Chapter 8. Discussion
8.2. Storage value in single vs multiple markets
The results in Chapter 5 have shown that there are potentially substantial arbitrage revenues in the APX market and BM. These were equivalent to £55/kW-yr and £82/kW-yr for the APX market and BM respectively in 2013. Across the 10 years from 2005-2015, APX arbitrage revenues ranged from £33/kW-yr to £102/kW-yr. These substantial variations, across mechanisms and across the years, imply that storage projects should factor these in building an economic case, otherwise an overestimation or underestimation of revenues may occur when these results are extrapolated over long timescales. Although not directly comparable due to design and parameter differences, Sioshansi et al., (2009) found wholesale arbitrage revenues in the PJM market to range from approximately £40/kW-yr to £75/kW-yr from 2002-20078. Connolly et al., (2011) showed the large variations of revenues across
several markets worldwide but did not specifically investigate whether volatility was the reason behind the large variations. They found the arbitrage revenues in Alberta, Canada to be the highest of all 13 markets they explored. Safaei & Keith (2014) argued that the large price differential is due to the large proportion of electricity being from baseload generation and uses a real time gross pool (as opposed to common day ahead pools). As a result, they explain, the low load factor peaking plants recover their costs through high bidding prices, which under a gross pool becomes the MCP. Subsequently, Safaei & Keith (2014) showed a strong correlation between the standard deviation of electricity prices and CAES profit.
In the Australian markets, McConnell et al., (2015) show the arbitrage value of storage to range between £26/kW-yr to £184/kW-yr9. They point out that the price volatility in the Australian market is
one of the highest in the world. Chapter 3 showed that the BM has more volatility than the APX market
8Using £1=$1.5 (USD) exchange rate 9 Using £1= $1.9 (AUD) exchange rate
and consequently, in Chapter 5, it was shown that the BM revenues were substantially higher than those from the APX market. In Chapter 7, part, the reason why an increased wind penetration increases storage value, despite its clear depressing effect on prices, is that the additional wind generation increases price volatility thereby creating opportunities for storage. Therefore, the strong link between price volatility and storage revenues has been shown in this thesis; furthermore, the findings from previous studies also support this finding.
In evaluating and comparing the revenues in each market mechanism, identical storage parameters were used in each of these markets. While this allowed for a basis of comparison in terms of revenues, these parameters have not been optimised for these markets. Thus the energy to power ratio is initially chosen as 12 hours’ equivalent to avoid restrictions as mentioned in section 4.2. However optimal energy to power sizing in the APX and BM market would likely be less than 5-6 hours of equivalent as the sensitivity analyses in Chapter 5 and 6 have shown. For the sole provision of FFR however this ratio is likely to be even smaller since the requirements for the provision of FFR is a minimum ratio of 0.5 hours. In Chapter 5 where a revenue comparison was made across mechanisms, smaller ratios would not substantially change the revenues considering the minimum requirements were met and that most of the revenues were derived from available capacity. Furthermore, it is unlikely that disturbances are consecutive and even so, in reality, recovery periods would be active, an aspect not explored in this thesis. However, for the calculation of NPV from FFR related revenues, smaller energy to power ratios are one of the key determinants of economic feasibility. Large power and energy capacities result in higher capital costs and since FFR revenues are mostly power capacity focused much smaller power capacities would likely be preferred.
In Chapter 6, when market constraints were imposed, revenues were reduced, some to a greater extent than others; the APX market showed great liquidity being almost completely unaffected. In the BM, although prices are higher, opportunities for arbitrage trades are limited by the state of imbalance on the system as well as the magnitude of the imbalance volume. As a result, the cross-system price arbitrage, or arbitrage between the system prices, is severely affected. Firm frequency response is not strongly affected in the absence of a utilisation profile since these have low utilisation rates; in fact, most of the revenue is derived from availability payments. FFR and STOR provision have special implications: they are niches for storage technologies that are highly sensitive to cycle life but with low self-discharge rates, for example, Lithium-Ion batteries.
These results highlight the need for caution when operating in the Balancing Mechanism alone whereas wholesale market arbitrage and FFR revenues are more easily accessible. Under a co- optimisation model, there is a synergy between the three mechanisms, being able to generate a high revenue which is lightly affected by market constraints. This occurs because constrained operation in
one mechanism can be compensated by operating in another. More specifically, the constrained BM operation is compensated by participation in the APX market, thus benefitting from the high-value arbitrage trades in the BM and the increased liquidity in the APX market, essentially getting the best of both. Idle times whereby storage energy capacity is non-zero can be offered for the provision of FFR. Furthermore, because the FFR service is not called on all the time, residual energy can be sold in the other markets, generating additional revenues for the same energy volumes. Therefore, a co- optimisation of revenues is more resilient to market constraints and also generates additional revenues due to the synergy of operations. This mode of operation is thus of particular significance to a private investor who is risk averse.
Storage operation and system benefits were shown to align when the system operates in a single revenue mechanism. However, Co-optimised storage operation does not always align with system benefits, as the model chooses trades that increase revenues at the expense of those that could bring about a system benefit. For example, if there is an underestimation of demand at the early hours of the morning when demand is the lowest, a shortage of energy in the Balancing Mechanism arises. Usually, storage would be charging at that time and thus, under co-optimisation may not participate in alleviating the imbalance. Imbalances at low levels of demand can bring about rapid changes in the system frequency, which highlights the need for FFR during those periods. Thus a situation arises whereby storage operation is desirable in the BM or for the provision of FFR but financially there is greater value in maintaining a status quo by charging (and purchasing power from the APX market). This conflict can be mitigated by providing aligned financial incentives, a finding that is of particular relevance to policy makers.