22 February 2021
January’s record high electricity prices indicates an interesting time for the GB power market. Here we investigate the key drivers behind this price surge.
Power markets have been making headlines recently. Tight margins in GB caused power prices to sky rocket to record highs in January, with the Day Ahead (DA) auctions on both N2EX and EPEX clearing at the price cap of £1,500/MWh and Bid Offer Acceptances (BOAs) reaching £4,000/MWh on several occasions. Notably, however, there were no loss of load events. Meanwhile, Texas has suffered some of the worst blackouts in recent memory due to unprecedented weather conditions causing the Electric Reliability Council of Texas (ERCOT) to implement rolling outages across the state.
The GB and Texas markets are of a similar size and have similar levels of renewable penetration. But a key difference is that GB uses a Capacity Market (CM) to ensure there is enough supply to meet demand over the winter peaks, while Texas’ does not.
The GB CM uses a security of supply metric known as Loss of Load Expectation (LOLE), targeting 3 hours per year, to procure a required amount of firm capacity to meet peak demand. The calculation underpinning this is based on creating expected distributions around uncertain events, such as wind generation, demand volatility or plant outages, and converting a target security standard (3 hours) into a capacity target.
There have been a range of responses to the high prices in GB, with criticism aimed towards the CM for not safeguarding against these prices, towards renewables for under delivering (an argument seen on both sides of the pond) and Brexit being blamed for inefficiency over the interconnectors.
In this blog, we explore how January’s outturns across wind, demand and plant unavailability fared against the expected distributions, what the main drivers of the price spikes were, and what this means for the future.
What drives power prices?
Suppliers buy power to meet the demand of their customers, with the prices they pay being largely dependent on what the marginal generating unit is. In summer, lower demand means power prices tend to be lower, while in winter higher demand means that higher cost sources of generation will need to be used to provide power.
The main driver of demand and the amount of renewable generation being produced is the weather, and this will have even more of an impact as we decarbonise the energy system. Commodity prices also play a large role in setting the power price. Thermal power stations (which have input fuels such as gas and coal) will need to cover the cost of burning these fuels plus the associated carbon costs. These commodity costs vary and feed through to the power price.
What happened in January?
Power prices exceeded £1,000/MWh on multiple occasions in January, with the market cap for the DA auction (£1,500MWh) reached twice. The previous record price in this auction was in September 2016, when prices hit £999/MWh. Along with high prices in the wholesale market we also saw a month of high prices being taken in the Balancing Market with West Burton B CCGT being accepted at £4,000MWh across several periods.
What was driving these high prices?
The three largest uncertain factors in driving high prices are wind speeds, demand (largely driven by temperature) and plant availability. In this section, we consider each of these in turn and explore how January 2021’s outturns fared against the expected distribution of each of these factors.
The chart below shows the distribution of expected wind load factors across the peak period of the day in January, based on observed historical load factors. Overlaid on this chart are the days in January 2021, coloured by their peak price.
As expected, the higher price days are concentrated on days with lower wind load factors, where margins are tighter and more expensive thermal units are needed. However, as shown by the distribution of days along the curve, January 2021, though having lower wind generation than average, was not an extreme case. The variation observed across the month was broadly as expected, there were no extremely low wind events (< 10% fleet load factor), and the highest price days were not exclusively the lowest wind days.
Performing the same analysis on demand, we see a similar picture.
The chart below shows the expected distribution of demand peaks (scaled as a percentage of expected peak demand) for January based on observed historical data. Again, overlaid on this distribution are the days of January 2021.
As expected, during high demand periods we see higher prices. However, again the peak demand levels are not unusually high for January, and the distribution of demand along the expected curve is broadly as expected. Therefore, demand levels alone do not explain the extreme prices seen last month.
Plant may be unavailable due to either planned or unplanned outages. Planned outages, such as those for maintenance or refuelling, will typically be scheduled over the summer period when demand and prices are low. Unplanned outages are the opposite and occur when a unit has unforeseen issues which can see it taken offline for hours, weeks or even months.
The graph below shows the distribution of expected availability across the fleet in a January, and plots the availability during the peak hours in each January 2021 day against this distribution.
It is immediately clear how the non-availability of plant has been a key driver in the high prices, with all days in the lowest percentile of expected availability.
Looking into the reasons for the low levels of plant availability, we can first look at “long-term” factors which lead to a significant amount of capacity being unavailable throughout the whole of winter 2020/21. There were three key factors here:
- Long term nuclear outages: The GB nuclear fleet is ageing and there have been several long-term outages including Dungeness B and Hinkley Point B. This meant that 3.3GW of capacity was unavailable in January.
- Company administration: Calon fell into administration last year with its three gas plant being mothballed while a new buyer is found. This saw 2.3GW of plant unavailable in January.
- Difference between Maximum Export Limit (MEL) and CM capacity: This is the difference between the capacity observed on the system and the capacity used in CM contracts, and resulted in a further 1.9GW unavailable in January. It is worth noting that NGESO take account of this difference when calculating the CM target.
* excluding short-term outages
Accounting for the long-term missing capacity, the availability of the remainder of the fleet shows a better fit to the expected distribution (first chart below). The distribution is still skewed to the lower end, and this is predominately driven by the relative unavailability of the coal fleet (second chart below), with the coal fleet operating at an average availability of 52% vs their CM derating factor of 87%. The age of the fleet, and the fact that the plant is nearing retirement (with all coal plant being required to close by October 2024), may explain these low levels of availability.
In addition to all of the above, BritNed, a 1GW interconnector connecting Britain and the Netherlands, suffered an unplanned outage on the 8th of December and was unavailable throughout January, only returning on the 9th February.
Although low wind and high demand both contributed to the high price events (and wind generation was lower than average across the month), our analysis shows that wind and demand conditions were within the range that we’d expect to see in a typical January. The main cause of the high prices was the unavailability of a number of plant, largely from plant nearing the end of their life (nuclear and coal), or plant that were unavailable due to events not accounted for in the CM calculation (Calon going into administration).
So does this mean these high price events were an anomaly or are they the “new normal”?
Looking at the CM, there may be a temptation to re-examine some of the parameters, such as the assumed availability from older plant. The recent announcement to increase the capacity requirement for the upcoming T-1 auction (for delivery in winter 2021/22) is an example of this already playing out. The requirement has been increased from 0.4GW to 2.4GW, though 1.85GW of this is to replace contracted plant who are known to not be delivering in 2021/22. The increase also includes an adjustment for the risk of non-delivery from other contracted plant and the risk of unavailability from coal and biomass conversion plant. This additional capacity should somewhat dampen peak prices in winter 2021/22.
Looking beyond next winter, the older coal and nuclear units that contributed to the high prices will soon come off the system. A greater and greater proportion of generation will come from intermittent renewables, which will mean more variation from one year to the next as weather patterns, particularly wind, drive system margins. The remaining thermal generation and new sources of flexible generation will increasingly need to recover costs over relatively few and infrequent periods (or through higher CM prices). This points to a power system where high prices such as those seen this winter could become more common.
How common will obviously depend on several factors, but a key one will be the level of prudence in the CM parameters, and how close the system actually gets to the LOLE standard of 3 hours. It is worth keeping in mind that despite the high prices and tight margins, this winter has (so far) not seen any actual system stress events. If we were to experience a winter with the 3 loss of load hours that the CM security standard targets, then we would likely see as high or even higher prices than those observed this winter.
One thing that is for sure, whether you believe that this winter’s high prices represent a market failure that requires attention, or a market providing the appropriate signals to incentivise flexible generation, these are interesting times for the GB power market.