One of the amendments to the BM in November 2015, briefly mentioned in Chapter 3, has special implications for values; a reduction in the price average reference volume from 500 MWh to 50 MWh increases the penalising effect of not meeting contracted demand or generation. According to the imbalance pricing method, an average of the most expensive 50 MWh of accepted offers will constitute the imbalance price (previously SBP) during periods of energy shortage on the system. Similarly, during periods of excess energy, the average the cheapest 50 MWh of accepted bids will constitute the imbalance price (SSP previously).
The amendment to the BM also included the addition of a function to reflect the impact of potential power loss to consumers (Ofgem 2015); this function is derived from the value of loss load and the loss of load probability. A value of loss load of £3000/MWh is initially implemented, rising to £6000/MWh
on 1st of November 2018. It is conceivable that during periods of high system stress, whereby the loss
of load probability rises, the imbalance price could be affected.
The elimination of dual prices (reverse price and imbalance price) for a single imbalance price has no impact on the co-optimised value of storage for several reasons; the reverse price which is almost identical to the half hourly APX market price is already accessible to the model under co-optimisation. Even in the case when storage operates in the BM only, market constraints mean that the storage system can only take actions that help the system and aligns operation to those of interest of the system operator. In this case, too the reverse price which applies to opposite actions, is irrelevant. The co-optimisation and BM model was re-run under the single price and confirmed that there was no difference in revenues from this change. However, with a greater incentive to help balance the system, parties may behave differently and this potential change in behaviour following this amendment is unknown at present.
7.6. Conclusion
This aim of this Chapter was to investigate the value of storage under increasing wind penetration levels. In order to do so, the impact of wind penetration on both the APX and BM prices was evaluated. The results of the econometric analysis show that the most influential variables on the APX price were a reflection of the merit order of generation; peaking plants have stronger effects, followed by mid- merit plants while nuclear plants as baseload generation have a very weak but positive impact on the spot markets.
These findings corroborate the economic theory of power dispatch whereby the least expensive generations are dispatched first followed the next least expensive. Furthermore, being a very short- term market, influences of the variables appear to be directly related to their level of flexibility, hence one of the reasons peaking plants have a stronger impact on the half-hourly spot market price. Wind generation is shown to have a negative impact on the APX price, under both the autoregressive model and the static models. Lower prices, as a result of increased wind penetration, however, do not necessarily mean a fall in storage arbitrage value; this depends on the extent off-peak prices are reduced relative to peak prices.
Using the APX price under a 20 GW wind penetration level scenario, the co-optimisation model was run, with special precautions to isolate model error and wind impact. The results show that a slight increase in storage value arises as wind penetration level increases for all four years from 2011-2014. This increase in storage value occurs under both AR and ST simulated prices. The econometric regression in the BM shows that the most influential variables on the imbalance prices, in descending order of magnitude are OCGT, NIV, the APX price and Oil. Thus, similar to the APX, peaking plants with greater flexibility have a strong influence in the BM.
Similar to the case in the APX market, a 20 GW wind penetration scenario is simulated in the BM, whilst adjusting for additional effects; the BM which opens after the APX market closes is influenced by the latter and therefore this effect has to be accounted for. The NIV which cannot be assumed to remain fixed under such a scenario is adjusted by generating wind forecast errors from known distributions of wind error forecasts, one hour ahead. The third adjustment is derived from the econometric result and assumes that wind displaces generation in descending order of the merit order stack i.e. peaking plants, followed by mid-merit plants…and so on.
The results of the econometric models, both AR and ST show a fall in prices as with the APX market case. However, when the co-optimisation model is run, the impact in storage value is not clear; in some cases, under the AR model storage revenue falls slightly whereas prices generated under the ST model lead to a small increase in value, for all years.
Thus, there is a stronger tendency for storage value to increase slightly under higher wind penetration. This result, which may at first appear counter-intuitive occurs due to the fact that wind generation is uncorrelated with demand (Coker et al. 2013) as well as both the APX and imbalance price. A lack of correlation between the prices and wind generation implies that the falls in peak and off-peak prices, translate into gains and losses which cancel each other out, although not completely. This explains the small magnitude of the increase in storage value under the 20 GW wind scenario.
The slight increase in storage value could arise because consistently, there could be a slight tendency for wind to depress prices further during an off-peak period than during a peak period. The other, more likely explanation, is that the additional wind power causes prices to fall further such that new arbitrage opportunities become feasible (as price differentials are now larger).
Large swings in wind power can change the case for a short optimisation horizon, shown earlier in Chapter 5, in favour of a longer one. An investigation on how the value of optimisation horizons changed from 2011-2015, under changing market conditions, was carried out. No discernible difference was found either in the APX markets or BM. Mitigation effects of different types of renewable energy Coker (2011) and the low wind-price correlation coefficients explain the existence of weak effects of higher wind penetration for storage value in the markets.
This Chapter has thus shown that the absence of perfect foresight is detrimental to storage value whereas an increase in wind penetration level, although detrimental to prices can be beneficial to storage arbitrage value. The findings of Chapters 5,6 and 7 are further discussed in the following Chapter, including the implication and significance of these results.