Frequency Response: The Past, The Present and The Future
The UK electricity is an Alternating Current supply at about 230 volts with a frequency of 50Hz (which means it changes direction and back again 50 times a second). Consequently, National Grid ESO (NGESO) has a licence obligation to control system frequency at 50Hz plus or minus 1%. The system frequency is a continuously changing variable that is determined and controlled by the real time balance between system demand and total generation. If electricity generation outweighs demand, the electrical frequency increases; if there’s more demand than electricity being produced, the frequency falls. If the frequency goes too far from its normal value of 50Hz, blackouts or equipment damage can result. NGESO buys frequency response services to make sure this doesn’t happen.
National Grid purchases frequency response from a range of sources and mechanisms. Such as, generators in the Balancing Mechanism (MFR), batteries through Enhanced Frequency Response, and monthly and weekly tenders for Firm Frequency Response. However, last year, NGESO released a new faster acting frequency response product, Dynamic Containment (DC). DC is designed to operate post-fault, i.e. for deployment after a significant frequency deviation usually resulting from major loss in generation. Increasingly, faster acting frequency products are needed because system frequency is moving away from 50Hz more rapidly. This is a consequence of imbalances and lower inbuilt system inertia as conventional thermal generation comes offline and increasing volumes of intermittent renewables are connected (see previous blog on the evolution of frequency).
Battery owners, such as Gresham House and Gore Street Capital presently see frequency response services as their main revenue income. In 2020 alone, 76 per cent of Gresham House Energy Storage's revenue came from frequency response, 10 per cent from trading, 9 per cent from triads and 5 per cent from capacity market contracts. These figures cite the importance of ancillary services for fund managers.
With clearing prices just under £17 / MW / hour, the value of DC is approximately double the volume weighted average value of the products it is set to replace. The introduction of DC has left a shortfall (Fig. 1) in the NGESO’s demand for frequency products versus the supply in the market and therefore, prices have rebounded slightly in previously saturated markets (Fig. 2).
Figure 1. Volumes NGESO aims to procure in the monthly DFFR/SFFR auctions, weekly DLH/LFS auctions, Enhanced Frequency Response and Dynamic Containment verses the size of the market for storage.
Figure 2. Pay-as-bid monthly DFFR auction prices. In the winter months, exceptionally high prices can be achieved by front-of-meter storage; the marginal price being set for Triad chasing hours (the weekday afternoon periods of highest demand on the grid between November and February each year). The orange line reflects the current market cap for Dynamic Containment.
Despite the attractive revenues, uptake has been gradual owing to the stringent technical requirements. DC service requires hourly Performance Reporting data submission to NGESO measured at 20Hz resolution in addition to secondly Operational Metering data that is provided to the ESO control room. In spite of the stringent requirements, DC provides little or no response, unless frequency exceeds 49.8Hz and therefore energy throughput is nearly seven times lower than DFFR. Consequently, the value of the service is currently based on the technical barriers to entry, the low supply of assets and the cost of alternative actions available to National Grid. NGESO published their forecast requirement for DC for the entirety of 2021, and current volumes make up only 50% of the 1.4 GWs required for this summer (Fig. 3).
Figure 3. Average weekly volumes in the DC auction since October 1s 2020 compared to the maximum and minimum volumes NGESO is aiming to procure.
From August, there will be the following changes to DC:
1. Moving from pay-as-bid to pay-as-clear – minimising the need for forecasting auction results.
2. 24 hr contracts to EFA block procurement - sending more accurate market price signals.
3. Manual validation to automated validation - immediate feedback if an order fails validation.
Change 2 will allow for more real-time market optimisation as currently providers must choose between wholesale energy trading for 24 hours or DC. In the second week of January, volumes in DC decreased on the days when the system price was high (Fig. 4). For example, on the 8th of January 256 MW of DC was accepted, whilst 124 MW was rejected/withdrawn as the system price reached 4000/MWh at 7pm. Consequently, a 1MW participating in DC on the 8th of January would have made £408, compared to an asset NIV chasing which could have made five times more (£2000). For context, two days later 371 MW of response was accepted in the DC auction and the system price was below £150/MWh. This highlights the importance of system forecasting and optimising across markets in as close to real-time as possible.
Figure 4. Daily DC accepted and rejected/withdrawn volumes due to signals in other markets. High system prices on January 8th resulted in over 100 MW more being unavailable for DC compared to the following day.
Later this year the final two fast acting frequency response products, Dynamic Moderation (DM) and Dynamic Regulation (DR) will go live. DM will be used to help manage sudden large imbalances between demand and generation by responding quickly when frequency moves towards the edge of the operational range. DR is a pre-fault service designed to slowly correct continuous but small deviations in frequency. So far, these two new products appear to require a more energy intensive response compared to DC and also DFFR (Fig. 5).
Figure 5. Response curve for current frequency products, DC and DFFR. The response curves for DR and DM are based on information currently available.
Looking to 2022 and beyond, the risk of future large losses is likely to increase due to the threat of single, unscheduled outages posed by very large assets listed to be connected, including two 1.4 GW interconnectors (NSL scheduled for Dec. 2021, Viking Link, 2023). Also of concern is the growing generation capacity in the south of Scotland, Hinkley Point C; and large offshore wind farms (1.4 GW Hornsea Two). Consequently, the need for faster acting response services will continue to outstrip supply.
Currently, there is approximately 1.2 GW of battery storage in the UK, with the majority coming online in the past 4 years (Fig. 6). Figure 6 illustrates the value of new markets providing strong revenue signals. In 2017/2018, 200 MW of storage came online with EFR contracts, and the majority of these assets were limited in duration (1 hour or under). Likewise, the introduction of DC and the move to slightly longer duration storage systems has driven more recent investments. Nevertheless, the average growth rate for storage over the last 4 years was 284 MW / yr and continuing at this rate, would give just under 4 GW of storage by 2030.
Figure 6. Annual battery energy storage deployments over the last 10 years categorised by duration of the asset. In 2017-2018 the majority of assets were of a duration less than 60 minutes. Note this chart does not include 100MW of BESS in Northern Ireland DS3 market (Tandragee and a 6 minute duration system at Drumkee). Both assets are owned by Gore Street Capital.
The UK currently has a healthy battery storage pipeline of 16 GW. This is likely to be driven by the shift in legislation to allow projects of over 50MW capacity (in England and 350 MW in Wales) to bypass the Nationally Significant Infrastructure Project planning status that larger projects previously had to attain, falling cell costs and the attraction of new revenue streams. However, to meet NGESOs demand, more technologies need access to ancillary service markets. In particular, this is vital for those that are smaller (sub-1MW) and require aggregation across multiple GSPs to meet the current 1MW limit threshold.
 SFFR energy throughput is approximately around 0MWh, DFFR is 30MWh and DC is 4MWh, calculated using the response curves and historic frequency data.  The cost of alternative actions refers to what it would cost NGESO to reduce the Rate-of-change-of-Frequency risk through alternative actions such as bidding off/down large generation units or reducing high levels of interconnector flows if there is insufficient inertia in the system.