Ofgem today published the interim report from National Grid Electricity System Operator into the power cuts last week caused by a significant frequency excursion on the transmission network. My initial thoughts on what happened can be found here.
The report is an interesting read, and while it adds to our understanding, it raises as many questions as it answers.
The report can basically be summarised by “a lightning strike on an overhead 400 kV transmission circuit caused it to trip but reclose successfully after 20 seconds as designed. This also caused Little Barford CCGT and Hornsey Windfarm to trip, removing 1,378 MW of generation from the system. NGESO only procure enough frequency response services to secure against a loss of 1,000 MW therefore the system frequency fell rapidly, but was successfully recovered by the Grid Code enforced Distribution Network Operator (DNO) Low Frequency Demand Disconnection (LFDD) scheme, by disconnecting 5% of the UK loads at the time, about 1,000 MW.”
Before discussing what does the above mean, here are some facts I think are important:
The weather was normal for this time of year
Load was normal for this time of year (32 GW)
500 MW of embedded generation disconnected, which the report refers to as ‘small’
Lightning strikes to the transmission system are common
NGESO was holding 1,023 MW of frequency response
What happened - Generators
The NGESO interim report tells us that a lightning strike was the root cause of the event. However, it is not clear how a lightning strike (a common occurring on the transmission system) caused the tripping of generation.
Considering Hornsea, immediately after the lightning strike the site de-loaded from 799 to 62 MW (2 modules tripped, 1 remained in service). NGESO state that the voltage fluctuation (as a result of the lightning strike) caused ‘automatic protective systems’ to activate. Subsequently these systems have been ‘fine-tuned’ and the wind farm is now ‘operating robustly’. Given no other large generator tripped, one possible conclusion is there was a protection setting issue revealed by the event.
At Little Barford, within seconds of the trip at Hornsea, the 224 MW steam turbine tripped. 1 minute later the associated gas turbine (now presumably operating in open cycle mode) was tripped as a result of high steam pressure in the steam bypass. The second turbine was then tripped manually by the power station staff, due to high steam pressures. What is not clear is if this was a reaction to what was going on in the first gas turbine and steam turbine or a separate issue with the second set. In any case, RWE state that the initial trip was caused by ‘discrepancy between three speed signals’. I am not sure what the means but it would appear to be coincidental to the Hornsea trip caused by the voltage disturbance from the lightning strike.
What Happened – Frequency
NGESO had enough frequency response services held in reserve to respond to maximum infeed loss of 1,000 MW. NGESO are clear in their report that because this event resulted in a combined loss of generation over this amount, the system was never going to be able to contain the frequency without resorting to extreme measures – the DNO LFDD systems. However the actual picture is a little more complex, as illustrated by the frequency trace below, adapted from the graph within the NGESO report (it is interesting the NGESO refused to provide me with the frequency second by second data when I requested it last week, telling me that all data is being reviewed!):
With the initial trips at Hornsea and Little Barford at 16:52, the cumulative lost transmission connected generation was 981 MW, which was within the ‘maximum infeed loss’ that NGESO had secured for. It is not clear therefore why the frequency fell rapidly all the way down to 49.1 Hz, which then tripped 500 MW of embedded generation. Given that between the first trip at 16:52:33 and the subsequent trip at Little Barford at 16:53:31 (58 seconds) the generation lost was 981 MW, it would appear NGESO was outside the ‘unacceptable frequency condition’, a potential statutory breach.
By 16:54, a total of 1,378 MW of transmission connected generation was gone, as well as 500 MW of embedded generation, causing the frequency to fall to 48.8 Hz causing the DNO LFDD relays (stage 1) to operate, disconnecting 1,000 MW of DNO MW of load, allowing the system frequency to recover. We all know the rest of the story – National Rail badly affected, millions of customers off supply.
NGESO secured against a single loss of 1,000 MW – but suffered a loss of 1,878, this is the cause of the frequency collapse. It is interesting that Ofgem approved a change to the ‘National Electricity Transmission System Security and Quality of Supply Standard’ in December 2014. Had this change not have not been made, it is likely NGESO would have had to carry additional frequency reserve – at a cost to all consumers. Given Ofgem are now investigating this event, is there a conflict of interest here?
NGESO calculated that the largest transmission infeed loss was 1,000 MW, as well as 500 MW of embedded generation. The frequency reserve carried is calculated as the maximum of the two – not a combination of the two. Given that the frequency drop triggered by a loss of 981 MW triggered the disconnection of this 500 MW of embedded generation, this now seems not the best approach.
The report makes clear that wind and solar variability was not a factor – but does not discuss the issue of limited inertia on the system. Much more work is required in this area to understand the true levels of frequency response it really needed – especially fast acing services like batteries.
More questions than answers…….