Tom Edwards, Senior Modeller at Cornwall Insight, discusses what published industry data tells us about the blackout of 9 August. This was originally published on 19 August 2019 in Energy Spectrum.
The Balancing Mechanism Reporting Service (BMRS) is the primary channel for operational data on electricity balancing and settlement arrangements. Under the EU REMIT directive, generators are also obliged to disclose when there are major changes to operational status. Piecing together information from both sources sheds some light on the events of that Friday evening.
Information from BMRS shows that around 16:54 BST on 9 August there was a dramatic fall in system frequency to below the statutory minimum of 49.5Hz. Some demand needed to be shed to maintain the integrity of the national network. Overall the frequency was outside of the lower statutory boundary of 49.5Hz for 135 seconds and remained outside the boundary of the targeted lower operational limit (49.8Hz) for an additional 90 seconds.
Elegy for falling frequency
According to REMIT data, Little Barford CCGT experienced an unplanned outage affecting 664MW of capacity at 16:57:40 (BST). This was followed by two REMIT notifications, starting from 17:00 BST, from Hornsea 2 and Hornsea 3 affecting 812MW – although according to physical notification data the loss was 754MW.
The BMRS data suggest an initial generator trip caused the frequency to fall rapidly but then stabilise at around 49.2Hz within 60 seconds. It has been speculated the second dip in frequency was caused by another generator. But media reports state that National Grid ESO has said it is treating the incident as one single event rather than two separate incidents.
The data presently available to the market is on a 15 second resolution and this might mask some volatility within the frequency that is not currently apparent. National Grid told The Times the data presented here is not correct, so it is currently not clear what the immediate cause of sequence of events is leading up to and during the blackout.
It is worth noting the Grid Code specifies generators need to cope with frequency deviations down to 47Hz for 20 seconds so Hornsea or Little Barford should be capable of remaining connected to the system despite a reduction in frequency.
Whatever the data says, frequency must have deviated below 48.8Hz because at this point the Low Frequency Demand Disconnection (LFDD) relays cut in as expected and shed 5% of the NGET area demand. This level is also set by the Grid Code (OC6) and each distribution network operator must set its system to lose a given level of demand below this frequency level. It is the DNOs’ obligation to ensure this loss of demand is evenly distributed across their networks.
Estimated inertia over the period was around 180GVA, above the 100GVA minimum (see Figure 1). Below this level the ESO would have to curtail wind farms and increase the active power output of synchronous generators to make sure the system was safe to operate. It has since been suggested the trip was caused by a lightning strike, although the system is meant to be protected against this event, which occur thousands of time a year.
Imbalance prices during the blackout periods were £58.81/MWh and £64.41/MWh, much lower than many expected for a period including a LFDD event. This indicates how well supplied the market was. But at this stage the prices do not incorporate the actions taken by the ESO as they have not yet fed through into the imbalance price calculations because of delays in adding in Disaggregated Balancing Services Data.
The Demand Control Volume – the volume of electricity not supplied from the transmission system during the period of disruption – has been determined at 931MWh between 16:54 and 17:40. It will be added into the imbalance price calculation process at the Value of Lost Load (£6,000/MWh). It will not directly set imbalance prices as it will be flagged as a system action but it could indirectly influence them through changing the Net Imbalance Volume (NIV).
The most significantly affected parties from the increase in imbalance prices will be those short over the periods, likely to include Little Barford and Hornsea. They will have been unable to trade for at least the start of the blackout period due to deadlines for reporting trades.
At this stage we believe renewables can take heart from the event. Even though a wind farm was one of the plants involved in the blackout, it seems renewables themselves were not responsible. The data supports this view. Inertia on the system was above the minimum levels required. The combined 1.4GW loss was above the level of the capacity the system operator is mandated to hold in reserve.
But it is undeniable that frequency has become more volatile, the distribution of frequency levels within the operational limits has become wider, a reflection of the increased variation of the system which is driven by renewables (see Figure 2). This has caused battery operators to import and export more power through their systems to provide response than originally expected and increased costs of Frequency Response provision.
There are several steps that could be taken to reduce the impact of these events. Some potential ideas in approximate order from cheapest to most expensive are to:
- change settings on the system to reduce the risk of high Rate of Change of Frequency (RoCoF) and voltage vector shift. This is already being done as part of the DC0079 change to reduce the effect of volatility on the loss of mains protection setting for embedded generators.
- change the settings on the LFDD relays, a study should be undertaken to see which load elements are covered by the current relays and what a more rational system which prioritised certain loads (such as medical and transport loads) could be achieved with greater transparency of the distribution networks
- incorporate wind into balancing services at full technical capabilities. Manufacturers are offering synthetic inertia capabilities to mimic the behaviour of turbines at conventional power stations. The ESO may also be able to do more to hold frequency response and reserve on wind farms. With the new Power Available signal the ESO has a real-time signal of the expected power output from a wind farm and could constrain output, to ramp up later and feedback into the system, and
- more, fast-acting, frequency response. The largest loss volume will grow as the Norwegian interconnector (1.4GW) and Hinkley Point C are completed (1.6GW). This will increase capacity held against the single largest failure. More renewables are also inevitable. The system will rely more on storage to provide fast response and alleviate supply and demand imbalance. Buying Enhanced Frequency Response cost £66mn in 2016. To buy the full 1.4GW loss experienced on that Friday would cost around £400mn over four years.
If you have enjoyed reading Tom’s views and want to read about the latest developments in the energy market and more, please contact us for a month’s trial of Energy Spectrum.
Energy Spectrum is a weekly publication covering all of the key policy, regulatory, market and transactional developments across the energy sector. It offers a timely, insight-driven overview of the need-to-know news and changes in the energy sector. Contact Nick Palmer to request a trial on 01603 604400.