Power cut: Five big questions
By Richard Black, Senior Associate @_richardblack
Published:20 August 2019
Following publication of National Grid's initial report into the power cut that took out rush-hour trains and about a million homes on 9th August, we're a little clearer as to what the causes weren't. As to what precisely did happen, and why... well, several questions remain.
We know the power cut wasn't an inevitable consequence of Britain's drive to renewable energy - that was never credible as a sole cause, but for a few sections of the media, credibility has been a secondary concern.
We know that lack of generating capacity isn't an issue: national demand was barely above 30GW at the time, three-fifths of the annual peak, and Grid reports that there was more than 4GW of spare generation waiting to be called up. We know that the responses of the regional distribution network operating companies (DNOs) and train systems were at least as important as the failure in generators.
Professor Tim Green of Imperial College London has written an excellent dissection of National Grid's report, so I'm not going to clutter up the Internet by repeating the exercise. And the report itself is pretty clear. Instead, here are five of the questions that haven't yet been answered.
1) If it was all sparked by lightning - then, how?
The first event that the Grid report mentions is a lightning strike on a high-voltage power line near Cambridge, producing a partial short circuit. Milliseconds later, the line disconnects, as it should. Milliseconds after that come a disconnection by two units of the Hornsea One wind farm and the automatic shutdown of one turbine, the steam-powered one, at Little Barford gas-fired power station.
At the same time, small generators connected to regional distribution networks (in which regions isn't yet clear, but probably the south-east) are also protectively shutting themselves down.
The report says that the Hornsea and Little Barford failures happened 'independently of one another - but each associated with the lightning strike.' But is this correct?
RWE, the owner of Little Barford, hints it's not proven.
And if Grid is right, what's the link from lightning to plant failures?
There were dozens of lightning strikes that day, in the UK and neighbouring countries - why did this one cause the outages when no others did, especially as the national high-voltage network did exactly what it was supposed to?
But if the Little Barford and Hornsea outages weren't related to the lightning strike, then we're talking about two big generators failing within seconds of each other without apparent cause - which isn't impossible, but highly unlikely. It seems probable therefore that the lightning was the spark... but the hows and whys remain less than clear.
2) Is the switch to renewables implicated at all?
One canard we can easily slay is the claim that this was all about 'unreliable' wind. The Hornsea One disconnection was a function of its automated control system: as the report puts it, 'as the reaction [to the voltage spike produced by the lightning strike] expanded throughout the plant, the protective safety systems activated.'
So, nothing to do with the fact that the energy comes from rotating wind turbine blades rather than rotating gas turbine blades.
The more interesting question concerns inertia and the stability of the system when something goes wrong.
Basically, when something feeding electricity into the grid trips out, the frequency of the grid falls. Generators that rotate - and indeed anything else connected to the grid that rotates - take time to slow down because they hold kinetic energy in a rotating mass, and this reduces the rate at which the frequency falls. This gives longer for backup systems to kick in, and also reduces the risk that other generators will sense what looks like a fault and shut themselves down.
Nowadays we have a higher proportion of generators without inertia on the grid: wind, solar and interconnectors. At the time of the 9th August power cut, these sources were providing 40-50% of our electricity.
Initially the grid frequency fell at a rate of 0.16Hz per second - faster than the threshold of 0.125 Hz/s at which many smaller generators automatically disconnect themselves. Hence, they did.
Would this frequency fall have been slower under an old-fashioned system where all of the electricity came from generators with inertia? Yes.
Would that have made enough difference to avoid the tripping out? Not sure.
In the UK system, about 30% of inertia comes from equipment connected to the regional distribution networks run by distribution network operating companies (DNOs) - and that share would be pretty much unchanged. In addition, some of the services National Grid has procured, for example battery storage, make up for reduced inertia by acting faster than old-fashioned back-up services could have done. Grid acknowledges the issue, but says it has it under control.
Grid has to ensure there is enough fast-acting reserve ready such that if the biggest source of electricity coming into the network at any given time fails, the fall in electrical frequency is manageable - it doesn't go outside the boundaries set by the regulator. At 4.50pm on Friday 9th August, it held 1GW in reserve.
National Grid could be required to hold more fast-acting capacity in reserve. But extra security costs money. As ECIU showed a while back, over three recent winters Grid spent £180 million on back-up generation that was never needed - a sum added, in the end, to people's energy bills.
Just as you can never have an entirely safe car, the idea of an impregnable grid is fantasy. The UK system is very reliable already - making it more reliable still is really about spending more money for an incremental gain.
There are two other caveats on this inertia argument. One is that automatic disconnection when the rate of change exceeds 0.125Hz/s is only supposed to apply to generation connected to distribution networks, not those attached to the National Grid high-voltage network - so would not have affected Hornsea One or Little Barford.
The other is that this problem is being rectified. Managers have realised that the 0.125Hz/s threshold is way more sensitive than it needs to be. Equipment is being reconfigured around the country in a programme that will be complete in 2022, setting a new threshold of 1Hz/s. If reduced inertia was a problem, it'll stop being one within the next three years.
3) Could Little Barford have prevented the outage?
As I noted above, all of Grid's frequency reserves kicked in very quickly after the initial loss of generation.
And it was almost enough.
After the initial spate of failures – with most of Hornsea One offline, Little Barford's steam turbine stationary and approximately 500MW of small-scale generation lost from distribution grids up and down the country – the system was almost stable. Grid frequency hovered around 49.2Hz - lower than the targeted band of 49.5-50.5, but high enough not to need any customer disconnections - for over half a minute.
Then, a minute after the lightning strike, the first of Little Barford's gas-fired turbines shut off. Only then did the frequency collapse far enough that the first households, businesses and railway signals lost power.
There are several questions here.
Firstly, why did the steam generator (which runs on steam heated by exhaust from the two gas-fired turbines) trip out so quickly? RWE told Grid it was because its control systems detected a 'discrepancy between the three speed signals' (coming from the three generators). But it isn't supposed to happen: as the report notes, the Grid's code says of generating stations that when the frequency is between 47.5Hz and 49.0Hz, 'Operation for a period of at least 90 minutes is required'.
The first gas-fired turbine shutdown, a minute after the steam generator tripped, was triggered by 'high steam pressure'; and a further 30 seconds later, operators turned off the second gas-fired turbine for the same reason. Two questions here:
- within those 60/90 seconds, couldn't operators, or the automatic control system, have turned the steam turbine back on, given that the 'fault' detected clearly wasn't real?
- given that the station is equipped with a steam bypass system (referenced in National Grid's report), couldn't that system have released pressure, thus enabling one or both of the gas turbines to keep operating?
One presumes that Ofgem will be asking these questions of RWE just as it's asking Ørsted questions about Hornsea One, because... if two of the three turbines had kept spinning, we probably wouldn't have had a power cut.
4) Why did losing 5% of generation for a short time cause such chaos?
For commuters stranded halfway along a railway journey without water or information, this is really the key question.
The first customer disconnections occurred five minutes before 5pm, the first reconnections at five past, and power was completely restored half an hour later. There was virtually no loss of power to electric rails or overhead lines, and only a limited impact on signalling.
Not enough, you might conclude, to cause the chaos we saw.
At this stage it looks as though there are a couple of key issues.
Firstly, why aren't services such as railway signalling considered priorities, to be disconnected only as a very last resort? In the case of Newcastle Airport, the report says it's because the airport hadn't asked to be a priority – which it's now remedied. But you might think it curious that airports, railway signalling and hospitals have to ask.
Secondly, what went wrong with trains? It seems that a number of units, maybe 60, turned themselves off when the frequency fell - and then couldn't be re-booted by the drivers, blocking up lines for hours. Seems crazy, seems unnecessary.... it has absolutely nothing to do with the power failure itself, but absolutely everything to do with the chaos that followed.
5) Ofgem is reviewing everyone else... but who's reviewing Ofgem?
As the energy regulator, Ofgem obviously has a key role in working out what went wrong and holding others to account if that is merited.
But as the energy regulator, Ofgem's part in the incident also, surely, merits investigation.
It approves the regulations on how much backup Grid procures, for example. It runs the RIIO system under which Grid and the DNOs are supposed to upgrade their networks to prepare for the changing nature of the electricity system. It regulates how much profit they make by running monopoly services (pretty vast amounts, as it turns out).
It's not hard to find energy industry professionals who'll allege that Ofgem hasn't exactly fallen over itself to advance technologies such as battery storage and demand-side response that are key to making a renewables-heavy system work. A succession of small rule changes in recent years, some itemised here by my colleague Jonny Marshall, smack, in many observers' minds, of favouring incumbents rather than rewarding innovation and preparing for the future.
Could Ofgem, for example, have forced DNOs to implement more quickly the change in frequency fall threshold from 0.125 Hz/s to 1Hz/s? Might it have been completed already? If it were, we might not have had a power cut on 9th August.
Is the regulator investigating a system which its own actions or inactions have made less resilient than it might be?
Ofgem and its supporters would dispute this. And maybe they're right.
But the point is that as far as I can see, no-one is asking the question.
Ofgem is investigating 10 companies, but can't investigate itself; and it's highly unlikely that National Grid or any other company under Ofgem's regulation would point the finger at this sensitive time. Neither is the government's inquiry looking into the regulator's performance - in fact, Ofgem sits on that panel too.
All of which is a little odd, given that any failure in a regulated industry can by definition be partially down to the regulator.