That blackout

Readers may recall the story I posted about a major power cut in the North of Scotland last month and the speculation that the underlying cause was windfarms. This has officially been put down to a faulty relay, but today Euan Mearns notes a letter by an electrical engineer in a local newspaper which tells another story:

SIR, I was amazed to learn that a Scottish Hydro Electric transmission spokeswoman said “repairs are being carried out on the faulty relay” that allegedly caused the power cut on April 16 (“works to fend off blackouts”, P&J, May 10).

I have been an electrical engineer for over 40 years and have never heard of anyone “repairing” a hermetically sealed relay switch.

The relay switch operated perfectly on the windy night of April 16 when it detected a sudden surge of voltage and frequency that fell outside acceptable parameters.

A relay switch has two states: on and off. All of these relay switches operated perfectly on the night, independent of the relay switch at Knocknagael Substation which is, itself fed by at least two windfarms, Farr and Moy.

This was what is known as a “rolling blackout”. It is ludicrous to suggest that all lights went out all over the north at 8.30pm exactly. My area went out at 8.43pm when the blast of wind reached Novar windfarm and toggled the relay switch to off to protect its local circuit and so on up the coast.

Grid operators can switch windfarms on and off remotely – if there is a risk of too much wind generating too much “wrong time” low-grade electricity with what is known in the industry as “flicker”. The grid cannot handle more than 10% of flicker contaminated electricity at any given nanosecond and this limit was exceeded on the night.

The operators were caught on the hop. With no electricity, all the windfarms had to be isolated manually.

The spokeswoman goes on to say that they will be making changes to how the protective equipment operates. This is code for shutting down windfarms even earlier in windy conditions so that the operators get more and more constraint payments for not generating when the wind speed is just right.

Andrew H Mackay, Tain

Comments

[Joe Public]

@ MikeHaseler May 12, 2014 at 10:33 PM & Euan Mearns May 13, 2014 at 9:39 AM

I suspect a Highlands crofter, switching off his kettle, could be deemed a contributing factor.

[MD]

May 13, 2014 at 8:39 AM | Martin A

It's not a typo. You have to distinguish between the relay's ability to monitor and the ability to keep many relays synchronised across the system. The relay needs to act as fast as possible to protect a fault, but also in order to fault find, etc it is necessary to have all the relays synchronised. The text from the datasheet on High Accuracy Time Keeping says:

Using high accuracy IRIG-B from a global positioning satellite clock, the SEL-421 can time-tag oscillography to within 10 μs accuracy. This high accuracy can be combined with the high sampling rate of the relay to synchronize data from across the system with an accuracy of better than 1/4 electrical degree. This allows examination of the power system state at given times, including load angles, system swings, and other system-wide events. Triggering can be via external signal (contact or communications port), set time, or system event. Optimal calibration of this feature requires a knowledge of primary input component (VT and CT) phase delay and error. A standard accuracy IRIG-B time-code input synchronizes the SEL-421 time to within ±500 μs of the time-source input. A convenient source for this time code is an SEL communications processor (via Serial Port 1 on the SEL-421).

The mains frequency is 50Hz, but this has no bearing on the need to isolate faults as soon as possible.

[gareth]

At the risk of getting into one of those XKCD "someone is wrong on the internet" moments: bearing in mind that 1/4 degree at 50Hz is equivalent to just under 14us, and the speed of light is about a nanosecond per foot, how does one distinguish between "1/4 electrical degree" and a positional estimation error of 2.7 miles?

Or maybe it really was the crofter's kettle wot dun it ;-)

[It doesn't add up...]

Martin A:

The spec goes on to mention sampling at 8kHz so the 10 microsecond time stamp accuracy seems quite plausible when events are recorded internally in the device. There appears to be a time sync signal connection on the rear. The 5ms figure only comes into play when the device is monitored over Ethernet.

[graphicconception]

Sounds like a system that was incorrectly designed ...

This comment is probably more true than was realised. The problem is that the current grid infrastructure was designed to work with constant and steady supplies of electricity from coal, gas and nuclear power stations. Now lots of small, intermittent generators have been bolted on. The previous, relatively steady, flow of electricity has now been disrupted. The grid cannot cope. Research is underway to improve the wind turbines but a complete solution is still awaited. See example here: http://www.icrepq.com/full-paper-icrep/208-BENDL.pdf

In Germany, for instance, they cannot use all the wind power that they generate. So they claim the Brownie points for generating it then sell it cheaply to their neighbours. Neighbours lucky enough to have a lot of pump storage can then sell it back to Germany for a profit at a later time. German consumers lose out twice. They pay extra for the wind turbines and then pay extra again to get their own electricity back.