- Email: [email protected]
Long-Duration Electricity Storage Applications, Economics, and Technologies
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Perspective
Long-Duration Electricity Storage Applications, Economics, and Technologies Paul Albertus,1,4,* Joseph S. Manser,2,4 and Scott Litzelman3
The United States electricity grid is undergoing rapid changes in response to the sustained low price of natural gas, the falling cost of electricity from variable renewable resources (which are increasingly being paired with Li-ion storage with durations up to 4 h at rated power), and state and local decarbonization policies. Although the majority of recent electricity storage system installations have a duration at rated power of up to 4 h, several trends and potential applications are identified that require electricity storage with longer durations of 10 to 100 h. Such a duration range lies between daily needs that can be satisfied with technologies with the cost structure of lithium-ion batteries and seasonal storage utilizing chemical storage in underground reservoirs. The economics of long-duration storage applications are considered, including contributions for both energy time shift and capacity payments and are shown to differ from the cost structure of applications well served by lithium-ion batteries. In particular, the capital cost for the energy subsystem must be substantially reduced to 3 $/kWh (for a duration of 100 h), 7 $/kWh (for a duration of 50 h), or 40 $/kWh (for a duration of 10 h) on a fully installed basis. Recent developments in major technology classes that may approach the targets of the long-duration electricity storage (LDES) cost framework, including electrochemical, thermal, and mechanical, are briefly reviewed. This perspective, which illustrates the importance of low-cost and high-energy-density storage media, motivates new concepts and approaches for how LDES systems could be economical and provide value to the electricity grid. Introduction and Applications for Long-Duration Energy Storage The United States (US) electricity grid is undergoing rapid changes that create opportunities for new electricity storage applications and may benefit from new electricity storage technologies. First, the levelized cost of electricity (LCOE) from wind and solar photovoltaics is now lower than the new natural-gas-combined cycle power plants, even as sustained low natural gas prices are shifting the fuel mixture away from coal and, in some cases, nuclear.1 However, unlike the preceding century in which the electricity system was based on a stable supply of chemical fuel, the electricity output from wind and solar installations is variable over time scales ranging from seconds to years. The second major change is a move toward decarbonization. At the time of publication, six US states had policies of 100% clean energy or renewable energy targets signed into law with more under consideration within various state legislatures; many countries, regions, and cities are also pursuing such policies.2 Third, states, municipalities, and industries are considering technology options to increase the resiliency of their grids, in part due to rising weather variations and storms, such as Hurricane Maria, which devastated Puerto Rico and the US Virgin Islands.3 In each of these cases, cost-effective long-duration electricity
Context & Scale The feasibility of incorporating a large share of power from variable energy resources such as wind and solar generators depends on the development of cost-effective and application-tailored technologies such as energy storage. Energy storage technologies with longer durations of 10 to 100 h could enable a grid with more renewable power, if the appropriate cost structure and performance—capital costs for power and energy, round-trip efficiency, self-discharge, etc.— can be realized. Although current technologies such as lithium-ion batteries are suitable for a number of applications on the grid, they are not suitable for longerduration storage applications. Although 10 to 100 h energy storage will help facilitate the integration of renewable power on the grid, it is not long enough to last for seasons, and is not sufficient to enable a grid with 100% renewable power. Given the capital-intensive nature of scaling up a storage technology that can be impactful to the overall electricity grid, rigorous technical and economic evaluations at the laboratory and pilot scale are required. Several major classes of
Joule 4, 1–12, January 15, 2020 ª 2019 Elsevier Inc.
1
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
storage (LDES), which we consider as durations at a rated power of 10–100 h, could provide clear benefits.4 Electricity storage services on the grid today are dominated by pumped-storage hydropower (PSH) (in terms of cumulative installations) and lithium-ion (Li-ion) batteries (in terms of share of present annual installations). Molten salt storage at concentrating solar power plants and to a lesser degree on-site thermal storage, are also significant, but here, we focus only on systems that charge with and discharge electricity. PSH, which typically has duration at rated power of 6–10 h, provides peaking capacity and fulfills daily energy time shift applications. However, permitting and large-project financing difficulties have limited additions of PSH.5 Li-ion batteries are technically capable of being used in projects of a wide range of sizes (e.g., kW to GW and kWh to GWh) and for essentially any duration, but economically viable applications pursued today include frequency regulation (with charge and discharge durations on the order of seconds), reserve markets (tens of minutes), transmission and distribution upgrade deferrals (8 h), and others of %10 h of duration.6 Table 19 of reference 7 provides an overview of various electricity storage applications.7
storage technologies may address the long-duration electricity storage cost and performance framework, and efforts are accelerating to identify and develop the most promising storage systems.
To make use of low-cost wind and solar and advance decarbonization goals, LDES systems may be configured in at least two ways: (1) as grid-tied, standalone systems, and (2) integrated with solar and/or wind behind a point of common interconnection. Considering first the case of grid-tied systems, variable resources have traditionally been buffered with natural gas-fired peaker plants, though other approaches include transmission expansion, demand-side management (including on-site thermal storage, both hot and cold, of various durations), and electricity storage. For electricity storage, modeling studies have demonstrated that up to approximately 8 h of duration can increase the amount of annual energy from wind and solar that can be utilized on a large regional grid (e.g., CAISO or ERCOT).8–10 A number of studies have also looked at storage durations longer than approximately 10 h; these have also found that the addition of increasing durations of storage reduces curtailment and increases the use of variable assets like wind and solar, with a falling marginal impact.9–12 Further modeling work is needed to accurately quantify the impact of LDES on wind and solar penetration at the regional level and should include realistic handling of transmission power flow constraints, network stability, contingency requirements, opportunity costs of curtailed energy, limits to load flexibility, and other parameters necessary to capture the full complexity of delivering power within a large electricity system. It is also important to point out that while longer-duration assets (e.g., seasonal storage) can potentially serve shorter duration applications such as daily cycling (assuming the efficiency, conversion kinetics, and other attributes are suitable), the converse is not true, which may favor the development of longer-duration assets.13 In the context of these studies, Figure 1 provides a high-level and semi-quantitative relationship between the maximum storage duration required to meet demand and the fraction of annual energy from wind and solar. The colored region in Figure 1 indicates typical assumptions associated with renewable curtailment, transmission build-out, and grid flexibility (i.e., number of must-run generators, demand response), while the arrows indicate either more restrictive (to the left) or aggressive (to the right) assumptions. This figure shows that with capacity factors of wind and/or solar installations of 20% to 50%, little storage may be needed until higher fractions are reached, especially in regions with optimized blends of wind and solar.8,9 Depending on the characteristics (e.g., quality of solar and/or wind resource, transmission system capabilities, amount of natural gas versus low ramp rate resources) of
2
Joule 4, 1–12, January 15, 2020
1University 2Ionic
of Maryland, College Park, MD, USA
Materials, Inc., Woburn, MA, USA
3Advanced
Research Projects Agency – Energy, Washington, DC, USA
4These
authors contributed equally
*Correspondence: [email protected] https://doi.org/10.1016/j.joule.2019.11.009
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 1. Semi-quantitative Overview of the Maximum Duration of Electricity Storage Needed to Ensure Demand Is Met at All Times versus the Fraction of Annual Energy from Variable Generators, Such as Wind and Solar Arrows indicate the impacts of wind and/or solar curtailment, transmission capacity, geographic extent, and grid flexibility (e.g., number of must-run generators and demand response) on the maximum required duration of electricity storage. The values in this figure are based on previous studies. 8–15
the region, between about 50% and 80% of annual energy from solar and/or wind can be reached with storage durations of %10 h.11 However, longer storage durations are likely needed to achieve approximately 70% to 90% of annual energy from wind or solar, with storage durations of 10 to the low hundreds of hours.10,14 Given the need for completely reliable electricity supply, storage durations may enter the seasonal and even inter-year realm if the fraction of annual energy from wind and solar approaches 100%.10 Achieving a grid powered only by variable sources like wind and solar is generally considered both less economical and higher risk than using a suite of generation technologies, with some that are completely dispatchable and/or that provide baseload.15 For context on the storage durations in Figure 1, natural gas in the US is stored in quantities equivalent to durations of tens to many thousands of hours of consumption, either in the extensive pipeline infrastructure itself or in below-ground storage facilities that hold over 4 trillion cubic feet of working natural gas (1,200 TWh on a primary energy basis, enough to completely supply 2 months of US electricity consumption).16 Similarly, coal power plants typically store 30–60 days of fuel onsite, while nuclear plants run for 2 years before refueling. While it is sometimes stated that electricity is unique in that it is a commodity with little storage, in fact there is a massive, long-duration storage system for the fuels used to make electricity and large numbers of dispatchable generators, to ensure a highly reliable and dynamic electricity supply. LDES systems could also be directly integrated with a wind and/or solar installation behind a point of common interconnection, providing dispatchable output that would mimic, at the level of a single project, current fossil generators. For example, in June 2019, a US utility, NV Energy, announced three solar projects with a combined capacity of 1,200 MW with 590 MW of battery storage; the battery storage systems, which range from 4–5 h of duration, increase the availability of power from 30% to 65%.17 Although 4–5 h of storage doubles the availability of this solar installation, it remains far from the availability of a traditional fossil fuel or nuclear power plant. Depending on the location (e.g., Maine versus Arizona), asset type (solar, wind, or a mix), and desired output shape (peaker versus baseload), storage systems with tens to approximately 100 h of duration can in many cases deliver the desired output across greater than 90% of the hours in
Joule 4, 1–12, January 15, 2020
3
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
given year.12,18 Indeed, variable renewable-plus-storage plants with 100% expected availability over a 20-year period may be designed with output prices for specific generation profiles of <0.10 $/kWh given LDES storage costs of 1,000 $/kW and 20 $/kWh and durations around 100 h at rated power.12 Finally, LDES may also provide enhanced resiliency at the level of a (micro)grid or single building. The example of Hurricane Maria mentioned above provides a concrete example of the risks associated with a lack of system resiliency, and researchers have recently begun to quantify the value that energy storage brings in terms of resiliency.19 Although Hurricane Maria was an extreme event that left parts of Puerto Rico powerless for nearly one year, there are numerous instances where tens of hours of distributed or behind-the-meter storage would be sufficient for a system to remain online during a loss of power. This function has traditionally been served by diesel-fueled generators. However, concerns regarding reliability, fuel supply during extended disruptions, fuel costs, maintenance labor, and emissions are driving operators of sites such as hospitals, data centers, and wastewater treatment facilities to explore alternatives. LDES could fulfill such a role and would have the added potential to provide additional revenue by participating in other applications such as local ancillary services markets. Data centers, which can consume hundreds of MW of power, are indeed considering storage systems to provide both backup power and revenue from market participation,20 with one report showing that revenue from peak shaving and frequency regulation could reduce utility bills by as much as 12%.21 Economics and Operation of Long-Duration Electricity Storage Systems In recent years, efforts have centered on development of electricity storage systems with installed capital costs of 150 $/kWh and 5 h duration at rated power.22,23 The distinct characteristics of LDES applications entail a different storage technology cost structure, and result in requirements that offer both challenges and opportunities for technology developers. In the US, deregulated electricity markets typically provide compensation for both electricity generation and reliability services in the form of ancillary and capacity markets. It is technically feasible for storage systems to ‘‘stack’’ a range of services to maximize return on investment, resulting in duty cycles that superpose the contributions of several individual applications24 (although this may not be operationally possible due to market rules or contractual obligations). This breadth of value propositions, along with the ongoing policy-driven evolution of storage valuation in electricity markets (e.g., Federal Energy Regulatory Commission [FERC] Order 841 and the Electricity Market Design Directive in the European Union), prevents establishment of a simple or single set of technical and economic metrics that, if met, would result in wide-scale deployment of LDES technologies.25–27 Analysis of the potential role of LDES in all regions of the world should receive additional future attention.28–31 The aim of this analysis is to identify key economic and engineering tradeoffs for LDES systems with limited assumptions about use case and remuneration mechanisms. Economic frameworks developed for specific applications show similar results to what we present here.12,18 We explore the general LDES cost-performance parameter space using the discounted cash flow framework shown in Equation 1 (see the Supplemental Information for equation details). T X t =1
h i h i ð1 + rÞt DE;t nc;t + RP;t d 1 = CP d 1 + CE;th h1 D +
T X t =1
h ð1 + rÞt nc;t PC;t h1 RTE 1
i L=nc + nc;t VOMt + d 1 FOMt + CRe ð1 + rÞ (Equation 1)
4
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
The left side of Equation 1 represents revenue over the financial life of the project and the right side represents the total cost of ownership (TCO) of the system, both with units of $/kWh. The first revenue term on the left side of Equation 1 scales with the frequency and average price differential of charging and discharging modes (DE ) (representative of arbitrage applications). The second scales with power capacity (RP ) (representative of resource adequacy or capacity market remuneration). The TCO is broken into five principle components, with the first representing the installed capital cost. The discounted terms in the second set of brackets represent (in order) the expense related to efficiency losses, variable operating and maintenance cost, and fixed operating and maintenance cost that scales with the power capacity of the system. The final term explicitly accounts for replacement of the energy storage medium due to capacity fade (cycle life is considered here, but calendar fade may also be a significant factor for certain systems). Equating the present value of revenue and cost implies a zero net present value for a given project term and internal rate of return. Calculations throughout assume T = 20 years and r = 10%.32 We assume PC = 0.025 $/kWh in Figure 2, a representative value for the LCOE of future wind or solar generation in most locations in the US.33,34 This value does not include transmission or distribution costs and therefore reflects the price of electricity at the site of wind or solar generation. For simplicity, taxes, depreciation, inflation, and finance costs are not included. A single-factor sensitivity analysis that explores a broader range of values for the various revenue and cost parameters in Equation 1 is provided in the Supplemental Information (Figure S1) for a 50-h duration storage system. In addition, the Supplemental Information provides a discussion and a figure (Figure S3) that quantifies the impact of PC (including a value of 0 $/kWh, reflecting excess electricity on the grid) and round-trip efficiency (RTE) on the installed capital cost for a range of durations and values of RP . Of the parameters outlined in Equation 1, researchers pursuing new electricity storage technologies can most readily estimate the capital cost of a given system, even at relatively early stages of development. We therefore show installed capital costs that encompass both power and energy components (CP d 1 + CE;th h1 D ) in Figure 2. The z axis represents the available budget for capital expenditure as a function of hRTE and nc . We assume two revenue streams: (1) arbitrage electricity sales that scale with cycle frequency (DE ) and (2) payments for reliability services that scale primarily with power capacity (RP ). Consideration of both remuneration mechanisms is important given that, as an example, PSH units often spend over 50% of their operational hours in reserve shutdown mode, ready to supply power when needed.35 We assume an average price differential for charging and discharging modes of DE = 0.05 $/kWh-cycle, which would result in a compelling LDES technology under the cost structure of today’s grid. To illustrate, if half of the electricity produced by a wind or solar plant generated at 0.025 $/kWh passed through a co-located storage device with a cycle ‘‘premium’’ of 0.05 $/kWh-cycle (i.e., discharge price of 0.075 $/kWh-cycle), the average electricity price for the combined generator plus storage system would be 0.05 $/kWh, a price competitive with electricity generated by future combined cycle natural gas plants (which are in the range of 0.041 to 0.074 $/kWh).36 With 90% flowing through the storage asset, the combined LCOE would still be competitive at 0.07 $/kWh. Thus, storage systems operating in electricity markets with DE = 0.05 $/kWh-cycle across a range of durations that extend beyond daily cycling would enable a new class of combined variable renewable-plus-storage generation assets with dispatchability that much more closely matches flexible fossil generators. Regarding RP , as grids incorporate larger fractions of variable energy resources that contribute to low wholesale electricity prices but fail to provide dispatchability, capacity and ancillary service markets may have an outsized impact
Joule 4, 1–12, January 15, 2020
5
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 2. The Influence on Installed Capital Costs of Round-Trip Efficiency, Cycles per Year, Discharge Time, and Capacity Payments. Indexed colors show installed capital costs (CP d 1 + CE;th h1 D , in $/kWh) as a function of system round-trip energy efficiency (hRTE ) and number of cycles completed per year (nc ) for storage durations at rated power of (A) and (B) 10 h, (C) and (D) 50 h, and (E) and (F) 100 h and annual capacity payments of (A), (C), and (E) 25 $/kW-y and (B), (D), and (F) 75 $/kW-y. Values for other fixed parameters in Equation 1 are: VOM = 0.002 $/kWh-cycle and FOM = 1% of installed capital cost (representative of PSH 41 ), PC = 0.025 $/kWh-cycle, DE = 0.05 $/kWh-cycle, and no storage medium replacements. Note that the number of annual full system cycles cannot exceed 4,380 h/d, hence the different limits on the x axes. Figure S2 in the Supplemental Information presents results for DE = 0.10 $/kWh-cycle.
on the economic viability of storage projects,37 particularly for LDES systems that can provide nearly dispatchable output. We examine two different values of RP (25 and 75 $/kW-y) in Figure 2 to account for the uncertainty range associated with duration and capacity value that continues to be a topic of debate for system operators and market stakeholders, and provide additional parameter analysis in Figures S1 and S2 of the Supplemental Information.38–40 The upper right corners of Figures 2A and 2B reflect the capital cost targets (around 150 $/kWh) typical for daily cycling of efficient systems such as Li-ion batteries.22,23 As discussed in more detail in the context of Figure 3, it is important to note that these capital costs include the costs of both the power subsystem (Cp =d) and the energy subsystem (CE;th =hd ). While challenging, with an ability to leverage the major investments in Li-ion technology taking place for automotive applications, together with potential cost reductions in balance of plant components including HVAC and power electronics, such capital cost values may be approached in the coming decade.42 However, consideration of (C) and (D) for 50-h duration and (E) and (F) for 100-h duration (Figure 2), show that much lower costs are required than Li-ion is expected to achieve. For 50 h, installed capital costs of about 20 to 35 $/kWh
6
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 3. Relationship between Power and Energy Capital Costs Derived from Figure 2 for RP = 25 $/kW-y, Two hRTE , and Three Durations Values used for total capital cost were extracted from Figure 2 at 70% of the maximum annual complete cycles for a given duration. Figure S4 of in the Supplemental Information shows results for DE = 0.10 $/kWh-cycle, while Figures S5 and S6 show results for R p = 75 $/kWh-y and DE = 0.05 $/kWh-cycle and DE = 0.10 $/kWh-cycle, respectively. This figure also includes annotations for the approximate power and energy installed capital costs for underground chemical storage (shown to the left of the y axis because the value for CE;th is <1 $/kWh), 43 PSH, 7 Compressed air energy storage (CAES), 7 and future Li-ion. 44
are required, while for 100 h, installed capital costs of around 5 to 15 $/kWh are required. With DE = 0.05 $/kWh-cycle, the cost targets shown here for a 100-h duration are likely at or beyond the limit of what any manufactured and installed storage system may achieve. Also of key importance in Figure 2 is the RTE. While the sensitivity to RTE is higher with a capacity payment of 25 $/kW-y (Figures 2A, 2C, and 2E), the capital cost targets are extremely challenging if hRTE < 50%, and increasingly favorable with increasing RTE. It is also important to note that Figure 2 includes CE;th =hD for the capital cost of energy. In other words, the system efficiency has a 2-fold impact: lost revenue from purchased energy that is converted to noncompensated heat rather than electricity and (through hD ) a more challenging capital cost target for the energy subsystem. While Figure 2 provides the total installed capital cost for delivered energy (Cp =d + CE;th =hd ), Figure 3 splits apart the capital cost goals for the power and energy subsystems (assuming that the power block and energy blocks are separable), showing values for CP and CE;th . Three system durations are shown, along with two values of the RTE (as green and red lines). The results in Figure 3 emphasize the key importance and limitations of energy costs. Even excluding power subsystem costs, for durations of 50 or 100 h, energy capital costs cannot exceed 20 $/kWh; for 10 h duration the energy capital costs can be substantially higher. It is important to note that those expenses are inclusive of the energy storage medium, the container used to hold the energy storage medium, any additional balance of plant (BOP) costs, and installation. On the other hand, if such low energy capital costs can be achieved, capital costs for the power subsystem can be in the range of 500 to 1,000 $/kW, which are challenging, but not completely out of reach, in the context of present costs for turbines, flow battery stacks, and other candidate power conversion technologies. Technology Approaches for LDES Systems Figure 3 also includes annotations for several major classes of storage technology (power-to-gas, PSH, compressed air energy sequence [CAES], and Li-ion). Based on presently deployed technologies, PSH is a strong candidate for storage across a wide
Joule 4, 1–12, January 15, 2020
7
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
range of durations because of its low energy capital costs, adequate power capital cost (achieved in part through the use of reversible turbines), and high RTE. The cost structure of Li-ion positions it comparatively well for daily cycling applications, where the high capital costs for energy can be paid for by frequent cycling (and hence, frequent revenue collection with the stored energy). Another class of storage technology that is often discussed in the context of long-duration is power-to-gas (or other chemicals), and making use of either the existing pipeline capacity or underground reservoirs for storage.45 Such methods offer the exceptionally low storage costs required for long-duration storage (consistent, of course, with the fact that months of natural gas are stored in such structures, as indicated above), that is, <0.10 $/kWh.43 As one recent example, Air Liquide developed a hydrogen storage facility to provide 30 days of backup supply for steam methane reformers on the US Gulf of Mexico coast; this facility has a volume of 6 million m3 and likely holds 50 GWh of working hydrogen gas (on a primary energy basis).46 Unfortunately, when a chemical like H2 or CH4 is first synthesized (with electricity energy input) and then converted back to electricity, the RTE will likely be <40%, limiting the suitability for the LDES economics described in Figures 2 and 3. The use of solid-oxide fuel cells (SOFCs) may help overcome these low RTEs, as discussed below. Several references contain summary tables on major classes of storage technologies.7,47,48 The target area for LDES systems, which use containerized energy allowing flexible siting and with cost targets drawn from Figure 2, lies in the region spanning the plotted lines in Figure 3. Capital costs for energy may be substantially higher than approaches that use geologic features for storage (and hence have more limited siting), and the RTE should likely exceed 50% in order to provide sufficient revenue to cover capital costs. Given the critical importance of the energy capital cost, careful attention should be given to both the storage media capital costs as well as the costs of containerization (including the container in contact with the energy storage media, any secondary or tertiary containment, as well as the shipping container or building that may be used to help with thermal conditioning), shipping, and installation. Figure S7 shows typical storage media and container capital costs, which constitute a lower bound on storage costs. As shown in Figure 3, acceptable installed energy capital costs range from just a few $/kWh (for 100-h duration, 50% RTE) to around 75 $/kWh (for 10-h duration, 80% RTE). To pick one value as an exemplar, for a 50-h duration and 80% RTE, to ensure that the costs of containment as well as shipping, site preparation, and installation are consistent with an energy subsystem cost targets of <20 $/kWh, the energy density of all storage media should preferably be R0.1 kWh/L. Several technology classes are under development that may approach the cost and efficiency targets depicted in Figure 3. Here, we focus on relatively new approaches that may be particularly well suited for 10- to 100-h duration that is a focus in this paper. In the area of thermal systems, latent, sensible, and thermochemical mechanisms are all under development. A key requirement for achieving a system RTE of R50% is Tcold =Thot < 0:2, given Carnot constraints on thermal storage. This can be achieved either by increasing Thot , reducing Tcold (which is often the ambient, but need not be), or both. For example, pumped heat thermal storage uses both cold and hot reservoirs for storage,49,50 and several system development efforts are underway.51 Another option is to substantially increase Thot , either to the upper temperature limits for gas turbines52–54 or even substantially beyond through the use of non-turbine heat-to-electricity conversion.55 High temperatures (>1,000 C) may be achieved with resistive heaters, which can offer extremely low capital costs on a $/kW basis depending on the upper temperature.56 One challenge for thermal systems is scaling; to minimize thermal losses (or insulation costs), large containment vessels are needed, and when turbines are used, that also contributes to system
8
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
sizing constraints, with the most favorable economics in the tens to hundreds of MW. Electrochemical systems are another technology class in which LDES systems are being developed. From the point of view of active materials alone, numerous battery chemistries have a capital cost that could meet the LDES targets outlined above.57–60 Indeed, the principal grid storage battery prior to Li-ion was the Na/S system, for which active material costs are <10 $/kWh. While starting with lowcost reactants is required, we also emphasize the need for a sufficiently high energy density (e.g., for a duration of 50 h the energy density should be R0.1 kWh/L, including all active materials) to limit container and installation costs. Another important class of electrochemical technologies for LDES is reversible SOFCs that have a clear technical capability to achieve high RTE when cycling between CO2 and/or H2O (on discharge) and CH4 and/or H2 (on charge).61,62 Mechanical systems are a third technology category under development with potential to address LDES markets. Adiabatic compressed air storage with integrated thermal storage offers substantially better RTE than conventional natural-gas fired systems.63 While the raw materials for mechanical storage systems generally range from low cost to free,64 the issue has been overcoming low energy and power densities that necessitate large earthworks or engineering projects to achieve appreciable capacities. New approaches that utilize existing or readily generated subterranean structures may help overcome this longstanding challenge.65 Beyond careful selection of low-cost energy storage media, containment, and other system attributes, the unique requirements of LDES—in particular the fact that the system may rarely undergo a full cycle of its stored energy—may offer opportunities for fundamentally different and innovative approaches and system configurations. As one example, Figure 4 provides a graphical depiction of an approach in which, rather than a ‘‘single tank’’ for the energy storage media, storage media with different attributes—which may include energy density, material identity, or purity, amount of thermal insulation, pressure, temperature, etc.—could be accessed in a staged and sequential manner. This staging could allow optimization for cost, efficiency, cycle life, and other metrics that result in superior overall system cost and performance. As one concrete example, in a Zn/Br2 flow battery when the Br2 exceeds the solubility limit in the ZnBr2 aqueous solution, it will precipitate into a separate phase with a higher density, providing a natural means of separation and storage.66 The energy density of 2.5 M ZnBr2 (aq) is 170 Wh/L (assuming 80% can react), while the energy density of liquid Br2 and solid Zn exceeds 1,400 Wh/L. If a Zn/Br2 system could be designed with a small volume of ZnBr2 (aq) compared to solid Zn and liquid Br2, the cost of the container alone could potentially be reduced by >5 $/kWh (with other costs such as site preparation and shipping also lowered), substantially improving the economics for LDES. Another innovative approach to the challenging economic targets of LDES is to add an energy storage subsystem on to an existing power conversion unit.67 For example, a thermal storage unit could be added to an existing turbine, with heat added to the thermal store using resistive heating or another method, and the existing turbine powered by either its intended source of thermal energy (e.g., natural gas), thermal energy from the store, or a combination of the two. In this approach, the power costs may consist only of the additional BOP required to connect the thermal store with the existing turbine, as well as the resistive heaters (for charging power), providing more room for energy costs while still meeting the overall LDES cost targets. These approaches are just two examples of the types of fundamentally different approaches to designing energy storage systems that are uniquely suited to the new technoeconomic design space of LDES applications.
Joule 4, 1–12, January 15, 2020
9
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 4. A Storage System with a Single Power Block and Storage Media with Attributes that Depend on Cycling Frequency Conceptual schematic of a fundamentally different system architecture in which the varied cycling frequencies associated with LDES applications can be leveraged to create a series of energy storage tanks with attributes optimized to reduce capital costs, rather than using a single, uniform, tank across all durations.
Outlook and Perspective Long-duration electricity storage systems could be one important route to make use of wind and solar and achieve zero-carbon electricity goals as well as serve other applications like backup power. In this work, we focused on durations between 10 and 100 h, with the lower limit at the upper end of daily cycling, and the upper end at or beyond the likely economic limits of any current containerized system that can be flexibly sited. Our economic analysis, and those of others, have found that achieving low installed energy capital costs (between 5 and 35 $/kWh) is of critical importance, which in turn will require low-cost and high-energy-density reactants. In addition, for the values assumed in our economic analysis, an RTE of >50% is likely required to provide enough revenue to finance the up-front capital costs. While approaches like power-to-gas can achieve extremely low (<0.10 $/kWh) capital costs when using underground storage, the RTE of these approaches is usually <40%. However, longer-duration systems that can also serve shorter duration applications (e.g., daily cycling) can potentially provide greater value to systems with durations of >100 h, even if the RTE is <50%. As solar and wind capacity continue to be built, and economic incentives and policies put in place to achieve 100% carbon-free grid goals, LDES systems may offer an important option to advance low-cost and carbon-free electricity in markets around the world. However, a challenge for any new stationary electricity storage system is the low-margin, risk-averse nature of the customers in the utility industry. Without a substantial (ideally, tens of billions of dollars cumulatively) and high-margin set of first markets (which Li-ion had in the form of portable electronics), it will be difficult to achieve the manufacturing volume and project quantity to consistently drive down costs through scale and learning. As such, it is likely that LDES systems will require long-term vision, development, and investment strategies in both the public and private sectors if these potentially game-changing technologies are to become a cornerstone of a future grid that is low cost, low carbon, and resilient.
10
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j.joule. 2019.11.009.
ACKNOWLEDGMENTS Paul Albertus and Joseph S. Manser acknowledge the Advanced Research Projects Agency–Energy, of the United States Department of Energy, where they were employed when a portion of this manuscript was conceived and written. This manuscript describes, in part, the motivation and framing of the ARPA-E DAYS program.4
REFERENCES 1. U.S. Energy Information Administration (2019). Levelized cost and Levelized avoided cost of new generation resources in the annual energy Outlook 2019. https://www.eia.gov/outlooks/ aeo/pdf/electricity_generation.pdf. 2. Clean Air Task Force (2019). Fact Sheet: state and utility climate change targets shift to carbon reductions, technology diversity. https://www.catf.us/wp-content/uploads/ 2019/05/State-and-Utility-Climate-ChangeTargets.pdf. 3. United States Department of Energy Washington, DC (2018). Energy resilience solutions for the Puerto Rico grid. https://www. energy.gov/sites/prod/files/2018/06/f53/ DOE%20Report_Energy%20Resilience% 20Solutions%20for%20the%20PR%20Grid% 20Final%20June%202018.pdf. 4. The Advanced Research Projects AgencyEnergy (ARPA-E) (2018). Duration addition to electricity storage (DAYS) overview. https:// arpa-e.energy.gov/sites/default/files/ documents/files/DAYS_Program Overview_FINAL.pdf. 5. U.S. Energy Information Administration (2017). Energy storage and renewables beyond wind, hydro, solar make up 4% of U.S. power capacity, Today in Energy. https://www.eia. gov/todayinenergy/detail.php?id=31372. 6. National Grid. (2017). National Grid develops innovative solution for an island community’s unique energy challenges. https://news. nationalgridus.com/2017/11/national-griddevelops-innovative-solution-islandcommunitys-unique-energy-challenges/. 7. Akhil, A.A., Huff, G., Currier, A.B., Kaun, B.C., Rastler, D.M., Chen, S.B., Cotter, A.L., Bradshaw, D.T., and Gauntlett, W.D. (2013). DOE/EPRI 2013 Electricity Storage Handbook in Collaboration with NRECA (Sandia National Laboratories). 8. Denholm, P., and Margolis, R. (2016). Energy storage requirements for achieving 50% solar photovoltaic energy penetration in California (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/docs/fy16osti/ 66595.pdf. 9. Denholm, P., and Hand, M. (2011). Grid flexibility and storage required to achieve very high penetration of variable renewable electricity. Energy Policy 39, 1817–1830. 10. Shaner, M.R., Davis, S.J., Lewis, N.S., and Caldeira, K. (2018). Geophysical constraints on
the reliability of solar and wind power in the United States. Energy Environ. Sci. 11, 914–925. 11. Denholm, P., and Mai, T. (2019). Timescales of energy storage needed for reducing renewable energy curtailment. Renew. Energy 130, 388–399. 12. Ziegler, M.S., Mueller, J.M., Pereira, G.D., Song, J., Ferrara, M., Chiang, Y., and Trancik, J.E. (2019). Storage requirements and costs of shaping renewable energy Toward grid decarbonization. Joule 3, 2134–2153. 13. Jones, R. (2017). System perspectives on long duration storage, ARPA-E, Evolved Energy Research. https://docs.wixstatic.com/ugd/ a35761_773bbc576d8b4b02ae2c02ad93c6166f. pdf. 14. De Sisternes, F.J., Jenkins, J.D., and Botterud, A. (2016). The value of energy storage in decarbonizing the electricity sector. Appl. Energy 175, 368–379. 15. Clack, C.T.M., Qvist, S.A., Apt, J., Bazilian, M., Brandt, A.R., Caldeira, K., Davis, S.J., Diakov, V., Handschy, M.A., Hines, P.D.H., et al. (2017). Evaluation of a proposal for reliable low-cost grid power with 100% wind, water, and solar. Proc. Natl. Acad. Sci. USA 114, 6722–6727. 16. U.S. Energy Information Administration (2019). Weekly natural gas storage report. http://ir.eia. gov/ngs/ngs.html. 17. Merchant, E.F. (2016). NV Energy Announces ‘Hulkingly Big’ solar-plus-storage procurement (Green Tech Media), June 25, 2019. https:// www.greentechmedia.com/articles/read/nvenergy-signs-a-whopping-1-2-gigawatts-ofsolar-and-590-megawatts-of-stor#gs.w3wb9i. 18. Braff, W.A., Mueller, J.M., and Trancik, J.E. (2016). Value of storage technologies for wind and solar energy. Nat. Clim. Change 6, 964–969. 19. Laws, N.D., Anderson, K., DiOrio, N.A., Li, X., and McLaren, J. (2018). Impacts of valuing resilience on cost-optimal PV and storage systems for commercial buildings. Renew. Energy 127, 896–909. 20. Storrow, B. (2017). How big batteries at data centers could replace power plants, E&E News, July 19, 2018. https://www.scientificamerican. com/article/how-big-batteries-at-datacenters-could-replace-power-plants/. 21. Shi, Y., Xu, B., Wang, D., and Zhang, B. (2017). Using battery storage for peak shaving and frequency regulation: joint optimization for
superlinear gains. IEEE Trans. Power Syst. 33, 2882–2894. 22. The Advanced Research Projects AgencyEnergy (2018). Integration and optimization of Novel ion conducting solids (IONICS) program overview. https://arpa-e.energy.gov/sites/ default/files/documents/files/IONICS_ ProgramOverview.pdf. 23. Darling, R.M., Gallagher, K.G., Kowalski, J.A., Ha, S., and Brushett, F.R. (2014). Pathways to low-cost electrochemical energy storage: a comparison of aqueous and nonaqueous flow batteries. Energy Environ. Sci. 7, 3459–3477. 24. Davies, D.M., Verde, M.G., Mnyshenko, O., Chen, Y.R., Rajeev, R., Meng, Y.S., and Elliott, G. (2019). Combined economic and technological evaluation of battery energy storage for grid applications. Nat. Energy 4, 42–50. 25. Balducci, P.J., Alam, M.J.E., Hardy, T.D., and Wu, D. (2018). Assigning value to energy storage systems at multiple points in an electrical grid. Energy Environ. Sci. 11, 1926– 1944. 26. Katsanevakis, M., Stewart, R.A., and Lu, J. (2017). Aggregated applications and benefits of energy storage systems with applicationspecific control methods: a review. Renew. Sustain. Energy Rev. 75, 719–741. 27. Oh, U., Choi, J., Kim, K., Cha, J., and Lee, K.Y. (2018). A study on capacity credit and ESS evaluation for WTG and multi-ESS in power system. IFAC-PapersOnLine 51, 516–521. 28. Jentsch, M., Trost, T., and Sterner, M. (2014). Optimal use of power-to-gas energy storage systems in an 85% renewable energy scenario. Energy Procedia 46, 254–261. 29. Barbour, E., Wilson, I.A.G., Radcliffe, J., Ding, Y., and Li, Y. (2016). A review of pumped hydro energy storage development in significant international electricity markets. Renew. Sustain. Energy Rev. 61, 421–432. 30. Zafirakis, D., Chalvatzis, K.J., Baiocchi, G., and Daskalakis, G. (2016). The value of arbitrage for energy storage: evidence from European electricity markets. Appl. Energy 184, 971–986. 31. Blakers, A., Lu, B., and Stocks, M. (2017). 100% renewable electricity in Australia. Energy 133, 471–482. 32. McLaren, J., Anderson, K., Laws, N., Gagnon, P., DiOrio, N., and Li, X. (2018). Identifying critical factors in the cost-effectiveness of solar
Joule 4, 1–12, January 15, 2020
11
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
and battery storage in commercial buildings (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/docs/fy18osti/ 70813.pdf. 33. US Department of Energy’s Solar Energy Technologies Office (2017). The SunShot 2030 Goals: 3¢ per kilowatt hour for PV and 5¢ per killowatt hour for dispatchable CSP. https:// www.energy.gov/sites/prod/files/2018/05/f51/ SunShot%202030%20Fact%20Sheet.pdf. 34. U.S. Department of Energy. WindVision: a new era for wind power in the United States. https:// www.energy.gov/sites/prod/files/2015/03/f20/ wv_full_report.pdf. 35. Urı´a-Martı´nez, R., Johnson, M.M., O’Connor, P.W., Samu, N.M., Witt, A.M., Battey, H., Welch, T., Bonnet, M., and Wagoner, S. (2018). 2017 Hydropower market report (U.S. department of energy). https://www.energy. gov/sites/prod/files/2018/04/f51/Hydropower %20Market%20Report.pdf. 36. Lazard. (2018). Lazard’s Levelized Cost of Energy Analysis–12.0 version. https://www. lazard.com/media/450784/lazards-levelizedcost-of-energy-version-120-vfinal.pdf. 37. Jenkin, T., Beiter, P., and Margolis, R. (2016). Capacity payments in restructured markets under low and high penetration levels of renewable energy (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/ docs/fy16osti/65491.pdf. 38. Carden, K. (2019). Valuing capacity for resources with energy limitations – independent assessment, Astrape consulting. https://www.nyiso.com/documents/20142/ 5020603/Astrape%20presentation%20021519. pdf. 39. Hall, W., Zhang, B., and Legnard, T. (2019). Valuing capacity for resources with energy limitations, GE Energy Consulting. https:// www.nyiso.com/documents/20142/4358080/ 01082019%20Capacity%20Value%20of% 20Resources%20with%20Energy% 20Limitations_v2.pdf/3499da16-12d8-16b7b12f-be7650e64b63. 40. Pfeifenberger, J. (2019). Capacity markets and wholesale market outcomes. https:// brattlefiles.blob.core.windows.net/files/ 16366_capacity_markets_and_wholesale_ market_outcomes.pdf. 41. The International renewable energy agency (2012). Renewable energy technologies: cost analysis series-Hydropower. https://www.irena. org/documentdownloads/publications/ re_technologies_cost_analysis-hydropower. pdf. 42. BloombergNEF (2019). New energy outlook 2019. https://about.bnef.com/new-energyoutlook/. 43. Federal Energy Regulatory Commission (2004). Current state of and issues concerning underground natural gas storage. https://www.
12
Joule 4, 1–12, January 15, 2020
ferc.gov/EventCalendar/Files/200410 20081349-final-gs-report.pdf. 44. Nykvist, B., and Nilsson, M. (2015). Rapidly falling costs of battery packs for electric vehicles. Nat. Clim. Change 5, 329–332. 45. Go¨tz, M., Lefebvre, J., Mo¨rs, F., McDaniel Koch, A., Graf, F., Bajohr, S., Reimert, R., and Kolb, T. (2016). Renewable Power-to-Gas: A technological and economic review. Renew. Energy 85, 1371–1390. 46. Air liquide worldwide (2017). USA: air liquide operates the world’s largest hydrogen storage facility. https://en.media.airliquide.com/news/ usa-air-liquide-operates-the-worlds-largesthydrogen-storage-facility-edf8-56033.html. 47. Abdon, A., Zhang, X., Parra, D., Patel, M.K., Bauer, C., and Worlitschek, J. (2017). Technoeconomic and environmental assessment of stationary electricity storage technologies for different time scales. Energy 139, 1173–1187. 48. Divya, K.C., and Østergaard, J. (2009). Battery energy storage technology for power systems—an overview. Electr. Power Syst. Res. 79, 511–520. 49. Laughlin, R.B. (2017). Pumped thermal grid storage with heat exchange. J. Renew. Sustain. Energy 9, 044103.
55. Amy, C., Seyf, H.R., Steiner, M.A., Friedman, D.J., and Henry, A. (2019). Thermal energy grid storage using multi-junction photovoltaics. Energy Environ. Sci. 12, 334–343. 56. Forsberg, C.W., Stack, D.C., Curtis, D., Haratyk, G., and Sepulveda, N.A. (2017). Converting excess low-price electricity into hightemperature stored heat for industry and highvalue electricity production. Electr. J. 30, 42–52. 57. Wadia, C., Albertus, P., and Srinivasan, V. (2011). Resource constraints on the battery energy storage potential for grid and transportation applications. J. Power Sources 196, 1593–1598. 58. Li, Z., Pan, M.S., Su, L., Tsai, P., Badel, A.F., Valle, J.M., Eiler, S.L., Xiang, K., Brushett, F.R., and Chiang, Y. (2017). Air-breathing aqueous sulfur flow battery for ultralow-cost longduration electrical storage. Joule 1, 306–327. 59. Soloveichik, G.L. (2015). Flow batteries: current status and trends. Chem. Rev. 115, 11533– 11558. 60. Hawthorne, K.L., Petek, T.J., Miller, M.A., Wainright, J.S., and Savinell, R.F. (2015). An investigation into factors affecting the iron plating reaction for an all-iron flow battery. J. Electrochem. Soc. 162, A108–A113.
50. White, A., Parks, G., and Markides, C.N. (2013). Thermodynamic analysis of pumped thermal electricity storage. Appl. Therm. Eng. 53, 291–298.
61. Jensen, S.H., Graves, C., Mogensen, M., Wendel, C., Braun, R., Hughes, G., Gao, Z., and Barnett, S.A. (2015). Large-scale electricity storage utilizing reversible solid oxide cells combined with underground storage of CO2 and CH4. Energy Environ. Sci. 8, 2471–2479.
51. Hanley, S. Google X spins off Malta molten salt energy storage business. https:// cleantechnica.com/2018/12/21/google-xspins-off-malta-molten-salt-energy-storagebusiness/.
62. Wendel, C.H., and Braun, R.J. (2016). Design and techno-economic analysis of high efficiency reversible solid oxide cell systems for distributed energy storage. Appl. Energy 172, 118–131.
52. Randhir, K., King, K., Rhodes, N., Li, L., Hahn, D., Mei, R., AuYeung, N., and Klausner, J. (2019). Magnesium-manganese oxides for high temperature thermochemical energy storage. J. Energy Storage 21, 599–610.
63. Bullough, C., Gatzen, C., Jakiel, C., Koller, M., Nowi, A., and Zunft, S. (2004). Advanced adiabatic compressed air energy storage for the integration of wind energy. In Proceedings of the European Wind Energy Conference, EWEC, pp. 22–25.
53. Ma, Z., Mehos, M., Glatzmaier, G., and Sakadjian, B.B. (2015). Development of a concentrating solar power system using fluidized-bed technology for thermal energy conversion and solid particles for thermal energy storage. Energy Procedia 69, 1349– 1359. 54. Ma, Z., Zhang, R., and Sawaged, F. (2017). Design of Particle-Based Thermal Energy Storage for a Concentrating Solar Power System. In the Proceedings of the ASME 2017 11th international conference on energy sustainability collocated with the ASME 2017 power conference joint with ICOPE-17, the ASME 2017 15th international conference on fuel cell science, engineering and technology, and the ASME 2017 nuclear forum. ASME 2017 11th International conference on energy sustainability. (American Society of Mechanical Engineers).
64. Fyke, A. (2019). The fall and rise of Gravity storage technologies. Joule 3, 625–630. 65. Moodie, A. (2017). A Texas startup’s big energy idea: storing electricity underground, The Guardian. https://www.theguardian.com/ sustainable-business/2016/aug/25/electricitystorage-underground-quidnet-energy. 66. Biswas, S., Senju, A., Mohr, R., Hodson, T., Karthikeyan, N., Knehr, K.W., Hsieh, A.G., Yang, X., Koel, B.E., and Steingart, D.A. (2017). Minimal architecture zinc–bromine battery for low cost electrochemical energy storage. Energy Environ. Sci. 10, 114–120. 67. Forsberg, C., and Peterson, P.F. (2016). Basis for fluoride salt–cooled high-temperature reactors with nuclear air-Brayton combined cycles and firebrick resistance-heated energy storage. Nucl. Technol. 196, 13–33.
Perspective
Long-Duration Electricity Storage Applications, Economics, and Technologies Paul Albertus,1,4,* Joseph S. Manser,2,4 and Scott Litzelman3
The United States electricity grid is undergoing rapid changes in response to the sustained low price of natural gas, the falling cost of electricity from variable renewable resources (which are increasingly being paired with Li-ion storage with durations up to 4 h at rated power), and state and local decarbonization policies. Although the majority of recent electricity storage system installations have a duration at rated power of up to 4 h, several trends and potential applications are identified that require electricity storage with longer durations of 10 to 100 h. Such a duration range lies between daily needs that can be satisfied with technologies with the cost structure of lithium-ion batteries and seasonal storage utilizing chemical storage in underground reservoirs. The economics of long-duration storage applications are considered, including contributions for both energy time shift and capacity payments and are shown to differ from the cost structure of applications well served by lithium-ion batteries. In particular, the capital cost for the energy subsystem must be substantially reduced to 3 $/kWh (for a duration of 100 h), 7 $/kWh (for a duration of 50 h), or 40 $/kWh (for a duration of 10 h) on a fully installed basis. Recent developments in major technology classes that may approach the targets of the long-duration electricity storage (LDES) cost framework, including electrochemical, thermal, and mechanical, are briefly reviewed. This perspective, which illustrates the importance of low-cost and high-energy-density storage media, motivates new concepts and approaches for how LDES systems could be economical and provide value to the electricity grid. Introduction and Applications for Long-Duration Energy Storage The United States (US) electricity grid is undergoing rapid changes that create opportunities for new electricity storage applications and may benefit from new electricity storage technologies. First, the levelized cost of electricity (LCOE) from wind and solar photovoltaics is now lower than the new natural-gas-combined cycle power plants, even as sustained low natural gas prices are shifting the fuel mixture away from coal and, in some cases, nuclear.1 However, unlike the preceding century in which the electricity system was based on a stable supply of chemical fuel, the electricity output from wind and solar installations is variable over time scales ranging from seconds to years. The second major change is a move toward decarbonization. At the time of publication, six US states had policies of 100% clean energy or renewable energy targets signed into law with more under consideration within various state legislatures; many countries, regions, and cities are also pursuing such policies.2 Third, states, municipalities, and industries are considering technology options to increase the resiliency of their grids, in part due to rising weather variations and storms, such as Hurricane Maria, which devastated Puerto Rico and the US Virgin Islands.3 In each of these cases, cost-effective long-duration electricity
Context & Scale The feasibility of incorporating a large share of power from variable energy resources such as wind and solar generators depends on the development of cost-effective and application-tailored technologies such as energy storage. Energy storage technologies with longer durations of 10 to 100 h could enable a grid with more renewable power, if the appropriate cost structure and performance—capital costs for power and energy, round-trip efficiency, self-discharge, etc.— can be realized. Although current technologies such as lithium-ion batteries are suitable for a number of applications on the grid, they are not suitable for longerduration storage applications. Although 10 to 100 h energy storage will help facilitate the integration of renewable power on the grid, it is not long enough to last for seasons, and is not sufficient to enable a grid with 100% renewable power. Given the capital-intensive nature of scaling up a storage technology that can be impactful to the overall electricity grid, rigorous technical and economic evaluations at the laboratory and pilot scale are required. Several major classes of
Joule 4, 1–12, January 15, 2020 ª 2019 Elsevier Inc.
1
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
storage (LDES), which we consider as durations at a rated power of 10–100 h, could provide clear benefits.4 Electricity storage services on the grid today are dominated by pumped-storage hydropower (PSH) (in terms of cumulative installations) and lithium-ion (Li-ion) batteries (in terms of share of present annual installations). Molten salt storage at concentrating solar power plants and to a lesser degree on-site thermal storage, are also significant, but here, we focus only on systems that charge with and discharge electricity. PSH, which typically has duration at rated power of 6–10 h, provides peaking capacity and fulfills daily energy time shift applications. However, permitting and large-project financing difficulties have limited additions of PSH.5 Li-ion batteries are technically capable of being used in projects of a wide range of sizes (e.g., kW to GW and kWh to GWh) and for essentially any duration, but economically viable applications pursued today include frequency regulation (with charge and discharge durations on the order of seconds), reserve markets (tens of minutes), transmission and distribution upgrade deferrals (8 h), and others of %10 h of duration.6 Table 19 of reference 7 provides an overview of various electricity storage applications.7
storage technologies may address the long-duration electricity storage cost and performance framework, and efforts are accelerating to identify and develop the most promising storage systems.
To make use of low-cost wind and solar and advance decarbonization goals, LDES systems may be configured in at least two ways: (1) as grid-tied, standalone systems, and (2) integrated with solar and/or wind behind a point of common interconnection. Considering first the case of grid-tied systems, variable resources have traditionally been buffered with natural gas-fired peaker plants, though other approaches include transmission expansion, demand-side management (including on-site thermal storage, both hot and cold, of various durations), and electricity storage. For electricity storage, modeling studies have demonstrated that up to approximately 8 h of duration can increase the amount of annual energy from wind and solar that can be utilized on a large regional grid (e.g., CAISO or ERCOT).8–10 A number of studies have also looked at storage durations longer than approximately 10 h; these have also found that the addition of increasing durations of storage reduces curtailment and increases the use of variable assets like wind and solar, with a falling marginal impact.9–12 Further modeling work is needed to accurately quantify the impact of LDES on wind and solar penetration at the regional level and should include realistic handling of transmission power flow constraints, network stability, contingency requirements, opportunity costs of curtailed energy, limits to load flexibility, and other parameters necessary to capture the full complexity of delivering power within a large electricity system. It is also important to point out that while longer-duration assets (e.g., seasonal storage) can potentially serve shorter duration applications such as daily cycling (assuming the efficiency, conversion kinetics, and other attributes are suitable), the converse is not true, which may favor the development of longer-duration assets.13 In the context of these studies, Figure 1 provides a high-level and semi-quantitative relationship between the maximum storage duration required to meet demand and the fraction of annual energy from wind and solar. The colored region in Figure 1 indicates typical assumptions associated with renewable curtailment, transmission build-out, and grid flexibility (i.e., number of must-run generators, demand response), while the arrows indicate either more restrictive (to the left) or aggressive (to the right) assumptions. This figure shows that with capacity factors of wind and/or solar installations of 20% to 50%, little storage may be needed until higher fractions are reached, especially in regions with optimized blends of wind and solar.8,9 Depending on the characteristics (e.g., quality of solar and/or wind resource, transmission system capabilities, amount of natural gas versus low ramp rate resources) of
2
Joule 4, 1–12, January 15, 2020
1University 2Ionic
of Maryland, College Park, MD, USA
Materials, Inc., Woburn, MA, USA
3Advanced
Research Projects Agency – Energy, Washington, DC, USA
4These
authors contributed equally
*Correspondence: [email protected] https://doi.org/10.1016/j.joule.2019.11.009
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 1. Semi-quantitative Overview of the Maximum Duration of Electricity Storage Needed to Ensure Demand Is Met at All Times versus the Fraction of Annual Energy from Variable Generators, Such as Wind and Solar Arrows indicate the impacts of wind and/or solar curtailment, transmission capacity, geographic extent, and grid flexibility (e.g., number of must-run generators and demand response) on the maximum required duration of electricity storage. The values in this figure are based on previous studies. 8–15
the region, between about 50% and 80% of annual energy from solar and/or wind can be reached with storage durations of %10 h.11 However, longer storage durations are likely needed to achieve approximately 70% to 90% of annual energy from wind or solar, with storage durations of 10 to the low hundreds of hours.10,14 Given the need for completely reliable electricity supply, storage durations may enter the seasonal and even inter-year realm if the fraction of annual energy from wind and solar approaches 100%.10 Achieving a grid powered only by variable sources like wind and solar is generally considered both less economical and higher risk than using a suite of generation technologies, with some that are completely dispatchable and/or that provide baseload.15 For context on the storage durations in Figure 1, natural gas in the US is stored in quantities equivalent to durations of tens to many thousands of hours of consumption, either in the extensive pipeline infrastructure itself or in below-ground storage facilities that hold over 4 trillion cubic feet of working natural gas (1,200 TWh on a primary energy basis, enough to completely supply 2 months of US electricity consumption).16 Similarly, coal power plants typically store 30–60 days of fuel onsite, while nuclear plants run for 2 years before refueling. While it is sometimes stated that electricity is unique in that it is a commodity with little storage, in fact there is a massive, long-duration storage system for the fuels used to make electricity and large numbers of dispatchable generators, to ensure a highly reliable and dynamic electricity supply. LDES systems could also be directly integrated with a wind and/or solar installation behind a point of common interconnection, providing dispatchable output that would mimic, at the level of a single project, current fossil generators. For example, in June 2019, a US utility, NV Energy, announced three solar projects with a combined capacity of 1,200 MW with 590 MW of battery storage; the battery storage systems, which range from 4–5 h of duration, increase the availability of power from 30% to 65%.17 Although 4–5 h of storage doubles the availability of this solar installation, it remains far from the availability of a traditional fossil fuel or nuclear power plant. Depending on the location (e.g., Maine versus Arizona), asset type (solar, wind, or a mix), and desired output shape (peaker versus baseload), storage systems with tens to approximately 100 h of duration can in many cases deliver the desired output across greater than 90% of the hours in
Joule 4, 1–12, January 15, 2020
3
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
given year.12,18 Indeed, variable renewable-plus-storage plants with 100% expected availability over a 20-year period may be designed with output prices for specific generation profiles of <0.10 $/kWh given LDES storage costs of 1,000 $/kW and 20 $/kWh and durations around 100 h at rated power.12 Finally, LDES may also provide enhanced resiliency at the level of a (micro)grid or single building. The example of Hurricane Maria mentioned above provides a concrete example of the risks associated with a lack of system resiliency, and researchers have recently begun to quantify the value that energy storage brings in terms of resiliency.19 Although Hurricane Maria was an extreme event that left parts of Puerto Rico powerless for nearly one year, there are numerous instances where tens of hours of distributed or behind-the-meter storage would be sufficient for a system to remain online during a loss of power. This function has traditionally been served by diesel-fueled generators. However, concerns regarding reliability, fuel supply during extended disruptions, fuel costs, maintenance labor, and emissions are driving operators of sites such as hospitals, data centers, and wastewater treatment facilities to explore alternatives. LDES could fulfill such a role and would have the added potential to provide additional revenue by participating in other applications such as local ancillary services markets. Data centers, which can consume hundreds of MW of power, are indeed considering storage systems to provide both backup power and revenue from market participation,20 with one report showing that revenue from peak shaving and frequency regulation could reduce utility bills by as much as 12%.21 Economics and Operation of Long-Duration Electricity Storage Systems In recent years, efforts have centered on development of electricity storage systems with installed capital costs of 150 $/kWh and 5 h duration at rated power.22,23 The distinct characteristics of LDES applications entail a different storage technology cost structure, and result in requirements that offer both challenges and opportunities for technology developers. In the US, deregulated electricity markets typically provide compensation for both electricity generation and reliability services in the form of ancillary and capacity markets. It is technically feasible for storage systems to ‘‘stack’’ a range of services to maximize return on investment, resulting in duty cycles that superpose the contributions of several individual applications24 (although this may not be operationally possible due to market rules or contractual obligations). This breadth of value propositions, along with the ongoing policy-driven evolution of storage valuation in electricity markets (e.g., Federal Energy Regulatory Commission [FERC] Order 841 and the Electricity Market Design Directive in the European Union), prevents establishment of a simple or single set of technical and economic metrics that, if met, would result in wide-scale deployment of LDES technologies.25–27 Analysis of the potential role of LDES in all regions of the world should receive additional future attention.28–31 The aim of this analysis is to identify key economic and engineering tradeoffs for LDES systems with limited assumptions about use case and remuneration mechanisms. Economic frameworks developed for specific applications show similar results to what we present here.12,18 We explore the general LDES cost-performance parameter space using the discounted cash flow framework shown in Equation 1 (see the Supplemental Information for equation details). T X t =1
h i h i ð1 + rÞt DE;t nc;t + RP;t d 1 = CP d 1 + CE;th h1 D +
T X t =1
h ð1 + rÞt nc;t PC;t h1 RTE 1
i L=nc + nc;t VOMt + d 1 FOMt + CRe ð1 + rÞ (Equation 1)
4
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
The left side of Equation 1 represents revenue over the financial life of the project and the right side represents the total cost of ownership (TCO) of the system, both with units of $/kWh. The first revenue term on the left side of Equation 1 scales with the frequency and average price differential of charging and discharging modes (DE ) (representative of arbitrage applications). The second scales with power capacity (RP ) (representative of resource adequacy or capacity market remuneration). The TCO is broken into five principle components, with the first representing the installed capital cost. The discounted terms in the second set of brackets represent (in order) the expense related to efficiency losses, variable operating and maintenance cost, and fixed operating and maintenance cost that scales with the power capacity of the system. The final term explicitly accounts for replacement of the energy storage medium due to capacity fade (cycle life is considered here, but calendar fade may also be a significant factor for certain systems). Equating the present value of revenue and cost implies a zero net present value for a given project term and internal rate of return. Calculations throughout assume T = 20 years and r = 10%.32 We assume PC = 0.025 $/kWh in Figure 2, a representative value for the LCOE of future wind or solar generation in most locations in the US.33,34 This value does not include transmission or distribution costs and therefore reflects the price of electricity at the site of wind or solar generation. For simplicity, taxes, depreciation, inflation, and finance costs are not included. A single-factor sensitivity analysis that explores a broader range of values for the various revenue and cost parameters in Equation 1 is provided in the Supplemental Information (Figure S1) for a 50-h duration storage system. In addition, the Supplemental Information provides a discussion and a figure (Figure S3) that quantifies the impact of PC (including a value of 0 $/kWh, reflecting excess electricity on the grid) and round-trip efficiency (RTE) on the installed capital cost for a range of durations and values of RP . Of the parameters outlined in Equation 1, researchers pursuing new electricity storage technologies can most readily estimate the capital cost of a given system, even at relatively early stages of development. We therefore show installed capital costs that encompass both power and energy components (CP d 1 + CE;th h1 D ) in Figure 2. The z axis represents the available budget for capital expenditure as a function of hRTE and nc . We assume two revenue streams: (1) arbitrage electricity sales that scale with cycle frequency (DE ) and (2) payments for reliability services that scale primarily with power capacity (RP ). Consideration of both remuneration mechanisms is important given that, as an example, PSH units often spend over 50% of their operational hours in reserve shutdown mode, ready to supply power when needed.35 We assume an average price differential for charging and discharging modes of DE = 0.05 $/kWh-cycle, which would result in a compelling LDES technology under the cost structure of today’s grid. To illustrate, if half of the electricity produced by a wind or solar plant generated at 0.025 $/kWh passed through a co-located storage device with a cycle ‘‘premium’’ of 0.05 $/kWh-cycle (i.e., discharge price of 0.075 $/kWh-cycle), the average electricity price for the combined generator plus storage system would be 0.05 $/kWh, a price competitive with electricity generated by future combined cycle natural gas plants (which are in the range of 0.041 to 0.074 $/kWh).36 With 90% flowing through the storage asset, the combined LCOE would still be competitive at 0.07 $/kWh. Thus, storage systems operating in electricity markets with DE = 0.05 $/kWh-cycle across a range of durations that extend beyond daily cycling would enable a new class of combined variable renewable-plus-storage generation assets with dispatchability that much more closely matches flexible fossil generators. Regarding RP , as grids incorporate larger fractions of variable energy resources that contribute to low wholesale electricity prices but fail to provide dispatchability, capacity and ancillary service markets may have an outsized impact
Joule 4, 1–12, January 15, 2020
5
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 2. The Influence on Installed Capital Costs of Round-Trip Efficiency, Cycles per Year, Discharge Time, and Capacity Payments. Indexed colors show installed capital costs (CP d 1 + CE;th h1 D , in $/kWh) as a function of system round-trip energy efficiency (hRTE ) and number of cycles completed per year (nc ) for storage durations at rated power of (A) and (B) 10 h, (C) and (D) 50 h, and (E) and (F) 100 h and annual capacity payments of (A), (C), and (E) 25 $/kW-y and (B), (D), and (F) 75 $/kW-y. Values for other fixed parameters in Equation 1 are: VOM = 0.002 $/kWh-cycle and FOM = 1% of installed capital cost (representative of PSH 41 ), PC = 0.025 $/kWh-cycle, DE = 0.05 $/kWh-cycle, and no storage medium replacements. Note that the number of annual full system cycles cannot exceed 4,380 h/d, hence the different limits on the x axes. Figure S2 in the Supplemental Information presents results for DE = 0.10 $/kWh-cycle.
on the economic viability of storage projects,37 particularly for LDES systems that can provide nearly dispatchable output. We examine two different values of RP (25 and 75 $/kW-y) in Figure 2 to account for the uncertainty range associated with duration and capacity value that continues to be a topic of debate for system operators and market stakeholders, and provide additional parameter analysis in Figures S1 and S2 of the Supplemental Information.38–40 The upper right corners of Figures 2A and 2B reflect the capital cost targets (around 150 $/kWh) typical for daily cycling of efficient systems such as Li-ion batteries.22,23 As discussed in more detail in the context of Figure 3, it is important to note that these capital costs include the costs of both the power subsystem (Cp =d) and the energy subsystem (CE;th =hd ). While challenging, with an ability to leverage the major investments in Li-ion technology taking place for automotive applications, together with potential cost reductions in balance of plant components including HVAC and power electronics, such capital cost values may be approached in the coming decade.42 However, consideration of (C) and (D) for 50-h duration and (E) and (F) for 100-h duration (Figure 2), show that much lower costs are required than Li-ion is expected to achieve. For 50 h, installed capital costs of about 20 to 35 $/kWh
6
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 3. Relationship between Power and Energy Capital Costs Derived from Figure 2 for RP = 25 $/kW-y, Two hRTE , and Three Durations Values used for total capital cost were extracted from Figure 2 at 70% of the maximum annual complete cycles for a given duration. Figure S4 of in the Supplemental Information shows results for DE = 0.10 $/kWh-cycle, while Figures S5 and S6 show results for R p = 75 $/kWh-y and DE = 0.05 $/kWh-cycle and DE = 0.10 $/kWh-cycle, respectively. This figure also includes annotations for the approximate power and energy installed capital costs for underground chemical storage (shown to the left of the y axis because the value for CE;th is <1 $/kWh), 43 PSH, 7 Compressed air energy storage (CAES), 7 and future Li-ion. 44
are required, while for 100 h, installed capital costs of around 5 to 15 $/kWh are required. With DE = 0.05 $/kWh-cycle, the cost targets shown here for a 100-h duration are likely at or beyond the limit of what any manufactured and installed storage system may achieve. Also of key importance in Figure 2 is the RTE. While the sensitivity to RTE is higher with a capacity payment of 25 $/kW-y (Figures 2A, 2C, and 2E), the capital cost targets are extremely challenging if hRTE < 50%, and increasingly favorable with increasing RTE. It is also important to note that Figure 2 includes CE;th =hD for the capital cost of energy. In other words, the system efficiency has a 2-fold impact: lost revenue from purchased energy that is converted to noncompensated heat rather than electricity and (through hD ) a more challenging capital cost target for the energy subsystem. While Figure 2 provides the total installed capital cost for delivered energy (Cp =d + CE;th =hd ), Figure 3 splits apart the capital cost goals for the power and energy subsystems (assuming that the power block and energy blocks are separable), showing values for CP and CE;th . Three system durations are shown, along with two values of the RTE (as green and red lines). The results in Figure 3 emphasize the key importance and limitations of energy costs. Even excluding power subsystem costs, for durations of 50 or 100 h, energy capital costs cannot exceed 20 $/kWh; for 10 h duration the energy capital costs can be substantially higher. It is important to note that those expenses are inclusive of the energy storage medium, the container used to hold the energy storage medium, any additional balance of plant (BOP) costs, and installation. On the other hand, if such low energy capital costs can be achieved, capital costs for the power subsystem can be in the range of 500 to 1,000 $/kW, which are challenging, but not completely out of reach, in the context of present costs for turbines, flow battery stacks, and other candidate power conversion technologies. Technology Approaches for LDES Systems Figure 3 also includes annotations for several major classes of storage technology (power-to-gas, PSH, compressed air energy sequence [CAES], and Li-ion). Based on presently deployed technologies, PSH is a strong candidate for storage across a wide
Joule 4, 1–12, January 15, 2020
7
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
range of durations because of its low energy capital costs, adequate power capital cost (achieved in part through the use of reversible turbines), and high RTE. The cost structure of Li-ion positions it comparatively well for daily cycling applications, where the high capital costs for energy can be paid for by frequent cycling (and hence, frequent revenue collection with the stored energy). Another class of storage technology that is often discussed in the context of long-duration is power-to-gas (or other chemicals), and making use of either the existing pipeline capacity or underground reservoirs for storage.45 Such methods offer the exceptionally low storage costs required for long-duration storage (consistent, of course, with the fact that months of natural gas are stored in such structures, as indicated above), that is, <0.10 $/kWh.43 As one recent example, Air Liquide developed a hydrogen storage facility to provide 30 days of backup supply for steam methane reformers on the US Gulf of Mexico coast; this facility has a volume of 6 million m3 and likely holds 50 GWh of working hydrogen gas (on a primary energy basis).46 Unfortunately, when a chemical like H2 or CH4 is first synthesized (with electricity energy input) and then converted back to electricity, the RTE will likely be <40%, limiting the suitability for the LDES economics described in Figures 2 and 3. The use of solid-oxide fuel cells (SOFCs) may help overcome these low RTEs, as discussed below. Several references contain summary tables on major classes of storage technologies.7,47,48 The target area for LDES systems, which use containerized energy allowing flexible siting and with cost targets drawn from Figure 2, lies in the region spanning the plotted lines in Figure 3. Capital costs for energy may be substantially higher than approaches that use geologic features for storage (and hence have more limited siting), and the RTE should likely exceed 50% in order to provide sufficient revenue to cover capital costs. Given the critical importance of the energy capital cost, careful attention should be given to both the storage media capital costs as well as the costs of containerization (including the container in contact with the energy storage media, any secondary or tertiary containment, as well as the shipping container or building that may be used to help with thermal conditioning), shipping, and installation. Figure S7 shows typical storage media and container capital costs, which constitute a lower bound on storage costs. As shown in Figure 3, acceptable installed energy capital costs range from just a few $/kWh (for 100-h duration, 50% RTE) to around 75 $/kWh (for 10-h duration, 80% RTE). To pick one value as an exemplar, for a 50-h duration and 80% RTE, to ensure that the costs of containment as well as shipping, site preparation, and installation are consistent with an energy subsystem cost targets of <20 $/kWh, the energy density of all storage media should preferably be R0.1 kWh/L. Several technology classes are under development that may approach the cost and efficiency targets depicted in Figure 3. Here, we focus on relatively new approaches that may be particularly well suited for 10- to 100-h duration that is a focus in this paper. In the area of thermal systems, latent, sensible, and thermochemical mechanisms are all under development. A key requirement for achieving a system RTE of R50% is Tcold =Thot < 0:2, given Carnot constraints on thermal storage. This can be achieved either by increasing Thot , reducing Tcold (which is often the ambient, but need not be), or both. For example, pumped heat thermal storage uses both cold and hot reservoirs for storage,49,50 and several system development efforts are underway.51 Another option is to substantially increase Thot , either to the upper temperature limits for gas turbines52–54 or even substantially beyond through the use of non-turbine heat-to-electricity conversion.55 High temperatures (>1,000 C) may be achieved with resistive heaters, which can offer extremely low capital costs on a $/kW basis depending on the upper temperature.56 One challenge for thermal systems is scaling; to minimize thermal losses (or insulation costs), large containment vessels are needed, and when turbines are used, that also contributes to system
8
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
sizing constraints, with the most favorable economics in the tens to hundreds of MW. Electrochemical systems are another technology class in which LDES systems are being developed. From the point of view of active materials alone, numerous battery chemistries have a capital cost that could meet the LDES targets outlined above.57–60 Indeed, the principal grid storage battery prior to Li-ion was the Na/S system, for which active material costs are <10 $/kWh. While starting with lowcost reactants is required, we also emphasize the need for a sufficiently high energy density (e.g., for a duration of 50 h the energy density should be R0.1 kWh/L, including all active materials) to limit container and installation costs. Another important class of electrochemical technologies for LDES is reversible SOFCs that have a clear technical capability to achieve high RTE when cycling between CO2 and/or H2O (on discharge) and CH4 and/or H2 (on charge).61,62 Mechanical systems are a third technology category under development with potential to address LDES markets. Adiabatic compressed air storage with integrated thermal storage offers substantially better RTE than conventional natural-gas fired systems.63 While the raw materials for mechanical storage systems generally range from low cost to free,64 the issue has been overcoming low energy and power densities that necessitate large earthworks or engineering projects to achieve appreciable capacities. New approaches that utilize existing or readily generated subterranean structures may help overcome this longstanding challenge.65 Beyond careful selection of low-cost energy storage media, containment, and other system attributes, the unique requirements of LDES—in particular the fact that the system may rarely undergo a full cycle of its stored energy—may offer opportunities for fundamentally different and innovative approaches and system configurations. As one example, Figure 4 provides a graphical depiction of an approach in which, rather than a ‘‘single tank’’ for the energy storage media, storage media with different attributes—which may include energy density, material identity, or purity, amount of thermal insulation, pressure, temperature, etc.—could be accessed in a staged and sequential manner. This staging could allow optimization for cost, efficiency, cycle life, and other metrics that result in superior overall system cost and performance. As one concrete example, in a Zn/Br2 flow battery when the Br2 exceeds the solubility limit in the ZnBr2 aqueous solution, it will precipitate into a separate phase with a higher density, providing a natural means of separation and storage.66 The energy density of 2.5 M ZnBr2 (aq) is 170 Wh/L (assuming 80% can react), while the energy density of liquid Br2 and solid Zn exceeds 1,400 Wh/L. If a Zn/Br2 system could be designed with a small volume of ZnBr2 (aq) compared to solid Zn and liquid Br2, the cost of the container alone could potentially be reduced by >5 $/kWh (with other costs such as site preparation and shipping also lowered), substantially improving the economics for LDES. Another innovative approach to the challenging economic targets of LDES is to add an energy storage subsystem on to an existing power conversion unit.67 For example, a thermal storage unit could be added to an existing turbine, with heat added to the thermal store using resistive heating or another method, and the existing turbine powered by either its intended source of thermal energy (e.g., natural gas), thermal energy from the store, or a combination of the two. In this approach, the power costs may consist only of the additional BOP required to connect the thermal store with the existing turbine, as well as the resistive heaters (for charging power), providing more room for energy costs while still meeting the overall LDES cost targets. These approaches are just two examples of the types of fundamentally different approaches to designing energy storage systems that are uniquely suited to the new technoeconomic design space of LDES applications.
Joule 4, 1–12, January 15, 2020
9
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
Figure 4. A Storage System with a Single Power Block and Storage Media with Attributes that Depend on Cycling Frequency Conceptual schematic of a fundamentally different system architecture in which the varied cycling frequencies associated with LDES applications can be leveraged to create a series of energy storage tanks with attributes optimized to reduce capital costs, rather than using a single, uniform, tank across all durations.
Outlook and Perspective Long-duration electricity storage systems could be one important route to make use of wind and solar and achieve zero-carbon electricity goals as well as serve other applications like backup power. In this work, we focused on durations between 10 and 100 h, with the lower limit at the upper end of daily cycling, and the upper end at or beyond the likely economic limits of any current containerized system that can be flexibly sited. Our economic analysis, and those of others, have found that achieving low installed energy capital costs (between 5 and 35 $/kWh) is of critical importance, which in turn will require low-cost and high-energy-density reactants. In addition, for the values assumed in our economic analysis, an RTE of >50% is likely required to provide enough revenue to finance the up-front capital costs. While approaches like power-to-gas can achieve extremely low (<0.10 $/kWh) capital costs when using underground storage, the RTE of these approaches is usually <40%. However, longer-duration systems that can also serve shorter duration applications (e.g., daily cycling) can potentially provide greater value to systems with durations of >100 h, even if the RTE is <50%. As solar and wind capacity continue to be built, and economic incentives and policies put in place to achieve 100% carbon-free grid goals, LDES systems may offer an important option to advance low-cost and carbon-free electricity in markets around the world. However, a challenge for any new stationary electricity storage system is the low-margin, risk-averse nature of the customers in the utility industry. Without a substantial (ideally, tens of billions of dollars cumulatively) and high-margin set of first markets (which Li-ion had in the form of portable electronics), it will be difficult to achieve the manufacturing volume and project quantity to consistently drive down costs through scale and learning. As such, it is likely that LDES systems will require long-term vision, development, and investment strategies in both the public and private sectors if these potentially game-changing technologies are to become a cornerstone of a future grid that is low cost, low carbon, and resilient.
10
Joule 4, 1–12, January 15, 2020
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j.joule. 2019.11.009.
ACKNOWLEDGMENTS Paul Albertus and Joseph S. Manser acknowledge the Advanced Research Projects Agency–Energy, of the United States Department of Energy, where they were employed when a portion of this manuscript was conceived and written. This manuscript describes, in part, the motivation and framing of the ARPA-E DAYS program.4
REFERENCES 1. U.S. Energy Information Administration (2019). Levelized cost and Levelized avoided cost of new generation resources in the annual energy Outlook 2019. https://www.eia.gov/outlooks/ aeo/pdf/electricity_generation.pdf. 2. Clean Air Task Force (2019). Fact Sheet: state and utility climate change targets shift to carbon reductions, technology diversity. https://www.catf.us/wp-content/uploads/ 2019/05/State-and-Utility-Climate-ChangeTargets.pdf. 3. United States Department of Energy Washington, DC (2018). Energy resilience solutions for the Puerto Rico grid. https://www. energy.gov/sites/prod/files/2018/06/f53/ DOE%20Report_Energy%20Resilience% 20Solutions%20for%20the%20PR%20Grid% 20Final%20June%202018.pdf. 4. The Advanced Research Projects AgencyEnergy (ARPA-E) (2018). Duration addition to electricity storage (DAYS) overview. https:// arpa-e.energy.gov/sites/default/files/ documents/files/DAYS_Program Overview_FINAL.pdf. 5. U.S. Energy Information Administration (2017). Energy storage and renewables beyond wind, hydro, solar make up 4% of U.S. power capacity, Today in Energy. https://www.eia. gov/todayinenergy/detail.php?id=31372. 6. National Grid. (2017). National Grid develops innovative solution for an island community’s unique energy challenges. https://news. nationalgridus.com/2017/11/national-griddevelops-innovative-solution-islandcommunitys-unique-energy-challenges/. 7. Akhil, A.A., Huff, G., Currier, A.B., Kaun, B.C., Rastler, D.M., Chen, S.B., Cotter, A.L., Bradshaw, D.T., and Gauntlett, W.D. (2013). DOE/EPRI 2013 Electricity Storage Handbook in Collaboration with NRECA (Sandia National Laboratories). 8. Denholm, P., and Margolis, R. (2016). Energy storage requirements for achieving 50% solar photovoltaic energy penetration in California (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/docs/fy16osti/ 66595.pdf. 9. Denholm, P., and Hand, M. (2011). Grid flexibility and storage required to achieve very high penetration of variable renewable electricity. Energy Policy 39, 1817–1830. 10. Shaner, M.R., Davis, S.J., Lewis, N.S., and Caldeira, K. (2018). Geophysical constraints on
the reliability of solar and wind power in the United States. Energy Environ. Sci. 11, 914–925. 11. Denholm, P., and Mai, T. (2019). Timescales of energy storage needed for reducing renewable energy curtailment. Renew. Energy 130, 388–399. 12. Ziegler, M.S., Mueller, J.M., Pereira, G.D., Song, J., Ferrara, M., Chiang, Y., and Trancik, J.E. (2019). Storage requirements and costs of shaping renewable energy Toward grid decarbonization. Joule 3, 2134–2153. 13. Jones, R. (2017). System perspectives on long duration storage, ARPA-E, Evolved Energy Research. https://docs.wixstatic.com/ugd/ a35761_773bbc576d8b4b02ae2c02ad93c6166f. pdf. 14. De Sisternes, F.J., Jenkins, J.D., and Botterud, A. (2016). The value of energy storage in decarbonizing the electricity sector. Appl. Energy 175, 368–379. 15. Clack, C.T.M., Qvist, S.A., Apt, J., Bazilian, M., Brandt, A.R., Caldeira, K., Davis, S.J., Diakov, V., Handschy, M.A., Hines, P.D.H., et al. (2017). Evaluation of a proposal for reliable low-cost grid power with 100% wind, water, and solar. Proc. Natl. Acad. Sci. USA 114, 6722–6727. 16. U.S. Energy Information Administration (2019). Weekly natural gas storage report. http://ir.eia. gov/ngs/ngs.html. 17. Merchant, E.F. (2016). NV Energy Announces ‘Hulkingly Big’ solar-plus-storage procurement (Green Tech Media), June 25, 2019. https:// www.greentechmedia.com/articles/read/nvenergy-signs-a-whopping-1-2-gigawatts-ofsolar-and-590-megawatts-of-stor#gs.w3wb9i. 18. Braff, W.A., Mueller, J.M., and Trancik, J.E. (2016). Value of storage technologies for wind and solar energy. Nat. Clim. Change 6, 964–969. 19. Laws, N.D., Anderson, K., DiOrio, N.A., Li, X., and McLaren, J. (2018). Impacts of valuing resilience on cost-optimal PV and storage systems for commercial buildings. Renew. Energy 127, 896–909. 20. Storrow, B. (2017). How big batteries at data centers could replace power plants, E&E News, July 19, 2018. https://www.scientificamerican. com/article/how-big-batteries-at-datacenters-could-replace-power-plants/. 21. Shi, Y., Xu, B., Wang, D., and Zhang, B. (2017). Using battery storage for peak shaving and frequency regulation: joint optimization for
superlinear gains. IEEE Trans. Power Syst. 33, 2882–2894. 22. The Advanced Research Projects AgencyEnergy (2018). Integration and optimization of Novel ion conducting solids (IONICS) program overview. https://arpa-e.energy.gov/sites/ default/files/documents/files/IONICS_ ProgramOverview.pdf. 23. Darling, R.M., Gallagher, K.G., Kowalski, J.A., Ha, S., and Brushett, F.R. (2014). Pathways to low-cost electrochemical energy storage: a comparison of aqueous and nonaqueous flow batteries. Energy Environ. Sci. 7, 3459–3477. 24. Davies, D.M., Verde, M.G., Mnyshenko, O., Chen, Y.R., Rajeev, R., Meng, Y.S., and Elliott, G. (2019). Combined economic and technological evaluation of battery energy storage for grid applications. Nat. Energy 4, 42–50. 25. Balducci, P.J., Alam, M.J.E., Hardy, T.D., and Wu, D. (2018). Assigning value to energy storage systems at multiple points in an electrical grid. Energy Environ. Sci. 11, 1926– 1944. 26. Katsanevakis, M., Stewart, R.A., and Lu, J. (2017). Aggregated applications and benefits of energy storage systems with applicationspecific control methods: a review. Renew. Sustain. Energy Rev. 75, 719–741. 27. Oh, U., Choi, J., Kim, K., Cha, J., and Lee, K.Y. (2018). A study on capacity credit and ESS evaluation for WTG and multi-ESS in power system. IFAC-PapersOnLine 51, 516–521. 28. Jentsch, M., Trost, T., and Sterner, M. (2014). Optimal use of power-to-gas energy storage systems in an 85% renewable energy scenario. Energy Procedia 46, 254–261. 29. Barbour, E., Wilson, I.A.G., Radcliffe, J., Ding, Y., and Li, Y. (2016). A review of pumped hydro energy storage development in significant international electricity markets. Renew. Sustain. Energy Rev. 61, 421–432. 30. Zafirakis, D., Chalvatzis, K.J., Baiocchi, G., and Daskalakis, G. (2016). The value of arbitrage for energy storage: evidence from European electricity markets. Appl. Energy 184, 971–986. 31. Blakers, A., Lu, B., and Stocks, M. (2017). 100% renewable electricity in Australia. Energy 133, 471–482. 32. McLaren, J., Anderson, K., Laws, N., Gagnon, P., DiOrio, N., and Li, X. (2018). Identifying critical factors in the cost-effectiveness of solar
Joule 4, 1–12, January 15, 2020
11
Please cite this article in press as: Albertus et al., Long-Duration Electricity Storage Applications, Economics, and Technologies, Joule (2019), https://doi.org/10.1016/j.joule.2019.11.009
and battery storage in commercial buildings (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/docs/fy18osti/ 70813.pdf. 33. US Department of Energy’s Solar Energy Technologies Office (2017). The SunShot 2030 Goals: 3¢ per kilowatt hour for PV and 5¢ per killowatt hour for dispatchable CSP. https:// www.energy.gov/sites/prod/files/2018/05/f51/ SunShot%202030%20Fact%20Sheet.pdf. 34. U.S. Department of Energy. WindVision: a new era for wind power in the United States. https:// www.energy.gov/sites/prod/files/2015/03/f20/ wv_full_report.pdf. 35. Urı´a-Martı´nez, R., Johnson, M.M., O’Connor, P.W., Samu, N.M., Witt, A.M., Battey, H., Welch, T., Bonnet, M., and Wagoner, S. (2018). 2017 Hydropower market report (U.S. department of energy). https://www.energy. gov/sites/prod/files/2018/04/f51/Hydropower %20Market%20Report.pdf. 36. Lazard. (2018). Lazard’s Levelized Cost of Energy Analysis–12.0 version. https://www. lazard.com/media/450784/lazards-levelizedcost-of-energy-version-120-vfinal.pdf. 37. Jenkin, T., Beiter, P., and Margolis, R. (2016). Capacity payments in restructured markets under low and high penetration levels of renewable energy (National Renewable Energy Laboratory (NREL)). https://www.nrel.gov/ docs/fy16osti/65491.pdf. 38. Carden, K. (2019). Valuing capacity for resources with energy limitations – independent assessment, Astrape consulting. https://www.nyiso.com/documents/20142/ 5020603/Astrape%20presentation%20021519. pdf. 39. Hall, W., Zhang, B., and Legnard, T. (2019). Valuing capacity for resources with energy limitations, GE Energy Consulting. https:// www.nyiso.com/documents/20142/4358080/ 01082019%20Capacity%20Value%20of% 20Resources%20with%20Energy% 20Limitations_v2.pdf/3499da16-12d8-16b7b12f-be7650e64b63. 40. Pfeifenberger, J. (2019). Capacity markets and wholesale market outcomes. https:// brattlefiles.blob.core.windows.net/files/ 16366_capacity_markets_and_wholesale_ market_outcomes.pdf. 41. The International renewable energy agency (2012). Renewable energy technologies: cost analysis series-Hydropower. https://www.irena. org/documentdownloads/publications/ re_technologies_cost_analysis-hydropower. pdf. 42. BloombergNEF (2019). New energy outlook 2019. https://about.bnef.com/new-energyoutlook/. 43. Federal Energy Regulatory Commission (2004). Current state of and issues concerning underground natural gas storage. https://www.
12
Joule 4, 1–12, January 15, 2020
ferc.gov/EventCalendar/Files/200410 20081349-final-gs-report.pdf. 44. Nykvist, B., and Nilsson, M. (2015). Rapidly falling costs of battery packs for electric vehicles. Nat. Clim. Change 5, 329–332. 45. Go¨tz, M., Lefebvre, J., Mo¨rs, F., McDaniel Koch, A., Graf, F., Bajohr, S., Reimert, R., and Kolb, T. (2016). Renewable Power-to-Gas: A technological and economic review. Renew. Energy 85, 1371–1390. 46. Air liquide worldwide (2017). USA: air liquide operates the world’s largest hydrogen storage facility. https://en.media.airliquide.com/news/ usa-air-liquide-operates-the-worlds-largesthydrogen-storage-facility-edf8-56033.html. 47. Abdon, A., Zhang, X., Parra, D., Patel, M.K., Bauer, C., and Worlitschek, J. (2017). Technoeconomic and environmental assessment of stationary electricity storage technologies for different time scales. Energy 139, 1173–1187. 48. Divya, K.C., and Østergaard, J. (2009). Battery energy storage technology for power systems—an overview. Electr. Power Syst. Res. 79, 511–520. 49. Laughlin, R.B. (2017). Pumped thermal grid storage with heat exchange. J. Renew. Sustain. Energy 9, 044103.
55. Amy, C., Seyf, H.R., Steiner, M.A., Friedman, D.J., and Henry, A. (2019). Thermal energy grid storage using multi-junction photovoltaics. Energy Environ. Sci. 12, 334–343. 56. Forsberg, C.W., Stack, D.C., Curtis, D., Haratyk, G., and Sepulveda, N.A. (2017). Converting excess low-price electricity into hightemperature stored heat for industry and highvalue electricity production. Electr. J. 30, 42–52. 57. Wadia, C., Albertus, P., and Srinivasan, V. (2011). Resource constraints on the battery energy storage potential for grid and transportation applications. J. Power Sources 196, 1593–1598. 58. Li, Z., Pan, M.S., Su, L., Tsai, P., Badel, A.F., Valle, J.M., Eiler, S.L., Xiang, K., Brushett, F.R., and Chiang, Y. (2017). Air-breathing aqueous sulfur flow battery for ultralow-cost longduration electrical storage. Joule 1, 306–327. 59. Soloveichik, G.L. (2015). Flow batteries: current status and trends. Chem. Rev. 115, 11533– 11558. 60. Hawthorne, K.L., Petek, T.J., Miller, M.A., Wainright, J.S., and Savinell, R.F. (2015). An investigation into factors affecting the iron plating reaction for an all-iron flow battery. J. Electrochem. Soc. 162, A108–A113.
50. White, A., Parks, G., and Markides, C.N. (2013). Thermodynamic analysis of pumped thermal electricity storage. Appl. Therm. Eng. 53, 291–298.
61. Jensen, S.H., Graves, C., Mogensen, M., Wendel, C., Braun, R., Hughes, G., Gao, Z., and Barnett, S.A. (2015). Large-scale electricity storage utilizing reversible solid oxide cells combined with underground storage of CO2 and CH4. Energy Environ. Sci. 8, 2471–2479.
51. Hanley, S. Google X spins off Malta molten salt energy storage business. https:// cleantechnica.com/2018/12/21/google-xspins-off-malta-molten-salt-energy-storagebusiness/.
62. Wendel, C.H., and Braun, R.J. (2016). Design and techno-economic analysis of high efficiency reversible solid oxide cell systems for distributed energy storage. Appl. Energy 172, 118–131.
52. Randhir, K., King, K., Rhodes, N., Li, L., Hahn, D., Mei, R., AuYeung, N., and Klausner, J. (2019). Magnesium-manganese oxides for high temperature thermochemical energy storage. J. Energy Storage 21, 599–610.
63. Bullough, C., Gatzen, C., Jakiel, C., Koller, M., Nowi, A., and Zunft, S. (2004). Advanced adiabatic compressed air energy storage for the integration of wind energy. In Proceedings of the European Wind Energy Conference, EWEC, pp. 22–25.
53. Ma, Z., Mehos, M., Glatzmaier, G., and Sakadjian, B.B. (2015). Development of a concentrating solar power system using fluidized-bed technology for thermal energy conversion and solid particles for thermal energy storage. Energy Procedia 69, 1349– 1359. 54. Ma, Z., Zhang, R., and Sawaged, F. (2017). Design of Particle-Based Thermal Energy Storage for a Concentrating Solar Power System. In the Proceedings of the ASME 2017 11th international conference on energy sustainability collocated with the ASME 2017 power conference joint with ICOPE-17, the ASME 2017 15th international conference on fuel cell science, engineering and technology, and the ASME 2017 nuclear forum. ASME 2017 11th International conference on energy sustainability. (American Society of Mechanical Engineers).
64. Fyke, A. (2019). The fall and rise of Gravity storage technologies. Joule 3, 625–630. 65. Moodie, A. (2017). A Texas startup’s big energy idea: storing electricity underground, The Guardian. https://www.theguardian.com/ sustainable-business/2016/aug/25/electricitystorage-underground-quidnet-energy. 66. Biswas, S., Senju, A., Mohr, R., Hodson, T., Karthikeyan, N., Knehr, K.W., Hsieh, A.G., Yang, X., Koel, B.E., and Steingart, D.A. (2017). Minimal architecture zinc–bromine battery for low cost electrochemical energy storage. Energy Environ. Sci. 10, 114–120. 67. Forsberg, C., and Peterson, P.F. (2016). Basis for fluoride salt–cooled high-temperature reactors with nuclear air-Brayton combined cycles and firebrick resistance-heated energy storage. Nucl. Technol. 196, 13–33.