ABSTRACT
The electric grid was designed to move large amounts of energy through space, but decarbonization goals will require it to also move energy through time—from days and seasons with a surplus of energy production to days and seasons with insufficient production. Long-duration energy storage technologies that can hold a large amount of electricity and distribute it over periods of many hours to days and even seasons will play a critical role in the clean energy transition. But creating an environment in which these nascent technologies can develop and thrive will require changes in how the grid is planned and built.
Battery energy storage is booming in the United States. Driven by the need to integrate variable energy sources like wind and solar, as well as significant tax credits established by last year’s Inflation Reduction Act, utilities are aggressively pursuing energy storage technologies. At the end of 2019, there were 958 megawatts (MW) of battery energy storage on the US grid. By the end of this year, there is expected to be 18,530 MW—a nearly 20-fold increase in just four years. And more than 11,000 MW of new battery energy storage projects are already contracted for 2024.Footnote1
Most (about 88 percent) of those batteries are lithium-ion batteries. Lithium-ion batteries can generally only hold up to four hours of energy, and many projects have been built with less than that (the average duration of energy that can be provided by battery storage projects in the United States is about 2.7 hours).Footnote2 These short-duration storage technologies have helped integrate clean energy by doing things like shifting production during low-demand periods to high-demand periods and smoothing out momentary changes in output (such as a cloud passing over a solar plant or a sudden gust of wind).
But in a decarbonized energy future, the role of energy storage will expand. There will still be a need to manage short-term variability, but how will utilities capture excess generation while the sun is up and spread it across not just a few, high-demand hours, but all night long? How will they capture excess solar generation during sunny summers and excess wind generation during breezy autumns, and then spread that energy across winters when sunshine and wind are in shorter supply?
These questions point to the impending need for long-duration energy storage (LDES) technologies, those with 10 hours of duration or more. Right now, the only proven technology that operates in that space is pumped storage hydropower, which uses pumps to move water to a higher elevation and then releases that water to run back down through turbines to produce electricity. Despite the recent growth of battery technologies, pumped storage hydropower still accounts for about 95 percent of the total energy storage capacity in the United States.Footnote3 But pumped storage hydropower plants require specific geographic conditions that allow for two large bodies of water to be physically near one another but at much different elevations, so scaling up LDES to support decarbonized power systems will require the rapid development and proving of new technologies.
Many candidate LDES technologies are in the early stages of development. Some are in an early commercial stage, meaning that they have deployed a few, small projects. Most are in a pre-commercial stage, meaning that they have not yet built a large project (though they may have small demonstration projects or contracts in place to build large projects).
In making the commercialization leap, LDES technologies face three key challenges:
Planning. The tools utilities use to plan for future needs generally do not identify the benefits of LDES because they make simplifying assumptions about grid operations that obscure the value of saving excess generation in one period for use in another, particularly if that usage period is in a different day or season.
Procurement. Acquisition processes for new grid resources are only done 2–3 years in advance in most of the country, which does not provide sufficient lead time for nascent technologies to scale up and compete against more mature technology options.
Compensation. In the wholesale energy markets that serve a majority of US electric demand, compensation mechanisms for the services that LDES assets provide are not in place.
Providing a supportive environment for the rapid scale-up and deployment of LDES technologies, then, will require changes in the way the electric grid is planned and expanded. To identify what types of changes might be needed, it is helpful to look at the two US states with the most energy storage.
The swimming pool analogy
Nearly three out of every four megawatts of battery energy storage installed in the United States thus far have gone to either California or Texas. California was home to 5,476 MW of energy storage as of June 2023 (about 52 percent of the US total); Texas was home to 2,372 MW (about 22 percent).Footnote4 To put those numbers in perspective and facilitate some visual comparisons, let’s imagine California and Texas as neighboring houses with their energy storage investments represented as swimming pools.
California’s 5,476-square-foot pool is roughly 1.33 times the size of an Olympic swimming pool, while Texas’ 2,372-square-foot pool is 58 percent the size of an Olympic swimming pool. Both of these two neighbors have nice, big swimming pools.
But energy storage projects in California and Texas have an important difference: The average California project can supply 3.48 hours of energy to the grid, while the average project in Texas can provide only 1.26 hours. Returning to our swimming pool analogy, this means that the California pool is about 42 inches deep (assuming 1 hour of duration equals one foot of water), while the Texas pool is only about 15 inches deep.
Now, imagine you lost a bet and have to do a belly flop into one of those pools. Which one would you choose? Neither belly flop would be pleasant, of course, but the California pool would at least save you the added indignation of cracking your nose on an unforgiving pool bottom.
Sometimes, the electric grid does a belly flop as well. While the modern marvels of engineering, and the thousands of smart people behind the controls, make the US grid highly reliable most of the time, climate change and efforts to address it are placing new strains on the grid and the people who operate it. Stronger, more prolonged periods of high heat are increasing demand while pushing grid equipment to the limits of its operational range. More intense storms and wildfire activity can damage or destroy grid infrastructure and leave customers without service for extended periods. Increasing reliance on variable energy sources like wind and solar makes it more difficult for grid operators to match electric supply with customer demands.
Under these strains, grid failures and outages are inevitable. But when the grid does the occasional belly flop, a deeper pool (i.e. longer-duration storage resources) provides a softer landing place to lessen the pain and mitigate the impacts. Customers who would have lost service due to an outage at a power plant or on a transmission line can remain connected through a local energy storage resource. Shortfalls in generation driven by a prolonged period of inadequate sun or wind can be met with energy stored when those resources were plentiful.
But how do utilities, grid operators, and policy makers create an environment that can provide that depth? Why does the average storage project in California have a much longer duration than the average project in Texas (or any other state, for that matter)? And what does that say about what to do next?
Compare and contrast: California and Texas
Before we dive in (pun most definitely intended), there is a key disclaimer: This comparing and contrasting of California and Texas is not an analysis of the merits of either state’s approach to electric policy and regulation, nor is it meant to endorse or critique either state’s policies. The only intent is to tease out the differences in California’s approach that have resulted in higher-duration storage projects, on average, and see how they might inform the incremental reforms that will be necessary to facilitate the development and deployment of LDES resources.
One possible difference between the two states is energy mandates. California was the first state to adopt an energy storage mandate in 2013, requiring its utilities to contract for 1,320 megawatts of energy storage by 2020. But California has procured energy storage far in excess of what the mandate required, and the mandate did not contemplate duration, so the state’s utilities could have satisfied it with shorter-duration storage projects. Furthermore, Texas has also procured a large amount of energy storage absent any state mandate to do so. California’s energy storage mandate may be a factor in the total amount of energy storage in the state, but it cannot explain the higher average duration of those storage projects.
Renewable generation levels are another possible reason for the different outcomes; a state with more variable energy generation would have an increased need for storage to manage that variability. And while California did generate 38 percent of its energy from solar and wind in 2022, Texas wasn’t far behind, with 27 percent of its energy coming from solar and wind. A differential of 11 percentage points likely doesn’t explain the wide difference in storage outcomes in the two states.
A third possible factor is energy market structure. California and Texas both operate competitive electricity markets, though they differ in how those markets are structured and managed. In California, state-run planning processes identify future grid needs and then establish requirements for utilities to construct and/or contract with enough energy resources to meet those needs. For the utilities that participate in the California Independent System Operator market, daily energy needs are met through competitive auction processes. To ensure that enough resources participate in the market, the state operates a Resource Adequacy Program, which pays electricity-generating resources and energy storage projects for being available to meet grid needs. Under the Resource Adequacy Program, these grid assets receive two forms of payment: contracted Resource Adequacy payments for being available, and energy market payments for actual energy output on a day-to-day basis.
Texas, on the other hand, is an energy-only market. All energy needs are met through daily competitive markets, and resources are only paid for their actual energy output. When the market price for energy gets high, that provides a price signal for additional energy resources to enter the competitive market. The state is not directly involved in resource planning and does not compensate a resource for being available to meet grid needs.
These differences in market structures are the main reason for the two states’ different storage outcomes. To ensure that sufficient resources will be on the grid during daily periods of peak demand, California’s Resource Adequacy Program evaluates resources based on how they perform over a four-hour period. Under this approach, batteries with a duration of less than four hours are prorated and receive reduced compensation. This structure creates a clear incentive for four-hour batteries and is a key factor in explaining why California storage projects have longer durations on average.
In Texas, a developer’s incentive is purely driven by market outcomes. Each day’s highest-demand hour is also the hour with the highest energy costs; designing a low-cost, 1-hour battery system to charge when energy is cheap and then discharge during that most expensive hour yields the highest possible return on investment. Adding additional duration to provide service during other hours increases the cost of the battery system while offering diminishing returns; there is little incentive in Texas’ energy-only market to build a battery with more than an hour of storage.
California’s average duration of 3.48 hours is much higher than any other state but is still far short of the accepted definition of LDES (10+ hours). It is worth noting that the state and its electric utilities have also taken steps to encourage LDES technologies in recent years by adjusting the grid planning process to identify LDES needs, providing funding for LDES technologies, and issuing procurements specifically targeted at LDES technologies.
Three ways to support LDES technologies
Again, the point of this analysis is not to praise or criticize the energy policies of either state, but to answer two simple questions: Why does California have storage projects with longer durations? And what does that say about the types of additional energy market reforms that will be needed to support long-duration energy storage technologies?
The first question can be answered with the deliberate signals California has placed in its market design for four-hour energy storage. As for the second question, three key themes emerge: moving from an economic to reliability paradigm, evolving energy models, and lengthening the horizon for resource procurement.
Moving from an economic to reliability paradigm
Whether through a competitive market process or a regulated planning process, an energy asset must be cost-effective to be selected by a planning model or market operator. But in electric grid operations, everything revolves around hourly schedules. Markets transact on an hourly basis, and resources are scheduled and dispatched on an hourly basis. In that paradigm, the value of a resource is based on how it performs from one hour to the next. To capture the value of flexible resources, many markets have moved to even smaller dispatch windows, such as 5 or 15 minutes. The value of an LDES resource, which is realized over many hours or days, does not fit into the economic paradigm of the grid.
But, as California and other regions have demonstrated, market processes can be adapted to better reflect reliability needs. By tying a storage asset’s compensation to its duration, California has demonstrated that resource valuation can take place over a wider window. In the PJM wholesale market, which serves several Eastern US states, a similar policy requires a storage asset to have 8 hours of duration to be fully compensated in the capacity market (which serves the same objective as California’s Resource Adequacy Program and is similarly structured); anything shorter receives a prorated payment.
Moving from an economic lens to a reliability lens is already creating signals for longer-duration storage. As grid needs further evolve, market processes and compensation mechanisms will also need to continue evolving to identify the need for longer storage durations and compensate the assets that have them.
Evolving energy models
Grid planning is a complex process. To identify future resource investments, planners must consider many variables, such as temperature, customer demand, fuel costs, market prices, resource characteristics, and weather patterns. This process, which requires hours or even days for a computer program to solve, must then be repeated several times under different planning scenarios that represent different possible futures. To keep the modeling process manageable, planners only model a few days out of the year that are representative of the different seasons and conditions the grid will face.
This simplifying assumption was reasonable when the grid was dominated by fossil-fueled power plants that could adjust their output when reality diverged from the model. But it doesn’t work well for a decarbonized grid that is largely powered by wind and solar. Since production by these resources varies, the energy needed in one day may have been generated in another day (or season) that the simplified model can’t see. Identifying the need for LDES resources requires a model that can look across all hours of the year to see when there will be production surpluses and when there will be production deficits, and then use LDES to reconcile them.
California’s planning process hasn’t quite gone to this type of year-round modeling, but the state has adopted a hybrid approach moving in that direction. The California Public Utilities Commission (CPUC), which prepares a reference resource plan every two years to guide utility planning processes, operates two models. The first does the simplified, forward-looking analysis described above to determine what new resources are needed to meet future needs. The second model then takes that output and models how the resource portfolio performs over the full course of a year to see whether it meets all needs. The second model may also use sensitivity analysis to see how the portfolio performs under varying conditions. By using both models, the CPUC can incorporate a full-year view into its resource planning and identify energy storage needs.
The 2019 planning cycle illustrated how this process works. When the CPUC ran the first model, it did not select any LDES resources. But the commission engaged in an extensive, public process to identify weaknesses in its models and correct them. A detailed description of the modeling changes that the CPUC made is beyond the scope of this article, but suffice it to say that the agency compared its model assumptions to grid operations and made several adjustments to bring the model more in line with operational realities. It also improved the ability of the two models to work together. After implementing the changes and repeatedly running the two models, the process identified 973 megawatts of LDES resources that would be needed by 2026. The CPUC has slightly increased that target to 1,000 megawatts and pushed it out to 2028 in recognition that LDES resources will have a longer lead time for development, which leads into the third theme.
Lengthen the horizon for resource procurement
As the California commission recognized, the LDES market is still nascent, and projects will take longer to develop as a result. But where most utilities and wholesale markets are making procurement decisions for new grid assets on cycles of three years or less, the CPUC carved out a dedicated procurement requirement for LDES technologies and placed it on a separate timeline. This track, which also includes other developing technologies with long lead times, recognizes the need for unique asset classes and gives them time to develop. This forward-looking procurement process sends a clear, long-term investment signal to industry that LDES technologies will be needed, which provides the certainty needed for developers to attract investors and financing to develop their technologies.
The federal government has also provided funding to support the development of the LDES industry. The Inflation Reduction Act offers financial incentives to support the construction of new energy storage manufacturing facilities around the country, including some that will make long-duration systems using non-lithium components like iron, nickel, and zinc. The US Department of Energy also announced $325 million in funding in September 2023 that will support the development of 17 LDES projects that cover a wide range of technologies. Several of the projects involve the use of various types of flow batteries, which have giant tanks of liquid electrolyte that are pumped through a circuit and react with a membrane to either receive or release electricity. Other projects will test thermal technologies, which store energy by cooling and reheating a gas or solid.
Flow batteries help explain why lithium-ion batteries are not able to provide long-duration grid storage. For energy storage technologies, duration is a function of their power output, expressed in kilowatts or megawatts, divided by their energy component, expressed in kilowatt-hours or megawatt-hours. In a flow battery, the power component is determined by the size of the membrane, while its energy component is determined by the size of its tank. Because the numerator and the denominator can be independently set, a flow battery system can be configured with a wide range of durations. Lithium-ion batteries, on the other hand, consist of individual cells or pouches that have fixed power and energy ratings. Adding more cells or pouches increases both the numerator (power) and denominator (energy), so the duration remains the same.
Just as a swimming pool provides a welcome refuge on a hot day, LDES resources can provide much-needed relief for the electric grid when it is under its biggest strains. The emerging realities of climate change are already rendering the need for a clean, reliable, and resilient power system in sobering detail. And while it still may be several years before US power systems arrive at the level of variable power output at which LDES becomes critical, steps must be taken now to align grid planning and procurement processes with that impending need to ensure that developing technologies will be ready when their time comes.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Funding
This work was funded by the US Department of Energy, Office of Electricity.
Funding
This work was funded by the US Department of Energy, Office of Electricity.
Additional information
Funding
Notes on contributors
Jeremy Twitchell
Jeremy Twitchell is a senior energy analyst at the Pacific Northwest National Laboratory, where he leads PNNL’s work on the Equitable Regulatory Environment thrust area of the Department of Energy—Office of Electricity’s Energy Storage Program. His research focuses on identifying the policy and regulatory barriers that impede the deployment of energy storage technologies and best practices for reducing or eliminating those barriers, as well as providing technical assistance to states on energy-storage-related topics. He also supports other efforts at the lab in areas related to grid planning, utility regulation, and energy system equity. Prior to joining PNNL, Jeremy worked at the Washington State Utilities and Transportation Commission.
Notes
1. Based on July 2023 Form EIA-860M monthly inventory of planned electric generators compiled by the US Energy Information Administration 2023. Preliminary Monthly Electric Generator Inventory (Based on Form EIA-860M as a Supplement to Form EIA-860) for June 2023. https://www.eia.gov/electricity/data/eia860m/.
2. Id.
3. Id.
4. Id.