The UK’s August 9 blackout: Why did it happen and what can we learn?


The power outages that occurred in the United Kingdom on August 9 demonstrated our increasing dependence on secure electricity supplies and the extremely disruptive consequences when those supplies fail. The reasons for the extent of the disruption to transport and other critical infrastructure are beyond the scope of this blog. However, an examination of the events leading up to the supply failure provides useful insights into how decarbonisation is changing our electricity system, with conventional generation increasingly being replaced by wind and solar and the gradual transfer of generation capacity from the transmission to the distribution systems. As the nature of the electricity system changes, the way it is operated and regulated will also need to change.

What happened?

National Grid’s final report on the events of August 9 shows that the disruption resulted from the cumulative loss of most of the Hornsea wind farm (737 MW), the total loss of the Little Barford combined cycle gas turbine power station (641 MW) and the loss of around 500 MW of small distribution-connected generation, presumably mostly solar, wind and some reciprocating plant. In total, some 1,880 MW of output was lost, far greater than the 1,000 MW loss National Grid was required to cover at the time,  resulting in the automatic disconnection of around 3% of system demand. All three generation losses were associated with lightning strikes that tripped a high-voltage transmission line. There were no capacity shortages at the time.

What role did renewables play?

The events of August 9 were truly exceptional and there should be no knee-jerk reaction.

August 9 was windy. In fact, wind met 50% of demand — a record for the UK — earlier in the day. High levels of wind or solar output cause inertia levels to fall, with system frequency becoming more sensitive to sudden changes in the supply-demand balance. However, National Grid had ensured adequate levels of inertia to cover the largest expected loss of infeed on the day, and therefore the relatively high level of renewable output was not the primary cause of the incident. The loss of the Little Barford gas-fired plant contributed as much to the supply failure as did the loss of Hornsea, while several other offshore wind farms electrically closer to the initial fault were unaffected. It is also worth noting that a transmission fault caused by a lightning strike (or anything else) should not have resulted in the loss of either Hornsea or Little Barford. In that regard, the coincident loss of both was truly exceptional.

The role of distribution-connected generation

While the deployment of renewables was not the primary cause of the incident, the tripping of so much distribution-connected generation, much of it renewable, clearly did contribute. Distribution-connected or “embedded” generation is equipped with “loss of mains” protection designed to operate in the event of a failure of the local grid or its disconnection from the remainder of the network. Since the local grid operated just fine on August 9, the loss of local embedded generation capacity seems to have been entirely unnecessary.

As the nature of the electricity system changes, the way it is operated and regulated will also need to change.

The risk of loss of mains protection operating unnecessarily has been known for many years and there have been several instances of embedded generation losses following transmission faults. Consequently, a programme of de-sensitising this protection was initiated in 2014 with smaller installations having until 2022 to complete the work. The relaxed timescale associated with the programme is clearly relevant and, given the events of August 9, quite remarkable. Analysis of the sequence of losses that occurred on the day suggest that, had the 500 MW of embedded generation not tripped, then what turned out to be a disruptive event impacting over 1 million customers may have been relegated to a frequency excursion that few customers would have noticed.

What lessons do we need to learn?

Generation failures are a regular occurrence, and, like buses, they sometimes annoyingly all arrive at the same time. However, the fact that a large amount of embedded generation unnecessarily tripped is clearly a concern. As we continue to decarbonise our electricity system and the plant mix changes, more and more generation capacity will be embedded. This embedded, decentralised capacity has the potential to enhance the resilience of the grid but, sadly, at the moment it is needlessly reducing resilience. It is quite remarkable, given the known risk posed by loss of mains protection and concerns raised by National Grid about that risk, that more urgent action has not been taken to resolve the issue. One of National Grid’s sensible recommendations is that the programme timescales for de-sensitising the loss of mains protection be revisited and, hopefully, Ofgem will shut that particular door before any more horses have the opportunity to bolt.

In the longer term, the whole issue of how embedded generation should be protected needs to be revisited. As suggested above, decentralised embedded generation has the potential to enhance the grid’s resilience to severe events. However, for this potential to be realised, the current underlying philosophy of shutting down all embedded generation on loss of mains until the grid is restored needs to be replaced with something more appropriate for the future.

In the longer term, the whole issue of how embedded generation should be protected needs to be revisited.

It is a concern that National Grid have only been able to “estimate” the amount of embedded generation that tripped offline based on the increase in demand that occurred during the incident. In fact, it is still not clear exactly how much embedded generation actually tripped. With embedded generation set to contribute increasingly to meeting demand, its visibility to National Grid is unacceptably low, with too little information available to adequately assess the considerable uncertainties around its operation. As recommended by the Capacity Market Panel of Technical Experts, the distribution system operators should establish a complete register of embedded generation capacity. This would allow National Grid to more accurately forecast demand and predict the response of that embedded capacity to a range of operational circumstances.

Finally, it appears from their final report that National Grid’s actions before and during the events of August 9 were in accordance with the current requirements of the Standards of Quality and Security of Supply (SQSS), and that the transmission system responded to those events in the way it was designed to do. However, replacement of conventional generation with wind and solar will mean the loss of traditional sources of system inertia. Alternative sources have and are being developed, however it seems appropriate that existing arrangements for maintaining the resilience of the system to extreme events should be revisited. The SQSS have their origins in the middle of the last century and National Grid’s recommendation that the SQSS be revisited from the perspective of resilience, is sensible. The events of August 9 were truly exceptional and there should be no knee-jerk reaction. However, the energy mix and characteristics of the electricity network are changing significantly and the SQSS need to reflect the new realities.