What Next for Energy Storage?

Written by Ralph D. Masiello

Prospects for wide integration of energy storage into grid systems will be enhanced with the development of market mechanisms that allow for coordinated trading of charge and discharge time. However, introduction of such mechanisms represent a challenge to traditional thinking and require added complexity and computation. Conceptually, energy storage straddles our conventional categories of transmission, distribution and generation.

The U.S. stimulus bill of 2009 (ARRA) and the Energy Department's Advanced Research Project Agency (ARPA-E) have provided funds for a number of projects to develop and demonstrate energy storage for electric system applications. There are frequent press releases of new advances in storage technology at universities, industrial labs and start-up companies.

Yet established advanced battery manufacturers are struggling to keep their doors open, largely due to slower than hoped for electric vehicle sales. Could grid storage applications be a lifesaver? While some grid applications, especially regulation services, continue to attract merchant storage developers, other applications are slow to develop due to difficulties in making the finances attractive or in negotiating existing regulatory and market protocols.

One of the biggest difficulties is that energy storage is being forced into one of the traditional buckets of generation, transmission or distribution (G, T or D), or "customer," when in reality it is not exactly any of the above and a little bit all the above. That in turn compounds problems in: (a) monetizing all the value streams that storage can provide, (b) accurately estimating/calculating those value streams, and (c) convincing utilities and regulators that storage can be a prudent investment given the novelty of the technology.

Consider, for example, regulatory practices and wholesale market protocols that try to consider storage as a "G" or a "T" asset. Forcing storage to bid into the wholesale markets as a generator makes the storage operator a kind of trader seeking to gain from the time arbitrage of energy. But bidding as a generator means the storage operator has to enter bids for discharging at specific hours and charging at specific hours, and if the storage charges, say, at 2 AM to 4 AM, it probably cannot charge any farther at 6 AM. Thus it cannot bid to charge at all off-peak hours but must "choose," and the market loses the opportunity to co-optimize the storage charging. The same is true on the discharge side of the cycle.

This is definitely sub-optimal for the wholesale markets which could instead co-optimize the charge/discharge for best market efficiency/lowest cost of wholesale procurement and base that on a bid structure designed for the storage asset. In other words, the true service that storage provides is time shifting energy production, not actual production or consumption.

Given the perceived need for either large amounts of peaking generation, demand response or storage to integrate variable energy resources (such as wind and solar) one could imagine markets conducting capacity auctions for "firming"—that is to say, time shifting—capacity. Such a process would enable a market-driven and technology-independent solution to the need while providing a revenue stream that storage developers could use to obtain financing. Such a mechanism, albeit indirectly, may come about if variable energy resource (VER) developers are required to offer energy on a firm basis—this could lead to a secondary market for firming services even in areas that do not have wholesale market operators today.

Many market operators do co-optimize pumped storage operations to some extent. These are usually legacies of pre-deregulation practices for singular facilities with special market power. Market operators are leery of extending this model to all new storage facilities if only because of the computational complexities entailed.

System operators' requests of proposals (RFPs) for peaking generation for congestion relief could admit storage as a viable technical alternative if structured around simply congestion relief rather than a generation construct. Instead of only focusing on the power production capability, allowing for time shifting of demand would accomplish the same benefit. That is, rather than writing a specification around peaking turbines as a solution, a specification around the amount and duration of power and energy required to mitigate local congestion might be more attractive to storage developers.

Energy storage providers could be attractive bidders especially when congestion costs are caused by N-1 contingency limitations. (These would be situations arising from the requirement that market/grid operators must operate the grid so as to be safe against first generation/transmission outage events: Such events can result in "congestion" due to an overload that would occur as a result of an outage, even if before the outage the loading is within limits.) The ability of most storage technologies to respond very rapidly makes them a real candidate to provide 15 or 30 minutes of post-contingency relief rapidly while conventional generation is brought back online.

The relatively infrequent operational cycles of contingency relief, unlike regulation services, makes it a relatively attractive service for storage technologies. For this purpose, storage offers many advantages over peaking generation—no environmental issues to deal with, easy siting at transmission stations and, almost certainly, higher reliability.

To be sure, reliability criteria for successful starts have not been established for this conceptual application, and the successfulness of starts has not been measured for storage technologies.

Then there is the entire discussion about how to pay for such a service. By its successful implementation it would destroy the potential arbitrage gains from congestion pricing, making a merchant approach to its provision unlikely. Making the service directly rate-based as a transmission asset or indirectly remunerated as a long-term power purchase agreement (PPA) would resolve this problem. Doing so would require transmission planners, transmission operators and regulators to accept storage as a congestion relief asset.

The process leading to Federal Energy Regulatory Commission's order 755, which specifies "pay for performance" for regulation services, is a large step in the direction of technology-independent recognition of the potential benefits of resources with novel technical capabilities, like fast storage. At the same time, the ongoing process of developing the particular market protocols, product definitions and payment/settlement schemes illustrates once again that the devil is in the details.

The PJM filing with respect to this is especially interesting. PJM designed a fast regulation product that incorporated the concept of "zero energy" over a five-minute window (meaning that the fast regulation resource would deliver zero net energy to the grid in each five minutes of operations, in effect that the state of charge of a storage resource is maintained on a five minute basis). PJM then analyzed the operational benefits of increasing penetrations of fast storage and fast regulation as a share of the total regulation market. Results suggest that a longer energy duration would allow for positive benefits at higher penetrations, and that at the five-minute window there are definite points of diminishing returns for fast regulation penetration.

At the distribution level, storage appears to offer benefits of peak shaving and also smoothing of local photovoltaic production and accompanying voltage fluctuations. Pilots are underway to demonstrate these benefits. But, distribution engineering is necessarily conservative and distribution utilities are accustomed to apparatus that can safely be ignored once installed. Storage facilities will require condition self-assessment and communications technologies as part of advanced distribution automation systems in order to be accepted by distribution utilities. Not only proof of concept with the pilots but demonstrated longevity and reliability are required; just as important, planning software products will have to incorporate storage as effortlessly as they do all other apparatus.

These issues all point to a slow building market for electricity storage. The struggles of suppliers that were looking at grid storage as a short-term filler until the EV market developed suggests that utility applications will not be their lifesaver. Meantime, the estimated need for storage driven by growth in renewables at all levels in the system is very real. It is encouraging that General Electric has made major commitments to storage technologies with the Durathon, aimed at rail, renewables integration and grid applications. But the regulatory and asset categorization issues are part of a chicken-and-the-egg problem, holding back adoption of storage and making the forward cost reductions needed for widespread acceptance that much more difficult. Addressing these issues is as critical as proving the technology and reducing costs.




Dr. Ralph D. Masiello, a member of the National Academy of Engineering, an IEEE Smart Grid Technical Expert and an IEEE Life Fellow, is DNV GL’s Innovation Director and Senior Vice President. He has served as chairman of the IEEE Power Systems Engineering Committee and serves now on the editorial board of the IEEE Power and Energy Magazine. In 2009 he received the IEEE PES Charles Concordia Power Systems Engineering Award. He earned his B.S., M.S. and Ph.D. in electrical engineering from the Massachusetts Institute of Technology, where he worked on the very early applications of modern control and estimation theory of power systems.