated power. Batteries with a duration of two hours or more can be awarded ECRS responsibility up to their full rated power.
  • Non-spin: a one-hour battery can be awarded a maximum of 25% of its rated power to a contract, while a two-hour battery can be awarded a maximum of 50% of its rated power.
  • RRS and Regulation: Batteries with durations larger than one hour are not limited - a one-hour or longer-duration battery can be awarded responsibility up to their full rated power.

With the constraints and specifications defined, how does the battery’s decision-making process play out in the dispatch model?

What does a day of dispatch model operations look in the ERCOT forecast?

Let's step through a day of operations for a 10 MW, two-hour battery energy storage system in West Texas. In this example, the battery is dispatched against actual prices from a day of operations in the summer of 2023.

From midnight to 9 AM:

  • the battery begins with a state-of-charge of just over 10 MWh,
  • and throughout the morning, it participates exclusively in the Responsive Reserve Service (RRS) and Regulation Down.

RRS has a near-zero dispatch rate, meaning it is the most attractive when all other Ancillary Service prices are low and roughly equivalent.

Meanwhile, Regulation Down provides the added benefit of allowing the battery to earn revenues while charging and is the highest-priced service in the early morning hours.

From 9 AM to 2 PM:

  • solar generation increases from near-zero to maximum output,
  • the battery is dispatched to charge while Energy prices are low - reaching ~100% state of charge around 11 AM,
  • the battery also prioritizes ECRS responsibility - ECRS is the service with the highest clearing price in this period,
  • at 1:45 PM, the battery discharges at its full rated power when Energy prices rise above $500/MWh.

From 2 PM to 10 PM:

  • the battery allocates capacity mainly to ECRS,
  • as the prices of Ancillary Services fluctuate relative to each other, this is supplemented with RRS and Regulation Up,
  • the battery does not discharge, as Energy prices sit between $25 and $50/MWh.

Around 10 PM, Energy prices exceed $300/MWh, prompting the battery to discharge at its full rated power over multiple intervals. This depletes its state-of-charge to ~10 MWh, or 50%.

The forecast does this optimization over every 24-hour period using the Energy and Ancillary Service prices assembled by the production-cost model. This produces the end result - battery revenues and cycling rates through 2050.

How accurate is the dispatch model, and what are its limitations?

Operating a battery with the dispatch model over historical power prices from January through May 2024 results in an outperformance of around 15% - relative to the average battery in ERCOT.

This represents realistic levels of revenue capture for a battery in ERCOT over this period.

Additionally, the proportions of revenues from each market are relatively consistent with the average battery in ERCOT, though the dispatch model does tend to favor Energy arbitrage while trying to maximize overall revenues, rather than per-cycle revenues.

This resulted in revenues that would have corresponded to the 65th percentile of one-hour batteries in ERCOT over the first five months of 2024.

The outperformance - relative to the mean - is predominantly due to some of the assumptions of the dispatch model - the primary one being that the model uses perfect foresight.

This allows the battery to maximize the spread at which it charges and discharges in the Energy markets. Additionally, given the combination of clearing prices and historical activation rates, it is able to optimally allocate capacity to Ancillary Services throughout the day.

The dispatch model is also configured such that the battery acts as a 'price-taker'. This means that dispatching the simulated battery does not impact the outcomes of the production-cost model. In other words, dispatching a battery to discharge power will not impact the Energy price.

However, the assumptions the dispatch model uses to limit cycling in a given day result in realistic cycling. And typically, the battery captures only the largest spread of each day, which tend to be more predictable in their timing, day-to-day.

Head to the Modo Energy forecast documentation for additional details.

If you want to learn more about Modo Energy’s ERCOT Forecast for battery energy storage revenues, schedule a call today.