Look Before You Leap. The Role of Energy Storage in the Grid. By Devon Manz, Richard Piwko, and Nicholas Miller

Transcription

1 Look Before You Leap The Role of Energy Storage in the Grid By Devon Manz, Richard Piwko, and Nicholas Miller DYNAMIC GRAPHICS A FUNDAMENTAL TRUTH OF THE GRID HAS been that electricity must be generated at the precise moment it is demanded. It is the ultimate just in time system, where the laws of physics prevent carrying inventory. This characterization is under challenge, as the development of large-scale energy storage technologies is accelerating. A growing group of engineers, grid operators, and regulatory agencies believe that energy storage will be a critical component of the grid of the future. Over the past several decades, large-scale hydro and pumped hydro storage facilities dominated the energy storage landscape. Today, new and evolving battery chemistries, primarily for electric vehicle and backup power applications, are emerging as potential solutions for some of the challenges that face the grid today. Both batteries and high-speed mechanical flywheels connected to the grid through power electronics are enabling smaller and more modular energy storage systems. These storage systems are being considered for a variety of applications, from time-shifting wind Digital Object Identifier /MPE Date of publication: 18 June 2012 july/august /12/$ IEEE IEEE power & energy magazine 75

2 The authors have not yet identified a single power system operation or performance issue for which storage is the only solution. or solar energy to deferral of new substation infrastructure to short-time-scale frequency regulation. The modular nature of these systems and the relatively small investment needed to defer what could be large expenditures offer effective technical alternatives to conventional solutions. The primary challenges that face the widespread deployment of energy storage are its relatively high capital cost, at least in the early deployments; the fact that there are relatively few installations in place with which to build confidence in the technologies; and, in the case of battery energy storage, cycle life limitations. Storage for Renewables: Solution or Bandwagon? The authors and their organization have performed wind integration studies for a variety of regions across North America, including New York, Texas, California, New England, the island of Oahu, and large portions of the western United States (see For Further Reading ). The results of these studies showed that wind power contribution levels approaching 30% by energy can be accommodated in large interconnected power grids. Integrating such high levels of wind power requires significant changes to power system operation and increases the need for operating reserves, generation flexibility (faster ramp rates, faster starts, lower turn-down), wind power forecasting, and balancing area cooperation. Energy storage could play a role as well, but the performance benefits and costs of energy storage must be compared with these more conventional, incumbent approaches. Other solutions to enable wind energy integration are also emerging. Demand-response programs are providing a growing number of ancillary services (e.g., operating reserves). Storage will have to compete against these technologies. In most applications, economics will likely be the deciding factor, unless policy or regulatory measures that favor energy storage significantly influence the value proposition so as to rank energy storage above competing alternatives. Where Does Energy Storage Really Fit? Power grids face a variety of challenges that could be totally or partially mitigated by installation of energy storage technologies. These include: widespread deployment of renewable energy, primarily wind and solar power, and the associated uncertainty and variability of the wind and the sun transmission congestion that periodically requires less economic operation, such as spilling wind power due to limitations in transmission paths increased need for frequency regulation due to variability in load and renewable energy production rapid growth of loads on feeders, requiring major equipment upgrades. Although energy storage can effectively mitigate these types of challenges, storage is not the only possible solution. In fact, the authors have not yet identified a single power system operation or performance issue for which storage is the only solution. There are many alternative approaches for solving any power system problem. The application of energy storage to solving a given problem must be evaluated in comparison with other methods or technologies on the basis of technical performance, environmental impact, reliability, and ultimate cost to energy consumers. This article is dedicated to that concept. Storage has a role to play in the power system. In this article the authors will consider five applications for energy storage. In each case study, energy storage will be compared against alternate conventional approaches. The Power-Energy Continuum The applications of energy storage can roughly be placed on a continuum of power and energy. Generally, energy applications can be defined as those that rely on a continuous delivery of energy over a considerable length of time, typically hours in duration. The steady flow of energy is more important for energy applications on the grid than is the sudden and rapid injection or absorption of energy under changing grid conditions. Energy applications may include peak shifting, economic arbitrage (i.e., capturing the difference between on- and off-peak pricing), and operationally necessary storage of energy generated at off-peak times (e.g., to avoid wind curtailment). Power applications, on the other hand, are those that typically require the rapid injection and absorption of energy over shorter durations. Power applications include regulating system frequency; providing ramp rate control for variable generation, such as wind and solar; and providing inertial/droop frequency response characteristics, to name a few. Different energy storage technologies have varying degrees of efficacy in different applications. Like applications for storage, technologies for storage can be viewed as either primarily power technologies or primarily energy technologies. Storage technologies well suited to power applications, rather than energy applications, are those that possess a rated 76 IEEE power & energy magazine july/august 2012

3 power-to-energy ratio (MW:MWh rating) of about 2:1 or greater. These technologies can deliver or absorb significant energy over short time frames and include technologies such as flywheels, some types of batteries (various lithium chemistries, for example), and ultracapacitors. Storage technologies well suited to energy applications, on the other hand, are typically those that have the capacity to store many more MWh than they have the power rating to deliver in a short time frame and have a rated power-to-energy ratio of about 1:2 or smaller. Storage technologies well suited to energy applications are flow batteries (e.g., vanadium and zinc bromine) and certain more common battery chemistries (e.g., valve-regulated lead acid and sodium). There is, of course, a continuum of power-toenergy ratios, as shown in Figure 1. In addition to this ratio, there is the issue of overall power rating. Two large-scale energy storage technologies, compressed air energy storage (CAES) and pumped hydro storage, generally have ratings of hundreds of MW and storage capacities of hundreds or thousands of MWh. Because of their scale, they can provide significant levels of both power and energy. Even though they may typically have relatively low power-to-energy ratios (in the range of 1:4 or 1:8), they can still provide large amounts of power and can be highly maneuverable. Each storage technology can serve a range of power and energy applications. Economic and physical limitations will dictate the power and energy range for each technology. As an example, economics might suggest that building a pumped hydro facility for only one hour of storage may not make sense, because the marginal cost of more energy capacity might be relatively low (larger pondage). As another example, the laws of physics may limit the amount of energy a single flywheel device can provide: material strength will limit the rotational speed of a flywheel module. Physics may also limit some battery chemistries from injecting or removing large amounts of energy over very short time frames. Such economic and physical limitations suggest that each storage technology has a sweet spot and the technology should be appropriately selected for the power and energy needs of a given application. Power Applications Applications of Energy Storage The power-energy continuum analysis showed that some applications demand a small amount of energy at high power while other applications demand a large amount of energy at a somewhat lower power. Beyond the ratings (in MW and MWh) of a storage system, some applications demand that energy be rapidly injected or absorbed while others demand Power (MW) Power : Energy 4:1 3:1 2:1 1:1 1:2 1:3 1:4 4:1 Time (h) MW : MWh Power (MW) Time (h) Energy Applications that energy be shifted from one time to another. Table 1 summarizes a range of possible applications for energy storage and lists the reasons why storage may be considered for a specific application. The value of energy storage is the measure of the benefit achieved for a given application. The value of energy storage, in turn, can be compared with the value of other approaches for serving the same application. The upper section of Table 1 shows applications for energy storage that in themselves could justify the investment in a storage facility. The lower section of the table shows other functions or ancillary services that a storage facility could provide as fringe benefits but that would not in themselves be likely to justify the full investment. Functions such as voltage regulation and harmonic suppression are made possible by the power electronics required to interface the storage media to the grid. For these functions, it is the electronics that provide the benefit and not the storage medium itself. In some instances, it is possible for storage systems to perform multiple functions. Two or more functions could be performed simultaneously or sequentially. A storage system must generate enough revenue to cover its costs and remain profitable. The costs include the capital investment to purchase and build the system, the replacement cost of components with life expectancies shorter than that of the facility, and the cost of the energy that must be purchased to cover the round-trip efficiency losses of the system. In all these applications, the key metrics that determine the cost of any storage asset are the upfront capital cost (purchase price), the round-trip efficiency (net energy consumed), and the replacement time (life of the storage medium). While costs for any individual technology can vary, the range of these costs is reasonably well understood. Round-trip efficiencies can be similarly estimated within a reasonable range. Of these three elements, the largest variable, however, is the life of the asset. For some of the storage technologies particularly some emerging battery chemistries the life of the battery cells has not been proven over a wide range of duty cycles over multiple years. Further, in Power (MW) 1:4 Time (h) figure 1. The continuum of energy storage applications and technologies. 1:1 july/august 2012 IEEE power & energy magazine 77

4 table 1. Applications of energy storage. Application Description Value Type Financial Energy Arbitrage Buy low, sell high Displaces most expensive generation Energy Generation Capacity Contribute to adequacy/reserve margin requirement Defers investment in new generation Energy Equipment Capacity Reduce flow through overloaded lines and transformers Defers investment in new equipment Energy Applications for Energy Storage Line Congestion Wind and Solar Power Smoothing Time shift delivery of renewable energy during congestion Reduce ramp rates of wind and solar plants Delays transmission line reinforcement Contributes to reserve and regulation requirements Energy Power Frequency Regulation Rapidly inject and remove power for short intervals Contributes to regulation requirements Power Spin and Non-Spin Reserve Dispatch power in <10 min Contributes to system reserves Power Governor/Inertial Response Provide dynamic functional equivalents of synchronous generators Reduces severity of frequency excursions events Power Ancillary Applications for Energy Storage Power Quality/ Harmonics Black-start Voltage Regulation Suppress system harmonics Support system during system restoration Manage delivery of reactive power to maintain voltage Contributes to power quality Contributes to system black-start capability Reduces need for new reactive power sources Both Both Both Capital Cost (US$/kW) 3,000 2,500 2,000 Rate of Return 0% 20% 100% Efficient 1,500 80% Efficient 1, Peak/Off-Peak Price Spread (US$/MWh) figure 2. Financial arbitrage value model for a storage facility with a 50-MW, 400-MWh energy storage system at 80% and 100% round-trip system efficiency. some applications, the duty cycle is not well known because new ancillary services markets are still evolving or the rules in mature markets are being changed to accommodate energy storage technologies. The early deployments of storage will provide a much-needed learning opportunity in order to optimize the design and operation of storage to maximize life expectancy. One of the successful energy storage deployments took place in 1996 in Metlakatla, Alaska [see Battery Energy Storage System (BESS) for Frequency Regulation in Metlakatla, Alaska ]. In the following sections, several of the energy storage applications in Table 1 are discussed in more detail, and frameworks for assessing the value of each are provided. Financial Energy Arbitrage Financial energy arbitrage is defined here as purchasing and storing energy ( charging ) when energy prices are low, then selling energy ( discharging ) when energy prices are high. The most expensive power generation resources are dispatched when the load is the greatest. A storage facility installed for energy arbitrage attempts to displace expensive generating resources during peak load periods with less expensive energy captured during light load conditions. A few key factors establish the viability of storage for this application: the round-trip efficiency of the system the capital cost of the system life expectancy under the prescribed cycling duty the average difference between peak and off-peak energy prices the duration and timing of the spreads in prices the accuracy of price spread forecasting. Historical wholesale prices in the energy markets are widely available, and by making some basic assumptions about 78 IEEE power & energy magazine july/august 2012

5 Battery Energy Storage System (BESS) for Frequency Regulation in Metlakatla, Alaska The value proposition for energy storage providing frequency regulation is enhanced when the costs of alternate means for frequency regulation are relatively high. In 1995, General Electric (GE) partnered with GNB Technologies to design and build a 1.2-MW, 1.2-MWh (a 1:1 power-energy ratio) BESS for Metlakatla Power and Light (MP&L), in Alaska (Figures S1 and S2). The BESS was deployed on the island of Metlakatla to help manage voltage and frequency fluctuations caused by load variations in the 1,000-resident community. Prior to the installation of the BESS, small reciprocating diesel units provided a significant portion of the island s frequency regulation. The diesel units were fueled by imported diesel fuel that arrived on barges. As part of this project, GE provided a power conversion system, battery management system, user interface, and the complete integration with the utility control system, as well as the engineering, procurement, and project construction. The BESS, at a cost of about US$2.2 million in 1997, saved MP&L approximately US$6.6 million in fuel costs over the following 11 years of operation, at which time the batteries were replaced. The battery life exceeded the eight-year life expectancy and nearly eliminated the use of fossil fuel at MP&L (see For Further Reading ). The value of energy storage is higher in regions with high fuel costs and limited resources capable of providing frequency regulation. figure S1. Nicholas Miller and the battery energy storage system in Metlakatla, Alaska, constructed in 1995 through (Source: George Hunt.) figure S2. Substation interconnection for the battery energy storage system in Metlakatla, Alaska. (Source: George Hunt.) the technical characteristic of a storage system and a corresponding bidding strategy, the economic case for financial arbitrage can be established for a given market. An example is a 50-MW, 400-MWh energy storage system (a1:8 power-energy ratio) that was deployed to capture lower-cost US$20/MWh off-peak energy for eight consecutive hours and displace higher-cost peak energy for 6.4 hours (only 80% of the energy captured returns to the grid due to roundtrip efficiency losses). This was performed on every day of the year. To achieve rates of return between 0% and 20%, the capital cost and price spread must fall within the red triangular area shown in Figure 2. In this example, if the capital cost is US$1,500/kW, an average daily price spread of US$50/MWh must persist for eight hours of off-peak charging to achieve positive rates of return. This example also assumes that accurate energy price forecasts exist so that the storage facility owner can capture this spread every day. In other words, it was assumed that the eight hours of charging, for which the off-peak price sustained itself, was known in advance, such that the storage system was capable of charging during those hours. It is obvious that large, persistent spreads between peak and off-peak energy prices are critical to the economics of arbitrage. In many systems, especially those with large amounts of wind power, one could reasonably expect significant spreads between peak and off-peak prices for at least a few hours each day. But the ultimate economic viability of an energy storage facility for an arbitrage application depends on the frequency and duration of time periods with large price spreads. When quantifying the cost of energy storage, care should be taken to distinguish between the cost of the energy storage technology itself (i.e., the batteries or the flywheels) and the full capital cost for the entire facility, including the balance of plant. Both are normally expressed in US$/kWh or US$/kW, but the technology cost can be misleading, as it omits significant costs associated with engineering, installation, and the balance of plant. The vertical axis of Figure 2 describes the full capital cost in US$/kW for a storage system capable of charging at rated power for eight hours. Aside from the actual storage medium itself, the remaining portions of the full capital cost are related to the power rating of the system. This includes the significant portion applied to engineering, procurement, and construction (EPC). Therefore, the cost per kw of a storage installation with fewer kwh of energy capacity would not necessarily be proportionately less. july/august 2012 IEEE power & energy magazine 79

6 Several emerging trends will affect how grids meet their capacity needs in the future. Since the spread of peak and off-peak energy prices is an important parameter, it is important to consider factors that could affect this spread over the lifetime of the asset. If the fuel price for peaking generation were to increase, the spread between peak and off-peak generation would increase. As new wind power generation is built, wind energy will increasingly displace generation operating at the margin, potentially changing the mix of generation operating during peak hours and off-peak hours. Policies and regulations could affect the type of generation that operates at the margin, thereby affecting the price of peak energy and the price spread between peak and off-peak energy over time. Recent wind integration studies, such as a National Renewable Energy Laboratory study of the western region of the United States, suggest that incremental additions of wind power will further reduce the spread between peak and off-peak energy prices, even though there will be occasions when the price spread is dramatically larger (see For Further Reading ). Such occasional large price spreads tend to be of insufficient frequency and duration to substantially alter the overall trend in which renewables reduce the average spread between peak and off-peak energy prices. Generation Capacity Historically, resource adequacy has been one of the key drivers for the construction of new generation facilities. Most grids maintain a generation capacity margin (the amount of installed generation in excess of the peak load) of about 15%. The most common approach for resolving a capacity shortage is to build more dispatchable generation. Energy storage may offer a competitive alternative to constructing Power Demand Demand Exceeds Transformer Rating: Storage Discharges Transformer Rating Time Storage Charges figure 3. A storage device can mitigate occasional transformer overload conditions. new plants to meet system capacity needs. In fact, pumped storage hydro plants have contributed to grid capacity for several decades. Several emerging trends will affect how grids meet their capacity needs in the future. Older generating plants that do not meet new environmental requirements are being retired. Renewable energy is increasing, dominated by wind and solar generation facilities that typically contribute only 10% to 30% of their plant ratings toward grid capacity. As a result, grids will have increasing needs for generation capacity. The value proposition for providing generation capacity through storage can be approached in a manner similar to financial energy arbitrage. The capital cost of the storage system can be compared with the capital cost of a generating plant. For example, the cost of a peaking plant is generally less than US$1,000/kW installed. While the peaking plant has a marginal cost of operation tied closely to its fuel price, the storage system has a marginal cost of operation tied closely to its round-trip efficiency and the energy prices in the region. A comparison of the up-front capital cost and the operating costs could be made for these two approaches to determine the lowest-cost approach for providing new capacity. Based on the costs of storage and fuel prices for peaking generation, storage may enable the deferral of new generation while providing additional services to the grid. Equipment Capacity Energy storage could also be applied to relieve a temporary periodic overload condition or other similar constraint. For example, a transformer experiencing loading that occasionally exceeds its rated capacity might constrain the delivery of electricity to downstream customers. One remedy would be to upgrade the substation (by reconfiguring feeders, adding a transformer, increasing the capacity of a transformer, or taking other actions). As an alternative to these conventional approaches, a storage asset situated downstream of the congested transformer could relieve the constraint and defer the investment to upgrade the substation. Figure 3 shows energy being shifted from one time to another to avoid overloading a transformer. The costs of incremental substation equipment upgrades are usually well known, as would be the specific nature of the constraint (its severity, duration, economic and regulatory cost if left unaddressed, and other such factors). For this application, the storage facility could extend the useful life of the substation with a potentially lower capital expenditure while deferring the larger investment needed to upgrade the 80 IEEE power & energy magazine july/august 2012

7 entire substation and install capacity that may not be needed for years. The duration, severity, and frequency of the overloads both those present and those projected for the life of the substation deferral will dictate the storage power and energy ratings and the preferred storage medium. Situations in which substation expansion is unusually expensive, such as locations where space for expansion is critically scarce, will be more attractive for this type of storage application. The conditions for profitable storage deployments in this context will be most apparent to the owner of the infrastructure. Line Congestion Curtailment of renewable energy due to transmission constraints can also be mitigated by storage and can be assessed through similar analysis. Much has been written about the reasons for and frequency of wind curtailment events. A well-known example occurs in Texas, where the amount of wind generation installed in some regions exceeds the transmission capacity at certain times of the year when the wind resource is especially robust. The costs and time required for building new transmission system infrastructure to accommodate all the wind energy this resource can generate are reasonably well known, as is the lost revenue from curtailed wind energy. Simply crediting the value of spilled energy against the cost of a storage device implies, however, that the wind plant owner has no other use for capital. This is, however, not the case. Let us examine two investment alternatives: 1) deploying storage to capture the wind energy and deliver it later, when the congestion is relieved 2) deploying a new wind plant, situated elsewhere in the transmission system, where this wind plant would be less likely to encounter congestion and would be able to deliver its energy to the grid. For the first case, consider a storage system designed to capture energy spilled by a wind plant. In this instance, the energy is being spilled because of a transmission constraint caused by the fact that many plants, including wind plants, are situated in the same region of the grid, far from the load. The storage system is used to capture the wind energy during congestion events and deliver it back to the grid when the congestion is relieved. Below we describe some of the factors that should be considered when sizing a storage system for this application. The size of the storage system is defined by the power that can be absorbed or extracted and the energy that can be stored in the system. To estimate the appropriate size of the storage system, assumptions must be made about the congestion pattern and the wind profile, both of which are uncertain and variable. If a wind profile and congestion pattern is defined by historical data, the amount of power being curtailed at the plant, how often the curtailment takes place, and the duration of each curtailment period can all be established. These factors are necessary to determine the most appropriate power and energy rating for the storage system. As an illustrative example, a single week of wind power production during each season of the year is shown in Figure 4 for a 200-MW wind plant. Each season has a specified congestion limit. For the summer week, the wind plant would encounter a congestion limit for almost four consecutive days. If a storage system were large enough to capture this type of wind curtailment event, it would be underutilized for the many smaller curtailment events experienced throughout the year. For example, during the spring week, the storage system would be used only modestly just on the last day of that week. For the storage system to generate sufficient revenue to cover the up-front investment, the revenue must be realized for many years of operation. Since the storage system was sized for an assumed wind profile and congestion pattern, care should be taken to consider the possibility that the wind profile and congestion pattern could change in the future. Some of the factors that could affect the wind profile and congestion pattern include: Fuel prices and the present and planned generation mix: If the wind plant is situated in an area with substantial nearby generation, the congestion will be affected by the variable cost of operating these plants. If the operating cost of these units drops relative to plants situated elsewhere in the system, the congestion periods could increase in frequency and duration. Transmission expansion plans: If congestion persists, the transmission owner could build out additional capacity to the region, thereby reducing or eliminating the congestion events. This could result in a significantly oversized and underutilized storage system. Wind and Solar Power Smoothing Energy storage is an approach often mentioned for smoothing the output from variable wind and solar plants. Storage, it is argued, offers the stabilization and buffering capacity that will be necessary to mitigate the variability of renewable generation and the operational challenges it might cause. The authors have contributed to studies evaluating the integration of levels of wind and solar power in excess of 30% annual energy penetration. These studies have concluded that the impacts associated with the variability and uncertainty of wind power can be addressed through conventional approaches and that energy storage is not necessary to reach high levels of wind and solar power penetration. In general, large interconnected systems are designed to manage load variability and can manage the additional short-time-scale variability (usually measured in seconds) introduced by the levels of wind and solar power likely to be brought online in the United States under existing renewable portfolio standard (RPS) scenarios. Smoothing the output of individual wind turbines or individual wind plants is not normally systemically necessary. In some situations, july/august 2012 IEEE power & energy magazine 81

8 Wind Power (MW) Congestion Limit 1 One Week in Winter 2.5 Days of Wind Curtailment Will Require a Large Storage Energy Rating Wind Power (MW) Congestion Limit 1 One Week in Spring Few Hours of Wind Curtailment Will Require a Modestly Sized Storage System Wind Power (MW) Congestion Limit 2 One Week in Summer Nearly Four Full Days of Wind Curtailment Will Require a Large Storage Energy Rating Wind Power (MW) Congestion Limit 2 One Week in Fall A High-Power Wind Curtailment Event Will Require a High Storage Power Rating figure 4. Wind power production and congestion limits for four different weeks of the year. however, the additional flexibility brought by storage could be very valuable. Some of these situations include: regions where the thermal generation fleet is relatively inflexible regions where the system is weakly interconnected or islanded from a larger grid regions where the power system is relatively small, with few generating plants regions where there is a lack of spatial diversity of the wind and solar plants (the plants are all near one another). In these cases, storage could provide ramp-rate control, enabling the variable power production to limit its up and down ramps in power and inject a smoother energy profile into the grid. The upside of this approach is that events at a single wind or solar plant, where the change in power could be substantial, would be smoothed and the power system would not have to cope with the sudden change in wind or solar power. When a sudden drop in power at one wind plant is observed, however, it is possible that a rise in power at other wind plants is taking place. In this instance, one storage system will be discharging while another storage system will be charging. If no storage were present, the changes in wind power at each plant would have effectively canceled one another out. To avoid this canceling effect, a centrally managed, grid-level storage system capable of responding not only to wind and solar power variability events but also to system events related to load variability could be particularly beneficial. Frequency Regulation Grid operators manage the balance between supply (generation) and demand (load), correcting for any imbalance by adjusting the output of a subset of the generation up or down when the grid frequency drifts away from nominal frequency (60 Hz in North America). Over time, any accumulated imbalance between supply and demand must be addressed by raising or lowering the output of generating units. The imbalance between supply and demand can be further exacerbated by scheduled interchanges of power between control areas. For large interconnected systems, the interties can contribute to or dominate this balancing equation, driving increases in area control error, (ACE), the standard industry measure of departure from scheduled interchange plus any accumulated error in the frequency. Functioning within a time scale of seconds, the automatic generation control (AGC) takes action by requesting a subset of the generation to increase or decrease its output. The AGC raises or lowers the output from generators, with the objective of driving ACE to zero. This service is normally procured as an ancillary service, usually simply termed regulation, for which many 82 IEEE power & energy magazine july/august 2012

9 systems have markets. Recently, many systems have modified their market rules to allow other resources, most notably storage and responsive demand, to participate in these markets. FERC has also helped to enable markets for energy storage systems in the United States. As an example, New York Independent System Operator (NYISO) has recently established a limited energy storage resource (LESR) market. This market will let energy storage systems participate in a 5-min short-duration frequency regulation market. Ramp rates for battery and flywheel energy storage systems are very high (they arrive at full power within a few seconds). ISO New England has recognized the value of higher ramping capability by providing incentives for resources that provide greater response; it rewards the generators that move upward and downward more frequently and more rapidly in response to changing regulation signals by offering a mileage adder to the typical regulation service payment. As more wind and solar generation penetrates the grid, the requirements for regulating resources needed to respond to AGC will also increase. It is unlikely that this increase will be substantial because of the geographic diversity of wind plants, which reduces the short-time-scale fluctuations in wind power within a balancing area, making the incremental requirements for regulation less than is often anticipated. The impact of variable generation on regulation also depends on how the longer-time-scale variability in wind and solar power is managed by the grid operators. For example, when all unexpected variations within an entire hour are managed by regulation, the requirement increases dramatically. But when these variations are managed by subhourly energy dispatch and reserve markets, less regulation is needed. Other Applications Energy storage systems integrated into the operation and management of the power system could be capable of providing desired governor and inertial response characteristics, managing harmonics to improve power quality, regulating voltage via power electronics, contributing to black-start capabilities, and offering other ancillary services at both the transmission and distribution voltage levels. Market mechanisms for quantifying the types of benefits offered by storage exist only rarely at present, so evaluating the costs and benefits of storage facilities providing these services is difficult. The value of these services alone is probably insufficient to justify the deployment of energy storage. In some instances, however, these additional services could be performed by storage systems deployed to serve the other applications listed in Table 1. For example, it may be possible for a storage system installed to perform financial energy arbitrage to also perform governor and inertial response. On the other hand, a storage system installed to perform frequency regulation would probably not be able to perform financial energy arbitrage unless the storage medium were increased in size to allow the performance of both services or each service were performed at different times. The type of storage medium must also be taken into account. One type of battery chemistry may be better suited for one of the applications but may not perform as well as alternative chemistries for the other application. Competing Technologies Interconnected power grids have operated reliably for decades without the benefit of storage (beyond the very real contributions of pumped hydro), and while the grid is undergoing some important changes, there remain a number of proven tools and mechanisms for addressing the operating challenges of today. Some of these may be deployed with equivalent or greater efficacy and economics to solve the same problems that storage assets are targeting. GE has performed wind integration studies in a number of regions around North America. From these studies, a common observation can be made: the power generation resources at the economic margin will be displaced by wind energy as the penetration of wind power increases. In most regions of the United States, this means that the vast majority of displaced generation will consist of gas-fired combustion turbines and combined-cycle plants. It could be reasoned that the underutilization of this generation in a power system with significant levels of wind power will free these units to provide other services for the grid, such as provision of nonspinning reserves, reserves to cover for shortfalls in wind power when a forecast has overestimated the amount of wind power available, and system capacity during peak load events. The conventional wisdom that gas in the ground offers many of the benefits of energy storage remains valid to a large degree. Displaced gas generation should be able to serve many of the same applications for which storage is targeted. In situations where gas supply has flexibility constraints, either from market or physical limitations, gas storage could be a competitive alternative to electric energy storage. The power system manages uncertainty in load forecasts and wind forecasts by carrying additional reserves to cover load and wind forecast errors as well as system contingencies, such as the loss of a large unit or transmission line. Load is uncertain, but wind and solar power production are even more uncertain. Forecasting load is a relatively well understood practice. While wind forecasting technologies have improved substantially over the past ten years and dayahead wind forecasts average less than 20% mean absolute error, there are inevitably errors that have consequences for systems with high penetrations of wind power. Wellunderstood practices for managing load forecast error do exist, however, and new technologies or new uses of existing technologies are emerging to meet such system needs. Energy storage is often mentioned as a technology useful in covering for generation shortfalls, but other technologies can also provide this service, including thermal generation and demand response. An emerging trend in markets with high local wind penetration (e.g., the area managed by Electric july/august 2012 IEEE power & energy magazine 83

10 Reliability Council of Texas, or ERCOT) is the deployment of relatively small (<10 MW) reciprocating engines as a means of firming large wind plants. These generators are attractive because of their low capital cost (<US$1,000/kW installed) and small size, which allow them to be deployed as needed. While they have a relatively higher operating cost than some alternatives, they are only operated occasionally. Low-cost generators challenge deployments of more expensive energy storage systems. Modifying system operation to procure more spinning reserve is another approach for addressing these needs. Each approach or mix of approaches will compete with others in power markets and will be assessed by utilities looking for the most economical way to provide the services desired. Another approach that competes with energy storage is demand response. Considering load as a resource with which to address the challenges that occur on the grid in the shortest time scales, such as those of frequency regulation, is an emerging trend. Today, in longer time scales (from 5 min to an hour), demand response could provide a suitable substitute for calling on the nonspinning operating reserves that are dispatched when the load is greater than anticipated or the wind forecast has overestimated the amount of wind generation available. A growing number of utilities have demand-response programs that are growing in both acceptance and maturity. Looking Forward The changes the electric grid faces over the coming years are arguably more radically transformative and will occur faster than anything seen in the history of the system. In emerging economies, the parallel evolution of their power systems will pose both similar and unique challenges. Changes in the generation mix, load behavior and management, and distribution automation all promise to alter how we think about and operate the grid. Deployment of wind and solar power will demand greater flexibility in the power system to manage the uncertainty and variability of these resources. These trends do not in themselves mean that wind and solar plants require storage or that power grids with wind and solar generation need storage. Rather, storage provides grid operators with a flexible resource that can free other generating resources to serve the grid in other ways, if and when the economics justify it. Grid-scale energy storage is one of the technologies that promises to bring about these changes. While energy storage may indeed find suitable applications, it is imperative that potential investors in these assets understand how to quantify the technical and economic suitability of energy storage on an application-by-application basis and in comparison with other approaches to achieving the same goals. To speak more simply, the potential applications for storage come down to a simple evaluation, guided by just a few questions. Seeking answers to the following questions will help frame the comparison between storage and alternate solutions: Energy Applications How much energy is available to be stored, and when? How often is the storage system needed? What size storage device is needed, i.e., what should its power and energy ratings be? What is the total capital cost of the storage device? What is the cost of the energy to be stored (i.e., the charging cost)? What is the value of the energy when it is released to the grid (i.e., the discharging price)? Power Applications What is the value of the service provided (regulation, operating reserves, and so on)? What is the capital cost of a storage device for that service? All Applications How much of the energy is lost in the storage device, i.e., what is the round-trip efficiency, and what are the other operating losses? What are the operating costs associated with owning and operating a storage system? What is the life expectancy of the storage medium? What are the other methods available to solve the same problems, and what are their costs? If funds are invested elsewhere, is there a better return on investment? Potential applications for energy storage in power grids are increasing, but alternative solutions are possible. The challenge is to find the sweet spot, where energy storage provides an essential function or service at lower total cost than alternative solutions. For Further Reading National Renewable Energy Laboratory. (2010). Western wind and solar integration study. [Online]. Available: Independent System Operators of New England (ISO- NE). (2010). New England wind integration study. [Online]. Available: comm_wkgrps/prtcpnts_comm/pac/reports/2010/newis_ report.pdf G. Hunt and J. Szymborski, Operating in a utility battery energy storage system (BESS) for 12 years, in Proc. ABSOLYTE Technology Professional Paper on Battery Energy Storage Systems, Aug. 2009, pp Biographies Devon Manz is with GE Energy Consulting in Schenectady, New York. Richard Piwko is with GE Energy Consulting in Schenectady, New York. Nick Miller is with GE Energy Consulting in Schenectady, New York. p&e 84 IEEE power & energy magazine july/august 2012