20% Price Declines for Wholesale Power? The Compelling Social Case for Electric Energy Storage and Why Regulated Utilities Are Likely to Roll It Out First
Eric Selmon Hugh Wynne
Office: +1-646-843-7200 Office: +1-917-999-8556
Email: firstname.lastname@example.org Email: email@example.com
SEE LAST PAGE OF THIS REPORT FOR IMPORTANT DISCLOSURES
March 22, 2017
20% Price Declines for Wholesale Power?
The Compelling Social Case for Electric Energy Storage
and Why Regulated Utilities Are Likely to Roll It Out First
The deployment of relatively small amounts of electric energy storage on the bulk power system can significantly reduce peak power prices while only slightly increasing off-peak prices. The resulting savings to consumers are compounded by the fact that the volume of electricity sold during peak hours is some 40% higher than that sold off-peak. The potential to reduce materially customer bills will provide the incentive for utility regulators to promote the rollout of storage by utilities and consumers – with potential disastrous consequences for competitive generators.
- Using historical load and price data from five regional transmission organizations (CAISO, ERCOT, ISO New England, New York ISO and PJM), we have modeled the impact that the deployment of storage can have on power prices. From 2011 through 2015, 500 MW of storage deployed on each of these RTOs would have reduced peak hour prices by an average of ~$11/MWh, or ~20% of the average price during those hours, while raising off-peak prices by only ~$3/MWh (Exhibit 2).
- On average across the five RTOs studied, 40% more electricity is sold during peak hours than off-peak. Thus the savings from cutting peak hour prices are compounded by the much larger volume of electricity sold on peak, while the cost of raising off-peak prices is mitigated by the smaller volume sold off-peak (Exhibit 3). The impact of storage on generators’ revenues (price x volume) is therefore much greater than its impact on prices alone.
- In states where power generation has been deregulated, consumers pay wholesale market prices for electricity. In these states, the ability of storage materially to reduce the cost of procuring electricity during peak hours, while only marginally increasing it off-peak, offers the potential for significant savings to electricity consumers (Exhibit 4).
- Over 2011-2015, we estimate the annual savings to consumers in ERCOT of deploying 500 MW of storage to arbitrage off-peak/on-peak power prices to be ~$1.3 billion, and in PJM, ~$0.6 billion.
- These annual savings, when divided by the 500 MW of storage deployed to achieve them, translate into ~$2,700 of consumer savings per kW per year in ERCOT and ~$1,200 per kW-year in PJM. These amounts materially exceed the annual revenue required to recover the cost of 500 MW of lithium ion storage (estimated at $389 to $813/kW-year by Lazard and Enovation Partners in Lazard’s Levelized Cost of Storage Version 2.0).
- Transmission constraints on the grid may limit the geographic impact of storage on prices. As we have not modeled these constraints, the actual amount of storage required to achieve the price and revenue impacts calculated in this note would likely need to be higher. Still, our analysis suggests that a very small amount of storage, as a percentage of total generation, could materially reduce costs to consumers.
- We believe the substantial savings to electricity consumers from the deployment of even limited amounts of storage will motivate state utility regulators to encourage investment in storage by regulated utilities and their customers.
- FERC regulations and recent legal precedents prevent state-regulated utility monopolies from deploying storage or generation for the purpose of influencing prices in FERC-regulated wholesale power markets.
- However, state regulators may require state-regulated utilities to deploy storage to achieve other objectives, such as enhancing the reliability of their distribution grids. The California Public Utilities Commission, for example, has required the state’s investor-owned utilities to deploy 1,325 MW of storage by 2020.
- Storage deployed by utilities to enhance grid reliability would do so by limiting system loads at peak hours. Storage deployed by large commercial and industrial consumers to reduce demand charges and capitalize on low off-peak rates would also curtail peak demand. These deployments would therefore erode peak hour wholesale prices.
- The impact of storage on peak hour prices and revenues will create challenges for CPN, DYN and NRG, all of which have substantial generation capacity in the markets where storage deployment is highest (Exhibits 9 & 10).
Exhibit 1: Heat Map: Preferences Among Utilities, IPP and Clean Technology
Source: SSR analysis
The deployment of relatively small amounts of electric energy storage on the bulk power system can significantly reduce peak power prices while only slightly increasing off-peak prices. Because the volume of electricity sold at peak hours is some 40% higher than that sold off-peak, the impact of this price differential on electricity revenues is even larger. The potential materially to reduce customer bills will provide the incentive for utility regulators to promote the rollout of storage – with potential disastrous consequences for competitive generators.
Our analysis assumes the installation of 500 MW of electric energy storage capacity in each of five regional transmission organizations (RTOs): CAISO, ERCOT, ISO New England, New York ISO and PJM. We have assumed this storage capacity is charged during the four hours of each day when power prices are lowest, and discharged during the four highest price hours. Since we assume the battery to be 90% efficient across the charge/discharge cycle, charging will it in fact take slightly longer than four hours: a battery discharging 500 MW of power for four hours during peak hours, delivering 200 MWh of electricity to the grid, must be charged with 222 MWh of electricity purchased off-peak over 4.4 off-peak hours.
To estimate the impact of this hypothetical storage capacity on wholesale power prices, we have reviewed historical load and price data from each of the five RTOs. For each year of the five year period from 2011 through 2015 (the last year for which data were available) we identified the four hours of each day when energy prices were highest, as well as the four hours of each day when they were lowest. We recorded the power demand (load) prevailing on the regional transmission organization’s system during each of these hours. To estimate how power prices would respond to the dispatch of 500 MW of electric energy storage during the four peak demand hours of each day, we identified those hours in each month when power demand on the RTO was 500 MW below the peak levels of the month and averaged the prices prevailing during those hours. The intuition is that the availability of 500 MW of storage would allow peak power demand to be met while dispatching generating capacity at a level 500 MW lower than would otherwise be required; the prices prevailing on the system during hours when the generation fleet operated at this level would therefore provide a good indicator of how prices might be reduced by dispatching 500 MW of storage.
We repeated this analysis in reverse to test the sensitivity of power prices to the charging of ~500 MW of storage during the 4.4 hours each day that electricity prices were lowest. For each month of the five year period, we averaged the electricity prices prevailing on the system when power demand was ~500 MW higher than that prevailing during those lowest demand hours of each day.
In competitive power markets, prices reflect the variable cost to supply electricity from the last generating unit dispatched to meet demand. The marginal cost of supplying electricity, in turn, is a function of numerous factors that vary with time, including the delivered prices of gas and coal, the availability of renewable energy, the level of demand on the system and thus the amount of generation capacity that must be dispatched to meet it, transmission constraints that vary with load, and the scheduled and unscheduled maintenance of major power plants. To ensure that our estimates of the impact of storage were sensitive to seasonal changes in these variables, we repeated our analysis for each month of the five year period, measuring the levels of demand during the four hours per day when prices were highest (lowest) and averaging prices during that month when demand was 500 MW lower (higher).
Limitations of Our Analysis
The simplified analysis of the impact of storage described above, while facilitating the manipulation of very large amounts of price and volume data, fails to take into account certain physical limitations of the power grid and thus may distort our results. First and foremost, we have not modeled how transmission constraints can limit the geographic impact of storage on prices. A very large RTO such as a PJM can have bottlenecks on the transmission system that prevent power from being imported to meet demand in transmission constrained areas (“load pockets”). Peak demand in these areas must therefore be met by resources within the load pocket itself, causing peak hour prices to differ, sometimes markedly, inside and outside the transmission constrained zone. Critically, storage deployed outside the load pocket would have no impact on peak hour prices within the load pocket. Conversely, if the supply curve within the load pocket were particularly steep at peak levels of demand, storage deployed within the load pocket might have a greater impact on peak prices than that deployed outside.
Even in the absence of a transmission constraints severe enough to create load pockets, the loading of the transmission system as demand rises can cause congestion as portions of the system reach their maximum capacity, causing the price differentials between different nodes of the system to vary markedly from off-peak to peak demand hours. Careful attention to the location of storage resources could reduce congestion on the grid during peak hours, allowing for lower equilibrium prices at any given level of demand.
The estimates presented in this note, in other words, do not reflect the physical constraints on the delivery of electricity across the grid as these change over the course of the day, in some cases isolating particular regions. As we have not modeled these constraints, the actual amount of storage required to achieve the revenue impacts calculated in this note would likely need to be higher than the estimates presented here. Our analysis does suggest, however, that a very small amount of storage, as a percentage of total generation, could materially reduce costs to consumers.
A second important limitation of our analysis is that we have assumed a minimum deployment of 500 MW of storage in each RTO. For the smaller RTOs, such as ISO-NE and NYISO Zone J, the economically optimal amount of storage capacity may in fact be smaller than this, in the sense that the bulk of the price reductions and customer savings achieved by deploying 500 MW of storage may be achievable with a smaller amount of storage capacity. In these markets, therefore, our analysis may underestimate the benefits of storage, measured in annual customer savings per kW of storage deployed.
Finally, we should note that the storage capacity installed to date on the bulk power system is often deployed in more than use, the most common being frequency regulation, spinning reserves and the arbitrage on off-peak and on-peak power prices. The use relevant to our analysis of the impact of storage on prices is the last, energy price arbitrage. Storage capacity deployed in uses that are not consistent with energy price arbitrage (as may be the case for example, with storage deployed for frequency regulation) would not bring about the reduction in prices and thus the consumer savings implied by our study.
Results of Our Analysis
As explained above, our model uses (i) the actual power prices prevailing when demand is 500 MW higher than it is during the lowest demand hours of the day to estimate equilibrium prices during the hours that our hypothetical storage facility is charged and (ii) the actual power prices prevailing when demand is 500 MW lower than it is during the highest demand hours of the day to estimate equilibrium prices during the hours when our hypothetical storage facility is discharged. We repeat this analysis for each month of the five years from 2011 through 2015 to take into account changes in fuel prices, the availability of renewable energy, and other factors that vary through time and have a material impact of the marginal cost of supply in wholesale markets.
Based on this analysis, one of our key finding is that the impact of storage is likely to be much greater on peak power prices than off-peak power prices, reflecting the relative steepness of the supply curve during the hours of minimum and maximum power demand. Across the five RTOs we analyzed, our analysis suggests that the average impact over the five year period from deploying 500 MW of storage would have been to increase prices during the five lowest demand hours of the year by an average of $2.88/MWh while cutting prices during the four highest demand hours by $10.89/MWh. The actual results by region are presented in Exhibit 2.
The favorable impact of storage on average prices is magnified by the fact that, on average across the five RTOs, system demand for electricity (load) is ~40% higher during the highest demand hours of the year than it is during the lowest demand hours. We illustrate the marked difference in the volume of peak and off-peak demand in Exhibit 3. The substantial price reduction achieved during peak hours through the deployment of storage thus affects a much larger volume of electricity sales than does the modest increase in prices during off-peak hours. The implication is that the deployment of storage reduces generator revenues (price x volume), and thus customer bills, by a much larger amount than its price impact alone.
Exhibit 2: Impact on Peak and Off-Peak Prices Exhibit 3: Load During the Four Lowest and Four
($/MWh) of 500 MW of Storage, 2011-2015 (1) Highest Demand Hours, 2011-2015 (MWh) (1)
1. Load and price data are calculated for the four lowest and hour highest demand hours of each day of the year for the five year period 2011-2015. Prices data shown reflect the average of all days, and load data the sum of all days, over this five year period.
Source: SNL, SSR analysis
The scale of this reduction in revenue is presented in Exhibit 4, where the red columns illustrate the reduction in generator revenues from the discharge of 500 MW of storage during the four highest demand hours of the year and the blue columns show the increase in generator revenues caused by the charging of the battery during the four lowest demand hours of the year. As can be seen there, our analysis suggests that generator revenues would fall in all five RTOs. The biggest impact by far, however, would be felt in ERCOT and PJM: over the five years 2011-2015, we calculate that the savings to consumers in ERCOT of deploying only 500 MW of energy storage would have been ~$1.3 billion annually, while consumers in PJM would have saved $0.6 billion annually.
Exhibit 4: Change in Cost of Procuring Peak and Off-Peak Power to Serve System Load Following Deployment of 500 MW of Storage, per RTO ($ millions)
Source: SNL, SSR analysis
In Exhibit 5 we divide (i) the estimated reduction in generator revenues in each of the five RTOs by (ii) the 500 MW of storage capacity deployed, to arrive an estimate of annual consumer savings per MW of storage deployed. These savings can be compared to the annual revenue required to recover the capital cost of deploying a 500 MW lithium ion battery on the power grid. Lazard and Enovation Partners have estimated the annual revenue required to recover the capital cost of a lithium ion battery at $389 to $813/kW-year (see Lazard’s Levelized Cost of Storage Version 2.0, published in December 2016 and available at https://www.lazard.com/media/438042/lazard-levelized-cost-of-storage-v20.pdf). This revenue requirement exceeds the annual customer savings from the deployment of 500 MW of storage in three of the five RTOs studied, CAISO, ISO New England and New York ISO Zone J. In ERCOT and PJM, by contrast, the scale of consumer savings is a multiple of the cost of the storage deployed. In ERCOT, we estimate, consumers would save ~$2,660 annually per kW of storage deployed, more than 3.3x its cost, while in PJM we estimate consumer savings at ~$1212 annually per kW of storage deployed, or 1.5x its cost. One conclusion of our analysis, therefore, is that
under the right circumstances the deployment of modest amounts of storage on the grid can materially reduce the cost of procuring electricity in the wholesale power market, so much so as to more than offset the cost of energy storage.
Exhibit 5: Estimated Consumer Savings from Deploying 500 MW of Storage Compared to Estimated Annual Revenue Requirement of Lithium Ion Storage ($/kW-year)
Source: SNL, Lazard and Enovation Partners, Lazard’s Levelized Cost of Storage, Version 2.0, SSR analysis
We believe the substantial savings available to electricity consumers from the deployment of limited amounts of storage will motivate state utility regulators to encourage the deployment of storage by regulated utilities and electricity consumers. Importantly, the authority of state regulators to do so is limited by FERC regulations and recent legal precedents that prevent state regulators and state-regulated utility monopolies from deploying storage or generation for the purpose of influencing prices in FERC-regulated wholesale power markets, such as the five RTOs we have analyzed. However, FERC’s regulatory authority extends only the sale or transmission of electricity on the high voltage power grid, while state regulators retain authority over the low voltage distribution system. On the distribution system, therefore, state regulators would be within their rights to encourage the deployment of storage to limit peak demand, much as they have authorized demand management and demand response programs in the past. The California Public Utilities Commission, for example, has required the state’s investor-owned utilities to deploy 1,325 MW of storage by 2020.
In addition, state regulators could require state-regulated utilities to deploy storage for other purposes, such as enhancing the reliability of their distribution grids or to firm renewable energy purchased to comply with state renewable portfolio mandates. As explained in our note of February 15, Electric Energy Storage and the Bulk Power System, almost two thirds of the battery and flywheel storage capacity on the U.S. grid is deployed in two or more uses, and half is deployed in three or more (Exhibit 6). The most common purposes of the battery and flywheel capacity deployed on the U.S. power grid to date is to provide frequency regulation (71% of installed capacity) and spinning reserves (31%), yet 30% of this capacity is also stated by its developers to be used for energy arbitrage (Exhibit 7). Similarly, we would expect storage capacity deployed by utilities for purposes of enhancing reliability and power quality to be deployed also in arbitraging between peak and off-peak power prices.
Exhibit 6: Percentage of U.S. Battery & Exhibit 7: Stated Applications of U.S. Battery &
Flywheel Capacity Allocated to Primary, Flywheel Capacity Deployed on the Grid to Date
Secondary and Tertiary Use Cases
Source: U.S. Department of Energy, Energy Storage Database (www.energystorageexchange.org) , SSR analysis
To provide an idea of the range of uses for which storage can be deployed on the power grid, Exhibit 8 breaks down the battery and flywheel capacity deployed on the U.S. power grid to date, showing the stated purpose, average scale and number of these projects. Of the 800 MW of battery and flywheel capacity on the power grid today, 700 MW has been deployed on the high voltage transmission grid. These projects average 5.6 MW in size and their stated purposed is most commonly frequency regulation (39% of the storage capacity deployed), electric energy time shift or energy price arbitrage (27% of capacity) and providing some form of reserve capacity (18% of capacity, if we add the MW whose stated purpose is electric supply capacity, spinning reserves and ramping). The remaining ~100 MW have been deployed on the low voltage distribution system for a much wider range of uses. The stated purpose of these projects, which average only 0.5 MW in size, is most frequently electric bill management (i.e., large commercial and industrial customers deploying batteries to reduce their maximum draw from the grid and thus the demand charge they must pay, or to purchase cheaper electricity at night for use during the day) or to enhance the reliability and quality of the power supplied by the grid.
The primary uses of the capacity deployed on the distribution grid to date are not inconsistent with the utilization of this capacity to reduce peak demand; indeed, in the case of the storage capacity deployed for bill management by commercial and industrial consumers, this is its primary purpose. In California, CAISO has encouraged the aggregation of this distributed storage capacity by firms such as Stem, Inc. so that it may be offered into the wholesale power market as a source of short term capacity during peak hours. To the extent that regulators, utilities and consumers see a benefit from the deployment of storage on the low voltage distribution system, we would expect the implications of these decisions to be felt in the wholesale power markets as well.
In conclusion, whether deployed by regulated utilities at the behest of state regulators, in order to upgrade the reliability and quality of power on the distribution grid, or deployed by large commercial and industrial customers for this purpose or to manage their electricity bills, once deployed, storage capacity on the distribution grid will be available and is likely to be used to supply peak demand with cheap electricity purchased when demand is low. Our analysis shows that even relatively small amounts of storage (500 MW) used for this purpose can materially reduce generator revenues (see Exhibit 4) and thus customer bills.
Exhibit 8: Electro-chemical and Small Scale Electro-mechanical Storage Projects Deployed on the U.S. Power Grid to Date, Showing Size in MW and the Primary Use Stated by Their Developers
Source: U.S. Department of Energy, Energy Storage Database (www.energystorageexchange.org), SSR analysis
Impact on Competitive Generators
We expect investment in electric energy storage to be strongest in those regions of the country where generation has been deregulated and utilities and retail electricity providers procure power for their customers in competitive wholesale markets. It is in markets such as these that the social benefits of storage – the suppression of peak power prices and generator revenues and the consequent reduction in customer bills – will be greatest, and where regulators, regulated utilities and electricity consumers will face the strongest incentives to pursue it. Critically, it is these same markets that the bulk of the capacity owned by competitive power generators is situated.
To date, the largest deployments of storage capacity have occurred in the California ISO, which saw 264 MW installed over the period 2011-2016, and PJM, with 260 MW (see Exhibit 9), followed by ERCOT and MISO, with 48 and 24 MW, respectively. By far the largest planned deployment of energy storage is also in CAISO, with 483 MW of capacity additions announced and a total of 1,325 MW of new storage capacity required by 2020 by the California Public Utility Commission for California’s investor owned utilities. Other regions where we would expect to see regulated storage investment by T&D utilities are those where state regulators are most consumer friendly or have developed plans to upgrade the grid through the integration of renewable energy, distributed generation and energy storage capability. These regions include ISO-NE (New England), the NYISO (New York) and PJM (Mid-Atlantic and Midwest). We also expect storage deployment in ERCOT (Texas), MISO (Midwest) and SPP (Midwest and Plains states) for the purpose of integrating intermittent wind power onto the grid.
As noted, almost all of the generation capacity owned by competitive power generators such as Calpine, Dynegy and NRG Energy is concentrated in the unregulated regions where storage is likely to be deployed. Indeed, each of these three companies has between 43% and 50% of its generation capacity in the two RTOs that have seen the largest rollout of electricity storage to date (CAISO and PJM) (compare Exhibit 9 and Exhibit 10). Given the very large reduction in generator revenues we estimate would result from the rollout of only 500 MW of storage (see Exhibit 4), we fear that that future revenues, gross margins and net earnings of these companies could be materially eroded.
Exhibit 9: Historical and Announced Exhibit 10: Breakdown of IPP Generation Fleets
Additions of Storage Capacity (MW) by ISO (excludes wind and solar assets)
Source: U.S. Department of Energy, Energy Storage Database (www.energystorageexchange.org)
Impact on Suppliers of Storage Technology
The potentially large adverse implications for competitive generators from the rollout of electric energy storage on the grid do not imply a correspondingly large market opportunity for the suppliers of storage technology. In some of the smaller power markets we analyzed (CAISO NP 15, CAISO SP 15, and New York ISO Zone J), the difference between peak and off-peak demand is only 2,000 to 4,000 MW during most months of the year (see Exhibits 11 and 12). Our analysis suggests that, in these markets, the deployment of relatively small amounts of storage (2,000 MW or less) would be sufficient to eliminate the difference between peak and off-peak prices (see Exhibit 13).
Exhibit 11: Average Daily Load During the Four Highest and Four Lowest Demand Hours
Source: SNL and SSR analysis
Exhibit 12: Average Daily Load During the Four Highest and Four Lowest Demand Hours
Source: SNL and SSR analysis
Exhibit 13: Price Differential Between Highest and Lowest Load Hours at Different Levels of Storage, per ISO ($/MWh)
Source: SNL and SSR analysis
Energy storage on the power grid may thus be materially more important for wholesale power markets than these are for the suppliers of energy storage capacity. We would expect future demand for energy storage, as in the past, to be driven primarily by the use of batteries in electric vehicles, with grid storage remaining a relatively small market by comparison (see Exhibit 11).
Exhibit 11: Sales of Large Format Lithium Ion Battery Packs, 2010-2016 (MWh of capacity)
Source: EV Sales Blogspot (www.ev-sales.blogspot.co.il), U.S. Department of Energy, Energy Storage Database (www.energystorageexchange.org), SSR analysis
©2017, SSR LLC, 225 High Ridge Road, Stamford, CT 06905. All rights reserved. The information contained in this report has been obtained from sources believed to be reliable, and its accuracy and completeness is not guaranteed. No representation or warranty, express or implied, is made as to the fairness, accuracy, completeness or correctness of the information and opinions contained herein. The views and other information provided are subject to change without notice. This report is issued without regard to the specific investment objectives, financial situation or particular needs of any specific recipient and is not construed as a solicitation or an offer to buy or sell any securities or related financial instruments. Past performance is not necessarily a guide to future results.
- In the three smaller RTOs, the deployment of storage in amounts smaller than 500 MW may offer superior economics; the first MW of storage has the greatest impact on prices and thus consumer costs as it displaces the most expensive generator on the system. Each subsequent MW offers diminishing marginal returns. ↑