Power Grid fluctuations in the UK

[madd0ct0r]

http://www.bbc.co.uk/britainfromabove/s ... tain.shtml

Everyday, a perfect storm of public programming (so no ad breaks), a densely packed island nation in a single time zone, and the British urge for a cuppa create a unique effect: The 'dun dun dun.dun.dun' power surge.

[Borgholio]

Pretty interesting actually. While I understand the need for human oversight, I am wondering why a computer can't be used to more quickly and precisely balance the load...especially in the given example where France couldn't generate enough power on their own.

[Lusankya]

Presumably because a computer would be reactive, rather than proactive. The demand peak is at a slightly different time each day, because the end of Eastenders would vary by a few minutes, so you need someone there actually watching the show in order to know when exactly to turn the dams on.

[Borgholio]

Well yeah I mean in a special case like this a human is needed, but for instance when the French link went dead, a computer could have spun up several power plants much quicker than the operator did.

[The Duchess of Zeon]

Yes, the problem is that you can't preprogramme the system for this kind of contingency since all programming of this sort is either time based or function based and you have to deal with both, plus the extreme variability within that five minute period of when the power actually spikes. You can't predict the spikes at any given moment over the five minutes in question, I suspect, you just know it will happen over those five minutes. Socially predictable creatures we are, but still random: How many people will get distracted talking about the show for four minutes before remembering to flip the kettle on, instead of two minutes, is a much higher variable, so you can't easily have a preprogrammed response pattern to the show's ending. The fact it doesn't end at a definite time each night also completes the absurdity of it.

Though really the cool thing is that we now have enough computers to let us actually do something like this with an integrated grid. Think of how easily the grid would have fallen to pieces without the ability to control powerplants from a central location. that's why a massively unified electrical grid like this was almost impossible a few short decades ago.

[slebetman]

Well yeah I mean in a special case like this a human is needed, but for instance when the French link went dead, a computer could have spun up several power plants much quicker than the operator did.

From my experience developing power grid equipment 10 years ago I'd say that it's thanks to computers that the operator managed to spin up those power plants in the first place. Power grids have been controlled by computers since the 50s. Without them doing things like balancing load and spinning up or shutting down powerplants would be very hard to do.

For the predictable stuff like when to service/replace transformers, balancing load during prime time (the problem is not really unique to the UK, though theirs is a more extreme case of the norm) etc. we can in fact automate the process. And indeed, some places do. Back when I was in industry 10 years ago, Canada and the US (especially California) was experimenting with automation. Not sure what it's like today but I believe most of North America and Western Europe is heavily automated.

Still, as mentioned there are cases where things are not quite predictable or scenarios that the designers either did not consider or were too unlikely to be worth considering. These are the corner cases. For these, employing people is more cost effective.

The other thing to realize is that systems vendors can and do implement almost any feature you can think of. But the utility companies don't have infinite budget, so they buy/implement only what's considered absolutely necessary. Doing anything beyond that is considered overengineering (but our marketing people will try to sell them stuff anyway).