In the electricity industry, there is growing excitement about the capabilities and potential of energy storage technologies, such as batteries, flywheels, compressed air storage, or pumped hydro.  These technologies are often mentioned as ways to store renewable energy for times when wind or solar are not generating or as solutions to peak load management.  However, power generation from natural gas is a good substitute for many energy storage applications, including frequency regulation, energy arbitrage, and renewables integration.  And the currently low price of natural gas is greatly constraining the value of grid energy storage for several important applications.  Right now, natural gas generation is an established technology, representing most of the new generation being built in the US, while energy storage is an emerging technology still exploring business models and experiencing relatively rapid price declines.  Right now, the economics of grid storage is captive to the cost of natural gas, but may eventually pose a challenge to natural gas generation in many grid services.

Historically, electricity systems have had limited energy storage capacity.  Energy storage today makes up less than 3% of total installed capacity in the U.S., almost all of which is in the form of pumped hydro storage [1].  Although energy storage can provide many services beneficial to the grid, high capital costs and technical issues have historically limited deployment.  However, trends over the last twenty years have increased interest in energy storage: the difference between peak and off-peak load is increasing in most regions of the US, variable and intermittent renewables are being added at a rapid pace, and new energy storage technologies are being created and improved.

The price of wholesale natural gas has a large effect on the price at which natural gas generation and services are offered on the market because the levelized cost of electricity for natural gas generators is mostly due to fuel cost.  For natural gas generation, 50-70% of the levelized cost of electricity is due to the fuel [2].  This is not the case for other technologies such as coal power, for which fuel costs are less than 30% of the total cost of electricity.  Thus, decreasing natural gas prices significantly decrease the cost of services from natural gas generation, which is the main competitor for grid energy storage in many applications.

Thus, falling natural gas prices in recent years may have significantly reduced the potential profit from U.S. energy storage projects, offsetting the technological improvements in storage technologies made over the same period.  If this hypothesis is correct, a thorough examination of the relationship between natural gas price and the economics of energy storage is necessary to forecast the success of efforts in the deployment of grid storage. As an emerging technology, energy storage is a “price taker”, dependent on favorable electricity market prices and niche applications until it becomes more mature.  Any significant shifts in the prevailing electricity prices, such as those occurring since 2009, will have an effect on price-taking technologies such as energy storage, putting economic competitiveness beyond any feasible technological improvement for some technologies.

In a recent paper published in the journal Energy Policy [3], we investigated the effect that natural gas price has on the profitability of storage projects.  We evaluated the relationship between natural gas price and the profitability of energy storage in two applications for energy storage: frequency regulation and energy arbitrage.  Frequency regulation refers to the constant need to inject or remove small amounts of power from the grid to maintain a stable grid frequency.  Both energy storage and natural gas generation are well suited to this application, as they can respond quickly to regulation requests.  Energy arbitrage refers to the ability of storage to profit on wholesale energy markets by storing energy at low-priced hours and discharging at high-priced hours.  Inexpensive natural gas competes with energy storage in this service by reducing peak prices and, therefore, the profitability of arbitrage with energy storage.

A brief description of the methods is provided here, but a full description can be found in the referenced journal article.  In each of the applications, we created an engineering-economic model of a storage device providing that service, and used electricity market price data to determine the net revenue produced by the storage.  Once we calculated the storage revenue, we varied the capital costs of storage to determine the “breakeven capital cost” – the capital cost at which the storage owner makes zero net revenue when accounting for amortized capital costs, operating costs, and revenues.  In all analysis, we assumed that the storage system is small enough that it displaces only the marginal generator and has no effect on market prices.  This means that our results assume the best-case scenario for storage, as the effect of non-marginal storage participants will lower potential storage revenue in either service.

Frequency regulation is an ancillary service where a grid-level plant (generator or storage) agrees to change their power output on a second-by-second basis, following a signal sent from the system operator used to balance short-term differences in supply and demand. Participating plants bid into a frequency regulation market and are paid both a participation rate, in dollars per hour for each MW of service, and paid (or charged) for their net energy production (or consumption) during the period.  Frequency regulation service has a limited market size but is an application where storage may be able to produce significant revenue.  Some energy storage technologies are well-suited to providing frequency regulation.  For example, Beacon Power designed their primary product to provide this service [4].

The revenue that can be produced by a flywheel providing frequency regulation service is highly dependent on the frequency regulation price.  Flywheel devices for frequency regulation generally do not provide any other services, nor do they consume any fuels that may have their own price volatility.  In the two markets that we examined (PJM and NYISO), average frequency regulation price is highly correlated with the delivered price of natural gas to generators (Figure 1).  When we compare these prices to the results of our engineering-economic model of a flywheel plant, we find a very strong relationship between the prevailing natural gas price and the breakeven capital cost of a flywheel plant, due to much lower revenue to the plant during periods of low gas price (Figure 2).  If our estimated capital cost for a flywheel plant is correct, then such a plant is profitable only when natural gas prices are above the $5-8/mcf range.


Figure 1: Monthly average delivered natural gas and frequency regulation prices in New York ISO from 2008 through 2013. 


Figure 2: Average monthly delivered natural gas price versus breakeven capital cost of a flywheel.  The estimated Beacon Power flywheel capital cost range and associated range of natural gas prices are overlaid onto the figure.  Our results indicate that a flywheel device can break even only under frequency regulation prices associated with $5-8/mcf natural gas.  Some points on this figure have a negative breakeven capital cost.  This occurs when the revenue from frequency regulation service is insufficient to cover even the operating costs of the flywheel plant.

Energy arbitrage is a service that can be provided by grid energy storage where the storage is used to maximize revenue from time-shifting electrical energy from low price periods to high price periods.  This is functionally similar to (but not exactly equivalent to) peak shaving, where a storage device attempts to reduce peak load by charging when demand is low and discharging during peak demand periods of the day.  The vast majority of existing grid energy storage is in the form of pumped hydro storage, which generally operates to provide the energy arbitrage/peak shaving service that we model [5]. The potential market for energy arbitrage is very large – a cost-effective solution could be scaled up until it has significantly shifted wholesale electricity price.  However, energy arbitrage is also an application that requires inexpensive storage, due to the low potential profit margins.

Our energy arbitrage engineering-economic model does not model a particular storage technology.  Rather, we model a generic storage device with attributes of existing or likely bulk storage technologies: pumped hydro, compressed air energy storage, and some battery technologies.  Specifically, we model a 20 MW / 80 MWh storage device.  The storage device has a round-trip efficiency of 75%, with the inefficiency divided equally between the charge and discharge portions of the cycle.  We model the operation of the storage plant under both perfect and imperfect information about future electricity prices. In either case, the storage owner pursues a strategy of maximizing revenue through energy arbitrage.

In energy arbitrage, a storage device attempts to maximize net revenue by purchasing charging energy when electricity prices are low and selling the stored energy when electricity prices are high.  Thus, energy arbitrage revenue is related to short-term variation in energy prices rather than average energy price.  Figure 3 compares annual averages of natural gas prices and the daily variability in wholesale electricity prices (daily high price minus daily low price) in Washington, DC (the PEPCO DC price node in PJM).  Natural gas prices strongly affect variability in electricity prices because they dictate the marginal costs of generation from peaker natural gas turbines.  Peaker turbines have low capital cost and moderate efficiency and are designed and operated to produce relatively expensive electricity for a few hours each day.  A significant fraction of their levelized costs is due to fuel, so a decrease in natural gas price will decrease the price of electricity during peak periods each day.  The marginal off-peak generator is likely to be a coal, combined cycle natural gas, or hydro plant (depending on location) and is less sensitive to fuel prices.

Figure 4 shows the monthly natural gas price and estimated storage revenue for a location outside of Washington, DC.  Because energy arbitrage revenue depends on large daily variations in electricity price, low natural gas prices greatly reduce the potential revenue to storage.  As a result, the breakeven capital costs of storage have decreased significantly from earlier estimates (Figure 5).  While a $250/kWh arbitrage storage device would be profitable in 2004-2007, the same plant would have to have a capital cost around $100/kWh to be viable, given current electricity prices.  Since 2008, the energy storage industry has faced a difficult trend: as the new storage technologies have become ready for the market and the more mature technologies have lowered their costs, the decreasing cost of natural gas has been reducing the potential revenue of energy storage.


Figure 3: Henry hub natural gas prices and daily variability in electricity prices (Washington DC, annual average) from 2002 through 2013.  When natural gas prices are high, daily price variability increases because of more expensive electricity from peaker natural gas turbines, which set the market price during peak hours.  When natural gas prices decrease, peak price falls closer to the low off-peak prices set by coal/hydro/combined cycle plants.


Figure 4: Monthly natural gas price and hourly energy arbitrage revenue from a storage device outside of Washington, DC under perfect and imperfect information.  Energy arbitrage revenue is greater during periods of high natural gas price.


Figure 5: Breakeven capital cost versus year for a storage device providing energy arbitrage service at three locations.  These results assume perfect information about future electricity prices and exclude all operating costs except for purchased electricity.  Data availability for New York City and Chicago limit the range of years analyzed for those locations.  In the 2005-2008 period, prevailing electricity prices would allow for storage with a cost in the $200-$300/kWh range to be profitable.  Since 2009, electricity prices have had lower daily variability, and storage capital cost must be less than $100/kWh to be reliably profitable.

Grid energy storage is often offered as a necessary solution to issues introduced by fluctuating wind and solar generation, implying that the existing grid assets are insufficient for accommodating renewable generation.  If this were true, then existing price signals should indicate the need for storage.  One would observe real-time prices collapsing when renewable generation picked up and price spikes when the power output dropped off unexpectedly.  Alternately, if significant short-term variability was introduced by renewables, the frequency regulation price would be expected to increase.  In actual fact, however, the opposite has been observed: since 2007, wind and solar capacity has increased almost ten times [6], but real-time price variability and frequency regulation prices have both decreased.  This is true even in West Texas, where significant wind development continues in an area of very little load.  Despite large additions of wind power causing frequent negative prices during windy periods and prompting a significant expansion in transmission capacity out of the area, real-time price variability in West Texas has been decreasing since 2007.  Significant additions of natural gas capacity, combined with decreasing natural gas prices, can explain this effect.

Average energy arbitrage revenue and average frequency regulation revenue or storage devices in the year 2012 are both approximately a third of what they were in the 2004-2008 time period.  This drastic shift in potential revenue is both a major obstacle to energy storage investment and a signal that energy storage is not currently needed for bulk energy arbitrage or frequency regulation.  However, while the two services we investigated were commonly discussed applications in the early 2000s, the focus of the storage industry has appropriately shifted somewhat towards distributed storage and other similar applications.  Even though there have been significant improvements in storage performance and price over the last ten years, this technological progress is slowly starting to result in appreciable deployments of energy storage.

                There are several important implications of the trends presented above.  First, industry and policy makers must be aware that the fundamental economics of energy storage have shifted significantly in the last ten years and that the conclusions of analysis using pre-2008 data may no longer apply.  Second, while regulatory reforms and policy support have increasingly acknowledged the unique contribution of energy storage, future efforts must shift their focus towards better support of currently valuable applications for the technology, such as distributed storage.  Third, decision makers should understand the relationship between natural gas prices and many of the grid-level services provided by energy storage.  This means that energy storage planning should be created with an appreciation that the storage industry is currently experiencing temporarily adverse market conditions which have hindered deployments in the last six years.  It also means that continued decreases in the cost of storage technologies may eventually threaten natural gas generation in many of its current grid services.

by Dr. Eric Hittinger, Assistant Professor of Public Policy, Rochester Institute of Technology



Eric Hittinger is an Assistant Professor of Public Policy at Rochester Institute of Technology and holds a PhD in Engineering & Public Policy from Carnegie Mellon University. His research combines engineering models and economic analysis to understand the capabilities and effects of likely changes to electricity systems in coming decades, with particular focus on energy storage, renewables, and alternative market designs.

[1] EPRI, “Electricity Energy Storage Technology Options,” 2010. [Online]. Available: id=000000000001020676. [Accessed September 2011].
[2] US Energy Information Administration, “Levelized Cost and Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2014,” April 2014. [Online]. Available: [Accessed October 2014].
[3] E. Hittinger and R. Lueken, “Is inexpensive natural gas hindering the grid energy storage industry?,” Energy Policy, vol. 87, pp. 140-152, 2015.
[4] Beacon Power, “Smart Energy 25 Flywheel,” 2011. [Online]. Available: [Accessed June 2011].
[5] E. Ela, B. Kirby, A. Botterud, C. Milostan, I. Krad and V. Koritarov, “National Renewable Energy Laboratory,” May 2013. [Online]. Available: [Accessed January 2014].
[6] US Energy Information Administration, “Electricity Data Browser,” October 2014. [Online]. Available: [Accessed October 2014].