Final Blackout Report - Comments

So, the final report by National Grid (NG). This is the third and final of my posts on the blackout! The report fills in a few of the blanks from earlier, so much of my discussions from the interim report still stand. I will try to pick out the additional information gleaned from the report rather than rehashing what we already know.

As previously noted, I am not going to bother looking into the detail of why any of the downstream systems like the electric trains, of hospitals emergency generators failed, as I don’t have enough information to make any informed comments on these systems.

My overall take on the report is that to blame the loss on bad luck and an unprecedented event is partly true, but also a slightly facile response - lightning strikes, whilst individually unpredictable are a fact of life in the UK and even a direct strike on a 400 kV line shouldn’t have caused this much chaos. The fact it did so indicates systemic failures at a system integration and resilience planning level. There were really three failures that occurred, following the lightning strike, which compounded to give the overall problem. These are summarized below:

• Hornsea offshore Wind failed to ride through the fault correctly
• Little Barford STG tripped, and then a subsequent steam bypass failure caused the two remaining GTGs to trip.
• The speed of the frequency change caused a significant amount of embedded generation to trip on vector shift / Rate of Change of Frequency. The actual magnitude of this is slightly unclear in the report, but seems to be somewhere between 350MW and 500MW.

Being fair, the first two incidents are outside of National Grid’s directly control, but the third one should have been anticipated for and planned into the contingency modelling. Whilst there is now a plan to address this, failure to plan for it in the first place does fall at National Grid’s feet.

One comment that is good to see confirmed that is the reserve margin held by National Grid is circa 1GW, to account for largest loss of infeed; although in the same sentence noticing that Sizewell can be up to 1200MW at full load. It is unclear exactly on how this reserve margin is calculated though, and the expected speed of response of the reserve, low inertia systems will typically need more and faster reserve, than high inertia ones.

First things then, interestingly the total generation numbers in the summary still don’t sum to 100%, perhaps this is me being pedantic, but it seems like a silly mistake. There are three big lessons, that National Grid mention, which I have summarized below, with a few of my own thoughts.

• Communication process and protocols, particularly in the first hour should be reviewed. Ok this is a valid point, but a bit of a red herring. The speed electrical system operate at, means response & protection systems are all automated. Communication would have helped in the restoration process, but not prevented the backout.
• The list of facilities connected to the Low Frequency Demand Disconnection (LFDD) should be reviewed. Absolutely agree with this – exactly what is needed.
• Settings on internal protection of electric trains should be reviewed. Well yes this is correct, but that’s looking at a symptom rather than addressing the cause.

A bit later in the report, National Grid then note that ere three further wide-scale policy issues that should be reviewed. I have paraphrased these below and added a few of my own comments.

• A review of the security standards (SQSS) to determine whether it would be appropriate to provide for higher levels of resilience in the electricity system. Agreed this is a fundamental requirement of an ESO, and one that should be embraced in terms of strategic vision for the zero emissions target.
• Assessing whether it would be appropriate to establish standards for critical infrastructure and services. This is perhaps a red-herring it is not for NG to determine the resilience levels of industry and private networks. Whilst there absolutely should be standards – most serious critical industries already have them.
• A review of the timescales for delivery of the Accelerated Loss of Mains Change Programme to reduce the risk of inadvertent tripping and disconnection of embedded generation. Yes this is one of the key requirements, tripping embedded generation made the whole scenario much more severe than it needed to be, and could have easily been avoided.

 The annotated timeline of events can be seen below (directly extracted from the NG report). This is more of less what we already knew from the interim report. What is perhaps interesting is that if you add Hornsea (737MW) and Little Barford ST (244MW) trip you get 981MW, which is below the reserve margin (although only just); we still get a large and sharp vertical descent of frequency that triggers a further Loss of Mains protection in embedded generators down loosing another 500MW. For the eagle eyed, there is a mismatch of 150MW between the graph and the timeline in the report. The rest of the system behavior is then broadly as expected.

In the next section National Grid confirm that there were three strikes near the 400kV Eaton Socon – Wymondley line, and although not conclusive, indicates a direct strike, rather than an indirect strike. Insulation coordination (protection against lightning and other nuisances) isn’t an overly exact art, but it is slightly surprising the tower line shield wires didn’t protect the circuit; but equally it is not really a smoking gun either.

A new and very interesting graph is the voltage sag experienced on the rest of the network during the fault, which shows how the rest of the system voltages sagged and the requirement for the other connected generators to ride through the fault. As readers may remember from my earlier article, fault ride through is specified by National Grid for precisely this reason i.e. a fault on one part of the system, will cause transient elsewhere, but these transients should be ridden through and not cause disconnection of other systems and generators.

There is also a nice graph of the transient response of Hornsea, and we can see that the system tries to respond but fails to ride through the fault elsewhere on the system, and power drops off rapidly. This is obviously a failure of Hornsea ride through, so I would expect to see some shuffling of the feet at Orsted. In the explanation from Hornsea, there is reference to some sub-synchronous resonance going on and lack of damping, unfortunately without seeing the Orsted report, there is not much light I can shed on this

RWEs report notes that the Little Barford STG tripped on a speed error detected signal from their speed sensors on the unit, and they are still investigating. Whilst not ideal, this is a sort of error that can and does happen, as a generating station, any of these kind of alarm signals do trigger a shutdown to prevent damage from the STG. More interestingly, is a point I had picked up earlier, that the steam bypass system appears to have failed, which is what caused the loss of the further GTGs due to excess steam buildup. RWE are still investigating this, so we will watch with interest.

As part of the report NG have also included some simulations of their system model for the actual event that occurred, gaining good correlation, and also a simulation for a loss of 1000MW – which unsurprisingly shows the system frequency is held within the required limits. I could poke into the assumptions of how this was modelled if I had the data, but without this it would be slightly irresponsible speculation on my part.

The use of frequency response services worked more or less as expected, 89% response of the contracted systems behaved as expected. So this is good news for the energy storage industry and proves that deployment of EFR and FFR services is worthwhile and should continue to be expanded. Although anyone whose system didn’t was contracted and didn’t respond may be feeling a little nervous.

So, overall I am generally sympathetic to National Grid the general system failures of the generation plant was mainly outside of their direct control, and were generally unexpected. However, the ROCOF relays tripping the embedded generation should have been anticipated and predicted with NGs modeling practices, so they are not entirely blameless. Orsted and RWEs failures were again unlucky and difficult to predict; particularly the RWE site, but perhaps Orsted should lead to a reassessment of the commissioning and testing of the various protection schemes by themselves and NG.

The overall feeling though is that there has been a systems integration failure. Individually no one thing was responsible, but when you do get a system wide failure for one event, that implies that National Grid was not sufficiently integrating the changing nature of the systems together correctly. This leads to the conclusion that their contingency modelling was not really as robust as it should have been, and a blanked assumption of 1000MW of reserve may not be the best was of ensuring overall system security.

But overall, as a system operator you can’t prevent failures from occurring, and there will always be some unexpected behavior in the system following an event, but to have three coincident failures following one lightning strike does raise concerns, about how the overall system is viewed. Several major institutions have been guilty of these system failures of thinking, and groupthink problems, which may well have been part of the problem of NGs contingency planning scenarios.

Really the issue, is that the dynamic of the electrical system is changing at every level and it is changing quickly. Renewable power is gaining increasing penetration, system inertia is dropping, NG have limited visibility of the large amounts of generation being connected on the DNO/DSO network and the DNO/DSOs are not really ready for the responsibility of operating a system.

The UK regulatory framework put in place by the department for Business Energy Industry Strategy (BEIS) and Ofgem, is fluctuating and uncertain and is preventing investment by developers, who have been caught out by sudden shifts in policy and the problems in the Capacity Market. Interesting, I was just at the Demand Side Reduction (DSR) event held by the Energyst in London today, and a common theme from many of the speakers was the unpredictability and uncertainty in the energy market policy area. This creates a level of uncertainty that National Grid alone cannot solve, there needs to be higher up coordinated strategy coming from Government and BEIS, but one supposes that they have their mind on other things at the moment(See I nearly got to the end without mentioning Brexit)! It is only when there is a coherent energy policy, can the electrical infrastructure be put in place to prevent future blackouts.