grid

That UK power cut in August

You probably all remember that power cut earlier in the year? There was all sorts of unfounded speculation about the causes. The initial report into the causes is out; it’s actually been out for a week or two, but I just got around to reading it. You can read it here if you want to – it’s not overwhelmingly technical – but here’s a brief summary, as I understand things:

First off, this was not because of any inherent problem with wind power – although it did involve a wind farm – and nor was it because we don’t have enough generation capacity in the UK (that’s a scare story that arises every winter, but this happened in summer in any case). The timeline went as follows,

  1. Lighting struck a 400kV power line in Cambridgeshire. That’s a fairly normal occurrence, and protective systems are designed for it. Those systems worked, and the transmission line was back to normal operation in a tenth of a second.
  2. Less than a half second later, as a direct or indirect result of the lightning strike, a large offshore wind farm stopped supplying the grid. This should not have happened, and the wind farm operators are investigating why it did. At the same time (within a second), part of a nearby gas power station also disconnected. It is not clear why that happened either, and that’s also being investigated.
  3. The grid was operating with enough reserve generating capacity to cover the loss of the single largest generator that was in use. This equated to a reserve of about 1GW. Losing two generators at the same time is not something that is normally allowed for, but as it happened the loss of generation from those two combined was a little under 1GW. The automatic systems detected the loss of generation and compensated with reserve generation 20 seconds later. So far, so good.
  4. About a minute after the initial event, another chunk of the same gas power station tripped. Again, it’s not yet clear why.
  5. This additional loss of generation was more than could be compensated for by the immediately available reserve. With demand greater than supply, an automatic system triggered the disconnection of about 5% of electricity users across England and Wales, in order to bring the two back into balance. This is the power cut that people noticed, and it is exactly what is supposed to happen in this scenario.
  6. In disconnecting 900MW of demand, the automatic system also caused about 600MW of embedded generation (mostly small-scale renewables) to be disconnected, so the net loss in demand was only 300MW. This is a problem in principal, but in this case dropping 300MW of demand was sufficient to stabilise the situation.
  7. Over the next four minutes, additional generation was brought online to return things to a normal operating condition.
  8. Customers disconnected in step 6 started being reconnected a few minutes later, and all were back to normal within 45 minutes of the initial event. Except for the railways, but that’s a separate question about how the railways respond to loss of power, rather than why the power was lost.

So, by and large things worked as they are meant to. The system carries enough reserve for any one generator to disconnect unexpectedly, but in this case two generators did so at the same time. Hence, the power cut.

There are three outstanding questions, which are being investigated further: (a) Why did the wind farm trip after the lightning strike? (b) Why did the gas power station trip after the lightning strike? and (c) How should embedded generation on the distribution system be handled when parts of that system are automatically disconnected?

The first two of those questions are purely technical ones: something happened that shouldn’t have happened, so the engineers need to know why so that it can be fixed. The third one is a bit more interesting, and is the only potential area in which the shift towards renewables is involved (although it’s important to note that in this particular case, it made no difference). It’s not clear to me at present, because I’m not a grid expert, whether there is supposed to be a mechanism by which distribution-connected consumers get dropped while distribution-connected generators stay online – in which case that system failed – or whether the issue is more that the generators are going to be dropped, and that the amount of the distribution grid that is automatically disconnected needs to be adjusted to account for that generation, so that the correct amount of net demand is removed.

Perhaps the final report in November will tell us more.

Posted by simon in The wider world