Merchant Storage Investment in a Restructured Electricity Industry.

AuthorSiddiqui, Afzal S.

    Recent years have seen a renaissance in the development of energy storage. Sioshansi et al. (2012) note that this interest in storage is prompted by a number of recent electricity-industry developments. One is that storage was viewed almost exclusively as an alternative to high-cost peaking generation in the 1970s, when much of the pumped hydroelectric capacity that is installed today was first built (cf. the work of EPRI (1976) as one example showing this). More recent analyses of energy storage, with the work of EPRI-DOE (2003) being a seminal example, recognise that storage can provide many services beyond avoiding the cost of installing and operating peaking generation. A second major development is the advent of restructured electricity markets, which provide transparent price signals for many of the services that energy storage can provide. Finally, Denholm et al. (2010) note that energy storage is expected to have a growing role in electric power systems as the penetration of variable renewable energy grows. As another example of this, von Hirschhausen (2014) surveys the potential role of energy storage in achieving the German energy-related policy goals underlying its so-called 'energiewende.'

    Although energy storage has the potential to deliver transformative benefits in the production and consumption of electricity, it is by no means a panacea. One important issue around the impacts of energy storage involves its interactions within an imperfectly competitive market environment. This is because charging and discharging storage, and the resulting purchases or sales of energy, can affect market prices and the welfare of energy producers (i.e., electricity generators) and consumers. Sioshansi et al. (2009) demonstrate the potential for energy storage to mitigate the impacts on wholesale electricity prices of limited natural gas supplies after Hurricanes Katrina and Rita in 2005. They show that merchant-operated energy storage could have delivered net social welfare increases, assuming that the generation sector is perfectly competitive. Sioshansi (2010) employs a stylised partial-equilibrium model, in which electricity generation is perfectly competitive but wholesale prices respond to storage use, to analyse the welfare impacts of storage use. He shows that a merchant storage operator tends to under-use storage compared to welfare-maximising use, because it is profit-maximising for the firm to maintain a larger price difference between on- and off-peak periods (relative to welfare-maximising use). Using actual data from Germany, Schill and Kemfert (2011) compare cases in which storage may be owned by generators, who themselves may exert market power (a la a Nash-Cournot equilibrium). They find cases in which having storage in the market reduces welfare compared to a no-storage case. Sioshansi (2011) uses data from Texas to conduct a welfare analysis of energy storage in a market with high penetrations of wind energy. As in the work of Schill and Kemfert (2011), he shows that adding storage to an imperfectly competitive generation market can reduce social welfare compared to the no-storage case. Virasjoki et al. (2016) develop an equilibrium model for a Western European test network to study the impact of storage on ramping costs and grid congestion in the presence of a realistic level of renewable energy. They show that the overall welfare impact of storage is modest, but that it can, nevertheless, reduce ramping and congestion costs. This benefit of storage is limited, however, if storage-owning producers exert market power. Sioshansi (2014) uses a stylised equilibrium model to consider various ownership and market structures involving energy storage. He concludes that storage can be welfare diminishing (compared to a no-storage case) in the presence of strategic generating firms. Surprisingly, he finds that welfare losses that arise with strategic generating firms can be greater if storage operations are perfectly competitive. Shahmohammadi et al. (2018b,a) take a bi-level approach to modelling strategic interactions among storage, conventional generation, and renewable units. They model strategic offering behaviour by these units at the upper level, subject to a least-cost market-clearing model, which determines prices and dispatch, at the lower level.

    An important limitation of these welfare analyses of energy storage is that they do not endogenise storage investment. In essence, these works all show that if a given amount of energy storage is in the system, there may be market structures under which social welfare is increased or decreased relative to a no-storage case. They do not, however, show whether investments in storage are individually rational in cases in which storage would be welfare-enhancing or -diminishing. Indeed, to our knowledge, there are only two works in the extant literature that endogenise storage investment within a market-equilibrium framework. However, neither of these works examines the welfare impacts of storage investment in an imperfectly competitive market. Nasrolahpour et al. (2016) develop a bi-level model with a profit-maximising storage investor at the upper level and market clearing at the lower level. Their model and case study account for uncertain renewable-energy availability and realistic generation constraints based on data from Alberta. They assume, however, a perfectly competitive generation sector and do not conduct a welfare analysis or a comparison of the merchant's decisions with those of a welfare-maximising storage operator. Thus, they do not model the full range of market imperfections that may yield welfare losses. Dvorkin et al. (2018) develop a tri-level model that includes a profit-maximising merchant storage investor. However, their focus is on using energy storage for alleviating transmission congestion. Thus, their modelling framework cannot unveil the types of market welfare impacts that ours does.

    The aim of this work is to fill this important gap in the existing literature. We do this by extending the stylised equilibrium model that is proposed by Sioshansi (2014) to explore the welfare implications of storage with an imperfectly competitive generation sector. However, unlike the extant literature, we investigate not only market operations but also the storage-investment decision itself. Specifically, we posit that storage capacity is owned by either a profit-maximising standalone merchant investor or a welfare-maximising storage operator. In either case, the storage operator is a leader in the sense that it anticipates the response of the generation sector when making its storage-capacity investment. Moreover, the storage operator is distinct from both the market operator and the generation firms. Thus, we use a bi-level modelling framework, in which the lower level reflects market operations consisting of storage use and multiple symmetric generators across two time periods (off- and on-peak periods, respectively). At the upper level, we have a single storage investor, i.e., either the profit-maximising merchant or the welfare maximiser.

    We find analytical solutions for the optimal storage capacity adopted by each type of investor and investigate the welfare effects of each. In particular, we prove that the welfare maximiser invests in more storage capacity than the profit-maximising merchant if the generation sector is relatively uncompetitive. This is because the welfare maximiser uses a large storage capacity to subvert the generators' strategy of withholding generation by moving energy to the on-peak period. Conversely, the profit-maximising merchant is content to profit from the high price differential that results from the generators' behaviour. It is, thus, reluctant to erode its profit by installing a large amount of storage capacity. In a more competitive industry, the welfare maximiser reduces its storage capacity to below that of the profit-maximising merchant as there is less welfare loss to mitigate from the exercise of market power by generators. With a relatively competitive generation sector the profit-maximising merchant maintains a relatively large storage capacity to increase its profit by trading a large volume of energy. The larger volume of energy transacted partially compensates for the lower price differential between on- and off-peak periods. If the generation sector is sufficiently competitive, then the behaviour of the profit-maximising merchant is actually welfare-diminishing vis-a-vis having no storage at all. This result runs contrary to those of Sioshansi (2014), who finds that there can be no welfare losses with a perfectly competitive generation sector if the storage capacity is fixed. Thus, our analysis shows that the welfare implications of energy storage is highly sensitive to representing the investment decision.

    Next, we show that the profit-maximising merchant may be induced to invest in the same level of storage capacity as that of the welfare maximiser via a ramping charge on generation. The ramping charge essentially penalises generators and the storage operator for having a large difference in the off- and on-peak load. This ramping charge can have the effect of mitigating the incentives of storage and generation firms to maintain large price differences between the on- and off-peak periods. Finally, through numerical examples, we illustrate that storage investment by a profit-maximising merchant in the presence of a ramping charge may increase social welfare to above the level that is attained by a welfare maximiser. This is because the ramping charge offers another layer of 'control.' This added control can mitigate potential welfare losses from inefficient storage use and the withholding of capacity by generating firms.

    Our work is limited in the potential value of energy storage that it captures, insomuch as we focus solely on using energy storage to shift...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT