The 2014 Highland power outage

Note: this was originally posted on another blog specific to Scotland.

In response to the news on bishop hill that MSP Alex Johnstone (Conservative, North East Scotland) was tabling the following question:

‘whether it has undertaken any further investigation into whether an over reliance on wind turbines as a source of electricity played a role in the grid failure on 16 April 2014′. (S4O-03258)

I decided to reveal what I had found in my comments on that blog. However, without the above graph the comments will not make much sense, so I’ve rushed out this post (apologies for the poor quality graph, click for a larger version). But before the graph, let me quickly look at the big picture:

What is the safe level of wind on a grid?

Back around 2000 it was well known that more than about 16-19% wind on a grid would be problematic. After the recent scandal of John Swinney misleading the public, I went to see what the current figure is. Eventually I found a paper, that quoted another paper, that quoted another paper, that eventually led back to a paper written about 1990.
In other words, there has been next to no (published) research on the effect of wind on an electricity network and no one really has any idea how serious the effect will be and at what level we will start to see these effects. This in itself is very worrying.
The analogy I would use is of a car driving down the road. As you drive the car faster and faster, the likelihood of a catastrophic accident increases because the effect of any curve (change in the wind) is all the more greater, and the time to respond is therefore all the smaller. Moreover, wind is displacing conventional backup as it is being made uneconomic – so not only is the destabilising effect of wind increasing, but the ability to stabilise the grid from conventional power is decreasing.
Just as one can make the car safer at higher speeds by getting better equipment – brakes, tyres, etc., but fundamentally, speed = more wind, can be partly offset by more and more money spent on protecting the network, but eventually no matter how much is spent too much instability will close down the grid as surely as too much speed will cause any car to go off the road. We just don’t know what that figure is!
But, in researching the background to this power outage, I became aware that the North of Scotland is now regularly having power outages. This is just what we would expect from a system under pressure. TO use the car analogy – we are already seeing the screeching of wheels and the excessive use of brakes.

The graph

This graph is a plot of the rate of change of power on the grid from various sources as well as the differently scaled change in phase (red dot). The most dominant feature is the massive change in phase which changed dramatically from 8:35-8:40 and then recovered from 8:40 to 8:45. This is like the heartbeat of the electricity network – it’s the single most important measure telling us if the system is under stress. It was clearly under a huge stress at 8:40, there is no obvious “stress” at 8:30 – 8:35.
The other important evidence on this graph is the way the wind curve (light blue) closes matches the rate of change in phase. The two match each other in their peaks and troughs showing that the main cause of changes in phase of the network was due to wind. In other words, wind power was the dominant feature that evening. To go back to the car analogy, it would be like plotting car speed against road curve. If the car is near its speed limit, each bend will cause the speed to decrease. If however the car is well within its speed for the road, it will hardly change the speed around the bends. The electricity network was clearly near its limit and having problems coping with the amount of wind.
What is perhaps most important is that the outage occurred at a time total demand was decreasing, with the result that Hydro (mid blue) was being turned down. So from 8pm till the massive power phase change at 8:35, hydro was being ramped down until (as seen on a graph of total power not change) very little was being produced. This is important because hydro is very quick to react and so a good proportion of hydro stabilises the grid. Turn off the hydro – and the grid becomes less stable – or as strongly appears to have happened on this night – we get a catastrophic instability!
These show that the right conditions existed for a catastrophic failure. However it does not give the actual cause. But in examining the supposed “relay fault” which we are told occurred at 8:30, it is very interesting that the main evidence of a power cut (a sudden and dramatic change in phase) occurred 8:35-8:40, at least five minutes after the supposed relay fault at 8:30. This is also backed up by a number of anecdotal accounts of lights going out well after 8:30 at around 8:40.
This strongly suggests that if there was a fault at 8:30pm that the actual event causing the power outage did not occur until at least 8:35 but instead showed at 8:40pm as a massive and sharp change in grid frequency. That suggests another event occurred around 5-10minutes after the supposed “relay fault” and so there is 5-10 minutes which has not so far been explained between the “fault” and the power outage.
This time discrepancy needs to be explained because I cannot see how a fault at 8:30 could lead to a sudden massive change in mains frequency at 8:40 unless there were an additional factor of which the large and dominant affect of wind, and the collapse in hydro output (presumably due to a bad forecast and general otherwise oversupply) that night on the whole system is very likely to be the cause.
In other words, even if there was a “relay fault” (which itself seems doubtful as given in the original letter) this was just coincidental was probably found as a result of having to cope with the massive power surge from wind – and is all in all just a very handy excuse.