Skip to content

Telling the Story of a Blackout

Dr John Constable, GWPF Energy Editor

There can be no doubt now that the UK’s electricity system is increasingly fragile. Will the consumer accept the extremely high costs of addressing renewable energy problems in order to secure this fragile electricity system? Or would they prefer a major policy rethink?

Figure 1. Annotated frequency trace describing events during the UK blackout on the 9th of August 2019. Source: National Grid ESO, Interim Report into the Low Frequency Demand Disconnection (LFDD) following Generator Trips and Frequency Excursion on 9 Aug 2019.

In an earlier post, this blog has noted the evident concern of National Grid ESO (Electricity System Operator) at the prospect of falling short-circuit levels in the grid. I took this as evidence of generally “Increasing Electricity System Fragility in the UK”. Low short-circuit levels were not, as far as we know, involved in the blackouts in the UK on the late afternoon of Friday, 9th of August, but there can be no doubt now that the UK system is indeed increasingly fragile. These blackouts disconnected 1.1 million customers, including several major railways and Newcastle Airport, and, because of its own internal protection systems, Ipswich Hospital. Roughly 5 per cent of the electrical load was lost, and while supplies for some customers were restored after fifteen minutes, others were without electricity for nearly an hour. The consequences continued for the rest of the day.

This was the most serious blackout in the UK for some time, and it was the first of any magnitude to occur since the advent of almost universal use of social media, meaning that the effects were reported with visual images and commentary in real time by those most affected, and quite independently of the news media. So long as their mobile phones were working those affected were no longer compelled to suffer in silence. This novel presentation led to a public perception of crisis that was to a degree not evident to National Grid ESO itself; and the company’s Director of Operations, Duncan Burt, will long be remembered for observing in his interview that the “systems worked really well”. He was, of course, patting himself on the back for avoiding a country-wide black system, but this was “mind-blind” tactlessness of the first rank. To smugly offer as corporate defence and general consolation the fact that the blackout could have been worse was bound to infuriate a public that had made it quite clear that they thought it shouldn’t have happened at all.

There are now three separate inquiries into the event, by National Grid, by the regulator, Ofgem, and also by the Department of Business, Energy and Industrial Strategy (BEIS), which has activated its Energy Emergencies Executive Committee (E3C) to undertake a review, which will report within twelve weeks.

The first of these, by National Grid itself, is to be composed of two parts, an interim report, and a detailed study. National Grid delivered the first instalment, Interim Report into the Low Frequency Demand Disconnection (LFDD) following Generator Trips and Frequency Excursion on 9 Aug 2019 to the regulator on the 16th of August, and Ofgem published this text on the 20th. The second, more detailed study is due on the 6th of September. While recognising that blackouts happen in milliseconds but need days of hard work to understand, this timetable does seem needlessly generous to National Grid and the generators involved. The public deserves prompt answers.

Furthermore, since much of the relevant data is held only by National Grid, it has been extremely difficult for external analysts to get a firm purchase on the matter; and this has allowed the interested parties, such as National Grid, the opportunity to control the discussion: an opportunity which they have exploited to maximum advantage, generally by dribbling information out, with the result that its impact has been diffused, and specifically by leading the public to conclude that the event was the result of highly unlikely coincidences, and thus by implication that there is nothing fundamentally or systematically at fault with the UK electricity network. The evidence suggests that neither of these is true.

But even with the meagre public data available from the Balancing Mechanism it was clear to some that National Grid ESO’s version of the events was not entirely persuasive. Readers of the Operational Forum reports, for example, would have been aware that the system has been experiencing what are known within the industry as “difficult days”, and Friday the 9th looked suspiciously like a difficult day that didn’t end happily. The system frequency trace, in particular, did not seem to be quite consistent with the narrative that the company was implying, a narrative that was in significant part driven by the clear desire to excuse wind power in general and the Hornsea Offshore wind farm in particular, from any significant blame in the event.

National Grid encouraged the view that the blackouts were the result of a simultaneous and highly unlikely fault at two power stations, Hornsea for one, and the other a Combined Cycle Gas Turbine (CCGT) at Little Barford on the border between Cambridge and Bedfordshire.

Doubts about the corporate story began to become public when the Guardian reported, on the 12th, that there had been “three blackout ‘near-misses’ in as many months before Friday’s major outage”. This information emerged as part of the ongoing sales campaign by Steve Shine of the battery storage company Anesco to persuade government to secure the system in the wake of the blackout by building more of his company’s equipment. Nonetheless, the information was correct.

National Grid had already submitted its report to Ofgem, but it had not yet been made public. In that hiatus it appears that National Grid leaked some of the key findings to both the Times and the Financial Times.

The Times published first, on the Friday (“Lightning strike and wind farm fault triggered blackout chaos”) in a story that disclosed for the first time to the public that the event was triggered by lightning, and supported the position earlier implied by National Grid that “The gas plant and wind farm failed within seconds of each other”, adding, however, that the strike was an event “which both should have been able to withstand”.

But not everything was going quite to plan. Simultaneously with the Times article on the 16th it was reported on the front page of the Daily Telegraph that on the night of Saturday and Sunday after the blackout there had been another system disturbance, with frequency dropping to just below the operational limits observed by National Grid, and that in response to this problem the Control Room had constrained Hornsea Offshore wind to reduce output, paying it £100,000 to do so, while at the same time requiring conventional generation to increase output. Far from being a one off, the problems appeared to be persisting, and it seemed likely that Hornsea was part of that problem, otherwise they would not have been paid handsomely to reduce output.

The Financial Times report on the 17th, “National Grid electricity blackout report points to failure at wind farm”, was explicitly based on a National Grid briefing and confirmed the lightning strike and added the additional information that a considerable quantity of “embedded generation”, that is to say generation connected to the low voltage distribution network and thus invisible to the Control Room, had tripped as a result of the lightning. The report also observed that the blackout was “caused by the world’s largest offshore wind farm accidentally going offline”, though Orsted, the owners of Hornsea were quoted as declining to confirm that they had tripped before Little Barford. If the mist was thinning in one quarter of the horizon it was becoming still thicker in another.

National Grid would have been reasonably content with the situation; the impact of the blackout story was being weakened by incomplete delivery, and there was much confused and incomplete comment from all sides, undermining public confidence in any particular conclusion offered. Faced with the inability to dominate public discussion, National Grid’s slow release policy had succeeded in making that discussion incoherent and of low credibility.

But things were not going forward entirely smoothly. On Sunday it was reported, again in the Telegraph, that Colin Gibson, a former Power Networks Director at National Grid, a main board position at that time with personal responsibility for keeping the lights on, had, with his former colleague Dr Capell Aris, raised concerns that National Grid had permitted hazardous decline in system inertia resulting from high levels of asynchronous generation, and pointed to this as the underlying cause of the weaknesses leading to the blackout (“Former National Grid director says ministers should impose limits to new wind and solar farms to help avoid power cuts”). Gibson and Aris added that government would need to call a halt to further installation of both wind and solar. Both are well known in the power networks field. This was not convenient for National Grid, and a company spokesman responded with the extraordinary remark that “There is no evidence at this stage that the power cut is related to wind or other renewables”, an observation scarcely compatible with the information in the FT’s report of the previous day.

On Monday the 19th, the so far helpful Times published a piece if anything still more awkward than Gibson and Aris’ criticism. This article reported that National Grid was actively limiting the utilisation of interconnectors with the continental networks in order to reduce the amount of asynchronous input and maintain inertia: “Blackout fears over National Grid cables from the Continent Company limiting use to guard against failure”.

The information in this article was attributed to a “senior National Grid source”, but may not have been part of the corporate public relations approach since it clearly tended to confirm the Gibson/Aris diagnosis of problematically low system inertia, and put into a more robust context the  fact of constraint payments to Hornsea on the night of the 10th and morning of the 11th, as well as the “near misses” reported by the Guardian. One suspects a whistleblower, or an inadvertent disclosure.

On Tuesday the 20th, over a week after the blackout itself, Ofgem released National Grid’s own analysis into this noisy and confused discussion. The study, a key figure from which is reproduced above, is entirely consistent with the tactical approach taken in the preceding week, in that it adds new information but raises as many questions as it answers. Needless to add, this is only an interim statement that does not permit conclusive analysis. We are kept guessing.

But the additional information provided is important and when combined with the perspectives that have emerged from quarters other than National Grid during the last week, a clearer picture does in fact emerge, and on the basis of the chart above, and the crucial timeline on pages 11 and 12, can be summarised as follows:

Overall ‘interim’ conclusion:

* The blackout was the result of a single phenomenon, not a rare coincidence of uncorrelated problems, as National Grid had intimated throughout the week. The single phenomenon consisted of a lightning strike on a grid line, which affected voltage and caused 500 MW of embedded generation and the Hornsea Offshore wind farm to trip because they could not ride through the fault. Little Barford CCGT subsequently tripped, probably as a result of the combined effects of the preceding events.

Further details

* The lightning strike precipitated the tripping of some 500 MW of embedded generation, which National Grid’s interim report identifies as solar and small diesel. This is likely to be an error, since there is a great deal of embedded wind, particularly in the relevant geographical area, and it seems extremely likely that there was a substantial perhaps preponderant wind energy component in the embedded capacity that disconnected. The exact timing of the embedded generation trip is not at present clear from the reports, and this is a matter that may become clearer in the detailed technical study. The trip of so large a total capacity of embedded generation suggests that the fault ride through settings on embedded generation are insufficient and must be tightened.

* It is probable though not yet certain that Hornsea and the embedded generation tripped together, and both in response to the lightning strike.

* It is now certain that Hornsea tripped at (hrs, min. seconds, milliseconds), and 165 milliseconds before the Little Barford CCGT began to trip.

* Hornsea “deloaded” very rapidly, in effect instantaneously, with output falling from 800 MW to 62 MW in 197 milliseconds. This trip suggests that the fault ride-through capability at this windfarm was not sufficient.

* Little Barford tripped in three stages: Stage One: At 16.52.34, the Steam Turbine element of the CCGT tripped, an instantaneous loss of 244 MW, a trip that made it inevitable that the two gas turbines would also now have to trip. Stage Two: At 16.53.31, the first gas turbine tripped, a loss of 210 MW. Stage Three: At 16.53.58 the second gas turbine tripped, a loss of 187 MW.

* It is not yet quite clear why Little Barford’s steam turbine tripped, and doubtless this will become clearer in the technical report.

So some progress in understanding has been made, but the interim report still leaves a good deal of mystery around the overall event, for example regarding National Grid’s preparation for frequency excursions of this kind. Was there sufficient frequency response? Was the recovery slow, or satisfactorily prompt in the circumstances. What was the system inertia at the time of the lightning strike and during the fault? Was the scale of the load shedding involved proportional to the problem, or was it excessive? Was the load shedding correctly targeted? Is there a general fault ride through problem with embedded generation, and with large scale renewables such as Hornsea offshore wind in the UK? What was the composition of the embedded generation that tripped. Why did the Little Barford steam turbine disconnect?

But over all these doubts hangs the larger and now strongly indicated problem of general system fragility as a direct consequence of renewable energy policies. Applying sticking plasters, some of Mr Shine’s batteries for example, will be very expensive, further degrading the productivity of the electricity supply industry. Will the consumer accept the extremely high costs of addressing the problems attendant on renewables so as to deliver a secure system? Or would they prefer a major policy rethink?

John Constable: GWPF Energy Editor