The National Energy System Operator (“NESO”) has been stung by my analysis of the blackouts near miss on 8 January, at one point apparently telling journalists, somewhat pompously that “WE are the National Energy System Operator, and SHE is “just” an independent consultant!” In this blog I will carry out further analysis of what happened on 8 January and what NESO has said about it at different times, and highlight the system operator’s disappointing lack of transparency.
Initial attempts to discredit me did not succeed
In the days following my blog it was picked up by the media and was widely read on X /twitter (430k views)! Here is a selection of the coverage:
- Britons paying ‘£2m an hour’ to keep gas-fired power stations running in freezing temperatures
- Britain ‘came within whisker of blackouts’ after plunge in wind speed
- UK’s blackout near-miss shows risks of Net Zero
- Blackout risk as energy supply warning issued during freezing temperatures as rush to net zero blamed
- UK households face ‘blackouts’ after energy supply warning issued
- Will Britain suffer a power outage? Freezing weather brought UK ‘within a whisker’ of blackout
- How the energy grid came ‘within a whisker’ of blackouts over sub-zero conditions
- The traders making millions by gaming Britain’s power crunch
It was even covered in foreign language press: Wielka Brytania o krok od blackoutu? Poważny incydent pierwszej zimy bez węgla (this Polish headline draws attention to this being the first winter without coal, and yes, I do speak and read Polish!) I was also interviewed on TalkTV and GB News.
I would particularly like to thank Matt Oliver at The Telegraph and David Rose at UnHerd who both pushed NESO to answer the questions I was posing, sadly without much success, but not for the want of trying!
The first public comment by NESO was:
“NESO operates Great Britain’s electricity network to one of the highest levels of safety and reliability anywhere in the world. Yesterday our control room engineers used our standard operational tools to manage the electricity network and ensure that we maintained enough electricity for our standard operating contingency. At no point were electricity supplies less than anticipated demand and our engineers were able to rebalance the system without the need to consider emergency measures. One of the standard operating reserves held by NESO at all times is for the largest power generator on the system, which last night was 1400 MW, not the 580 MW that has been quoted online,”
– Craig Dyke, Director of System Operations, NESO
Follow-up questions from journalists who had set a deadline of 5pm on Friday 10 January for more substantive information (including a list of the units which were providing the margin and reserve on 8 January) yielded a two-paragraph response to the effect that everything had been fine and we should all just take its word for it. And none of the details that had been requested.
But it did not end there. I continued to be dis-satisfied by the lack of a proper explanation from NESO, and on Monday 13 January published further analysis of NESO’s forecasting errors which contributed to the near miss. Later on Monday, Andrew Bowie, the Shadow Energy Minister said in the House of Commons:
“All our constituents will be aware of the freezing temperatures experienced across the United Kingdom last week, dipping to minus 18° in the north of Scotland. However, many will not be aware of just how close this country came to an energy shortage, blackouts, or demand control—closer than at any point in the past 15 years…
Earlier in the week the National Energy System Operator issued a call for electricity providers to step in to provide extra electricity to meet demand and limit the risk of blackouts, paying 10 times the average daily amount to keep the lights on, all of which will end up on the energy bills of our constituents. With an incredibly tight margin between demand and available power generation, we were once again forced to rely on reliable gas power plants to keep the lights on in this country, showing that gas is and will be a vital component of our energy security for decades to come.”
As those who follow me on X /twitter will know, I have been engaging with the Shadow energy team. Claire Coutino, the Shadow Secretary of State for Energy Security and Net Zero was kind enough to tweet that I should be among a group of market expo
erts advising the Government on energy.
On Tuesday 14 January I recorded a webinar with UnHerd, one of the media outlets that has been working with me to cover this story, with the resulting interview being published to their YouTube channel on the following day.
NESO invents a new concept to mislead the market
Later on Tuesday, NESO published a short 2-pager in which it claimed to have held 3.7 GW of “headroom”.
“Headroom” is defined in the Grid Code as “The Power Available (in MW) less the actual Active Power exported from the Power Park Module (in MW),” where a Power Park Module is an Offshore Power Park Module or an Onshore Power Park Module which are themselves defined as non-synchronous generating units powered by an “intermittent power source”. In other words, NESO is playing with words. We are concerned with Margin and Reserve. Headroom, as defined in the Grid Code, has nothing to do with it, or if it does, NESO is claiming, incorrectly there was 3.7 GW of intermittent generation spare at the peak on 8 January!
It’s also curious that in the Electricity Margin Notice, NESO said there was a 1,700 MW shortfall versus required contingency of 900 MW (later adjusted to 1,120 MW shortfall vs 448 MW contingency). If there was always this mysterious “headroom” of 3,700 MW, why did we need the Electricity Margin and Capacity Market Notices? Surely there was plenty of spare generation around?!
Elsewhere in the paper, NESO says that the market tightness was exacerbated by there being roughly 3 GW of interconnection to Europe that was unavailable. It said this is “not unusual at this time of year, as maintenance is carried out over the Christmas period and returns typically through January”. It also pointed out that the Winter Outlook published in early October highlighted this period as one in which margins are generally expected to be the most tight.
So it is a little surprising that in the Winter Outlook, NESO said it assumed that 6.6 GW (de-rated) net imports would be available via interconnectors at times of tighter margin. On 8 January, only 6.28 GW was actually secured at the peak. If there is an expectation of maintenance at this time of year, perhaps more conservative assumptions should have been made.
According to this paper, peak demand was 45.841 GW at 7:30pm. This corresponds to the Initial Demand Outturn, whereas the Initial Transmission System Demand Outturn data from BMRS was 46.825 GW (at 17:30). The Initial Demand Outturn takes into account transmission losses but does not include station transformer load, pumped storage demand or interconnector demand, whereas the Initial Transmission System Demand Outturn takes into account transmission losses, station transformer load, pumped storage demand and interconnector demand. This is the higher number, and represents the number that has to be met by generation, so there is a question as to why NESO mentioned the lower number.
Operational Transparency Forum is told another story
On Wednesday 15 January, a week after the event, a member of the control room team walked though what NESO did leading up to the peak on 8 January during the Operational Transparency Forum (“OTF”) call. Interestingly he did not say what happened at the peak, or what they learned afterwards, it was a purely forward-looking view. He described how in the morning of 8 January, a planned network outage was cancelled, an ongoing network outage was ended, planned commissioning work was deferred to reduce risks on the grid, and models were adjusted to take a more conservative view of the availability of units returning from outages.
In addition, Medway CCGT was asked to return to service and Sutton Bridge was asked to run when it had planned not to for reasons that are not public (there was no notice of maintenance or similar on REMIT, they just had a notice to deviate from zero of 360 minutes which is too long for normal BM participation – questions were asked about this during the call, but NESO batted them away saying they won’t comment on individual units).
I can’t quite figure out what was going on with Medway from its REMIT notices but I assume that Amira, whose availability data I used in my previous analysis included zero output from it since it was on an outage, and that on the day, it was able to generate 340 MW during the evening peak, increasing the margin by this amount.
NESO also apparently gained 150 MW from EWIC and Moyle interconnectors with Ireland, but I don’t understand this claim since BMRS indicates no positive flows on these interconnectors (exports are not shown) during 8 January. Amira shows exports that day for Moyle of 150 MW and that this was in line with the operational schedule. It’s possible this schedule was changed early enough to not appear on the chart. It was a similar story on the East-West interconnector (exporting 100 MW). Both exported larger amounts before and after the peak. However, when I ran my numbers I took this into account, so my analysis was after this addition, if there was an addition.
It’s worth noting that we almost always export to Ireland. So much so, that I didn’t even look at the Irish interconnectors when I made my initial analysis since I think of Ireland as demand on the GB grid. The additional capacity NESO says it secured on the Irish interconnectors was a reduction in exports, not an increase in imports, ie it effectively reduced GB by demand a small amount. Ireland was also experiencing similar weather conditions to us and a tight grid (there is a single unified power grid for the whole of island of Ireland, both Northern Ireland and the Republic of Ireland).
There was also 180 MW of Demand Flexibility Service which I overlooked in my previous analysis (pity it wasn’t the 2 GW demand-response that used to be available under Triad avoidance!)
Interestingly, one participant on the call challenged NESO about the availability of Short Term Operating Reserve (“STOR”) and Positive Balancing Reserve (“PBR”). He said that batteries and peaking plant often run in the wholesale market, despite having a STOR or PBR contract, even though this is explicitly against the rules, because plants that are running cannot provide reserve.
The questioner said this was a particular problem on 8 January – which makes sense because the amount of money that could be made in the wholesale market was significant – cashout prices went as high as £2,900 /MWh. The answer was that if units do this they do not receive availability payments and on top of that a penalty regime was recently imposed (the latest STOR terms and conditions are here). There was a further question about publication of the penalties and NESO said it may consider publishing aggregate penalties as it does not publish performance data on balancing services at the unit level.
NESO was asked if any non-BMUs were included in the margin. The answer was that most comes from BMUs but there is some from STOR and Fast Reserve. (A BMU is a Balancing Mechanism Unit).
What have I been able to find out for myself: my original calculation was sound!
With NESO continuing to refuse to explain which units were providing the spare margin and reserve on 8 January, I have looked (with the much appreciated help of one of my readers who posts as “It doesn’t add up…”) at the Maximum Export Limit of all BMUs connected to the transmission system on 8 January, excluding wind and interconnectors which I looked at separately, to see what generation would have been available to run if needed.
Peak demand was at 5:30pm ie at the end of Settlement Period (“SP”) 35 and the start of SP 36. SP35 generation was slightly higher than SP36 so I used that for my analysis. The API returns MEL for the following classes of assets: 2 = aggregated, C = consumption, E = embedded, M = non-pumped hydro, T = transmission-connected and V = virtual power plant.
I added up the MELs for each of these categories. Then I excluded wind and solar because their MELs indicate what they could generate on a windy/sunny day, and double-counting (within the settlement period some assets had reported MEL multiple times, but obviously could only run once, so I kept the entries closest to 5:30pm which is when I believe the peak was, rather than 5:00pm). I also excluded the MEL of units that had STOR contracts but did not run in the wholesale market since they should be providing reserve instead of margin. I then added back the wind which actually did run, and actual net interconnector imports.
This gave me total availability of dispatchable generation plus actual wind output and actual net interconnector imports of 47.290 GW ie 465 MW higher than peak demand, which is not far away from the 580 MW of my original analysis.
There was some easily identifiable headroom. As mentioned in the OFT call, Sutton Bridge had not planned to run on 8 January and was asked to make itself available in the Balancing Mechanism (“BM”). It was accepted at its Stable Export Limit (“SEL”), so there was about 400 MW headroom to the MEL. Connah’s Quay offered itself at £2,900 /MWh in the BM and was accepted at SEL for three units and MEL for the other, giving another c300 MW total. Similarly, Rye House, which was offering itself at £5,500 /MWh in the BM was also accepted at SEL giving another 300 MW spare, except that NESO stopped extending it at from 4:30pm and by 5:30pm it was fully offline. It actually reached zero at 5:15pm, and with a ramp time of about 25 minutes to MEL, it would not have been able to reach full load by the peak.
While these assets could be called on to meet regular demand, they should not be treated as reserve. It takes around 25 minutes for a CCGT to start from zero to MEL, and 15 minutes to increase from SEL to MEL. This is too slow to be part of the initial response to the loss of generation covered by reserve. Normally batteries respond immediately until Dinorwig, the largest pumped hydro plant in Europe at 1,800 MW starts up taking 16 seconds to reach full load (water drops through its six vertical turbines in a way which is incredibly cool!), and this can run for 5 hours at full load, giving time for gas power stations to ramp up, even from a cold start which would take over an hour.
Looking at what Dinorwig did on 8 January:
- Unit 1 was at MEL (300 MW) during the peak
- Unit 2 was unavailable all day
- Unit 3 was at MEL (300 MW) during the peak
- Unit 4 was at MEL (300 MW) during the peak
- Unit 5 was unavailable all day
- Unit 6 was unavailable all day
This means Dinorwig would not have been able to respond to generation loss since the available units were already running!
There was 848 MW of contracted STOR, but, noting the above comments, I checked the PNs ie actual running profile of the assets and found they generated 599 MW in SP35. A couple of the units generated a lot. Stripping out the units that ran, those that didn’t could have provided 532 MW of reserve and not the 848 MW contracted. Fast Reserve data for 8 January are not yet available on the NESO portal. Data available for 1 January suggest there might have been 50-150 MW of Fast Reserve at the peak. So total procured reserve was maybe 700 MW.
Both my bottom up and top down analyses have yielded essentially the same result – about 500 MW of margin at the peak on 8 January. On top of this there was about 700 MW of reserve. This means that the single largest infeed loss of 1,400 MW could not have been covered. The 3.7 GW “headroom” appears to be a complete fiction, which could only be met by units that are not BMUs as these are the ones I considered (other than some STOR units). However, NESO could clear this up very easily by simply publishing the list of units providing margin and reserve at the 8 January peak, including volumes.
Someone at NESO told me they always cover the Security and Quality of Supply Standard (“SQSS”) because they have to – well everyone that drives a car “has” to have car insurance, but many people don’t and that is generally only discovered either when a third party tries to claim on the non-existent insurance or the police check for some reason such as the driver committing a motoring offence or other crime. This seems analogous – NESO “has” to secure the SQSS, but is anyone actually checking? Or are Ofgem and DESNZ assuming it’s fine and will only check after demand control or a blackout take place?
The questions that remain
Unfortunately we still do not know:
- The actual difference between available supply and demand at the peak on 8 January
- What the actual 1-minute peak demand was (rather than the half-hourly peak) – ie the highest instantaneous demand that had to be met
- The corresponding actual available supply from generators, batteries, interconnectors and demand side response
- Which units exactly provided the margin by which available supply at the peak exceeded demand at the peak and which units provided reserve
- What the difference was between what actually happened at the peak and what NESO thought was going to happen when the EMN was cancelled
On this last point, NESO was conspicuously silent in the OTF call (which I was unable to join live otherwise I would have asked myself). Much has been made of the cancellation of the notices, but these appear to have been based on inaccurate demand forecasts, so their cancellation is not comforting!
My analysis suggests NESO was unable to meet the single largest infeed loss and was therefore in breach of the SQSS. I invite it to DEMONSTRATE ie not simply assert, how it did comply with the Standard.
On its website, NESO says “We actively embrace the need to share our data with our customers and the industry, fostering transparency, innovation, and collaboration.” Sadly, it is not putting this claim into practice!
.
POST SCRIPT
David Turver has published an analysis of the Winter Outlook which is well worth a read:
https://davidturver.substack.com/p/neso-margin-call
You may remember my take on it a few years ago where I criticised the use of average annual generation for the determination of the supply available to meet ACS demand, given there is often a high correlation between cold weather and still weather:
https://watt-logic.com/2022/10/06/electricity-winter-outlook-2022-23/
Kathryn Porter for Energy Minister please!!
I do not understand why NESO are not more conservative and also push back against the way our government is prematurely retiring dispatcgable generation. There will be no praise for them running close to the margin if and when a trip occurs. (With when being much more likely)
I would have thought, that as they will get much of the flack when there is a failure that they would resist government policy and tell them they will not be held responsible if the grid fails due to that policy against their recommendations?
I think they will just say they did everything the were asked to., eg using bad models because no-one asked them to improve them…
At which point the Government will hang them out to dry with the Civil Service pointing to the quasi authority as having gone rogue and it all should have been kept in house.
Very true, head of NESO Ed Milliband will ensure the crew take’s the blame.
Barry Wright, Lancashire.
A very concerning analysis.
Currently, the wind generation is still low at just 6.62GW, I hope we don’t get poor wind speeds every winter..
I wonder if there would be any point me writing to my new Labour MP asking why Ed Miliband has axed the fully cost analysis, as a way of showing up the latters’ duplicity.
“I think they will just say they did everything the were asked to., eg using bad models because no-one asked them to improve them…”
I’m sure the history surrounding control of transmission centre’s is relevant to where we are today. Pre privatisation (1990) it was System Operations (sys/ops) with five regional area control centres reduced to one national control (1980’s) working closely with generation, both always the main focus within the Central Electricity Generating Board (CEGB) & the media. Generation with a huge fuel & construction budget dwarfed all. Sys/ops national control selecting generation plants to deliver in real time to match national demand from a merit table, base load at the top (nuclear, large coal fired PS’s) working down based on day before bid-in cost. Variables factored in like weather, historical demand etc. Verbal exchanges, hands on throttles. Somewhat simplified but totally different from today; read predictable merit table v asynchronous renewables ?
In between, transmission area’s responsible for maintenance repairs & renewals, invisible yet a vital part of national network security. Miles of 400/275/132kv overhead lines plus hundreds of substations. Infrastructure from power station terminals to bulks supply points serving DNO’s. Rapid expansion of the super grid leading to transfer of the 132kv grid assets to localised DNO’s in the 1980’s. We in the areas received a 1/2 year ahead super grid outage plan populated with large construction outages in place involving lots of contractors requiring on site supervision & safe access at all levels. Area outage requirements for urgent repairs, renewals, uprating always a struggle leaving the impression that sys\ops lived in an unreal world of 100% network availability at all times. In contrast a real world involving lots of dedicated site engineers & work teams delivering major work programmes year on year. Safe access for working parties along side day on day engineering decisions, some task; Often unrecognised IMO.
Sys/ops switched out a circuit & transferred responsibility to site. Lots of temporary earths within close proximity to live equipment. Issuing/cancelling permits to working parties; a major responsibility. Overhead line teams spread over several miles involving manual application at point of work earthing to every conductor at each tower, a quad bundle needing 4 earths per phase, 12 at each tower. Not wishing to labour a point but feel many on this blog may appreciate the valuable input of on site personnel delivering major work programmes beyond the meeting rooms.
At privatisation (1990) National Grid was born including the area’s & sys/ops. It is my view that things continued as previous however the grid network has become a highly visible & important part of energy delivery. I think sys/ops have continued as previous sticking with systems that worked combined with an element of old culture wrestling with a variable generation mix this has overlapped into the new recently nationalised NESO. Hence the short comings highlighted by Kathryn in her excellent post.
Barry Wright, Lancashire.
Thank you for the clear explanation of what is known and what is hidden. Clearly the “unknown unknowns” such as how close individual generators came to outages will remain as such.
Could this be the same NESO that published a report Clean Power 2030 vaguely validating Miliband’s unfeasible plans for Green power in such a short timescale? Presumably they will continue to monitor and update us with progress towards that target even when it moves through things such as Starmer’s blank cheque for energy for the Emperor’s new AI clothes.
I’m getting in the popcorn to see how this ends. Experience of Government organisations leads me to believe that once they lose full honesty for the first time it becomes habitual.
Hi Kathryn – so good to see you keep pushing. Hope the best for you and life is good over there.
Very interesting to follow this series looking into what we all fear – that black outs are coming one way or another.
BR Mogens
Keep up the good work – I understand the basics of what your are talking about and I think you are covering it very well. Many, many more need to understand the operation of the grid system if they are to make sensible decisions about meeting our current and future electricity needs, both in term so security and in terms of economics. Your webinar with Unherd needs to be widely listened to and understood.
I am not holding my breath!
Excellent post!
I especially endorse the proposal for ‘red-teaming’ NESO’s plans.
Thank you, this is a information and well written and articulate response..
I have been( largely unsuccessful) trying to point out via a small you tube channel the perilous state of the grid, the increasing unrealistic demands being put on it year on year, over optimistic bordering on pie in the sky forecasting of future needs and overbudget and undercapacity projects for future needs..
Candles..
Buy candles.
Keep going Kathryn. Don’t let Neso grind you down. You clearly have them rattled.
On the 8th or 9th of January the BBC told us about this near miss luckly I had all my backup power fully charged, I lived through the powercuts of the 70s and early 80s thanks to the miners and I also live through the worst winter had to offer, I have learned not to rely or need the power companies and having bought the lowest power drawing appliances I can reduce my daily power fraw to as little as 300w per day if I need to,
As for lighting I never use Candles because I have LED lights that can run for over 300 and 500 hours non stop on one set of batteries, The big problem is No Disrespect but people in the UK and this Site just does not know what we can buy and is just a click away, I saw Kathryn Porter. in a Video and was impressed with how she knew all the facts without any Loony conspiracies, WTG KP.
I have found and tested over 33 Low Watt household Appliances over the last 4 or 5 years and it is possible to reduce your powerbills by as much as 80/85% if you make a plan and stick to it, Anyway Thanks everyone and stay safe.
From the NESO website “National Energy System Operator (NESO) is a new public corporation owned by the Department for Energy Security & Net Zero (DESNZ). From the launch of NESO in 2024 we become a public authority under the Freedom of Information Act 2000. NESO is also subject to the Environmental Information Regulations 2004.”
NESO will have to respond to an FOI within 31 days – perhaps a well tailored FOI request will get the missing information. If the not then it can be taken to the Information Commissioner for a ruling
A few of my journalist friends are doing exactly this!
As always it is difficult to do a thorough investigation when you are not privy to certain data but this work of Kathryn appears to be a master-class and similar to the work of Bellingcat. People want to hold onto their jobs and generally do it to the best of their ability. Maybe I’m too trusting but I imagine this is the case with the rank and file at NESO. They are not being attacked or accused of duplicity and are likely quite proud of the work they do. I imagine that if they [NESO] continue to stonewall there will be leaks or even a more senior whistleblower unhappy with the obfuscation will come forward. Being a government department don’t they come under the FOIA? Even if they use the Section 43 (2) excuse that it is commercially sensitive even that moves it forward. Why the secrecy – it is after all energy security!
Apologies – there’s a glitch in my comments form which wrongly attributes comments….not sure who actually said this one!
As always it is difficult to do a thorough investigation when you are not privy to certain data but this appears a master-class and similar to the work of Bellingcat. People want to hold onto their jobs and generally do it to the best of their ability. Maybe I’m too trusting but I imagine this is the case with the rank and file at NESO. They are not being attacked or accused of duplicity and are likely quite proud of the work they do. I imagine that if they continue to stonewall there will be leaks or even a more senior whistleblower unhappy with the obfuscation. Being a government department don’t they come under the FOIA? Even if they use the 43 (2) excuse that it is commercially sensitive even that moves it forward. Why the secrecy – it is after all energy security!
Apologies – there’s a glitch in my comments form which wrongly attributes comments….not sure who actually said this one!
Given that power cuts can result in deaths why are the NESO not required to publicly publish the information relating to this issue.
The transport industry (I worked on the railway for 20 years) has to hold an inquiry for near misses that could have resulted in a loss of life. Is the NESO held to a lower standard regarding potential loss of life?
The point of an inquiry to to find out the level of the failure in terms of whether it is caused by external factors like pressure from the government to achieve a particular goal or internal factors like structure, training or competence.
As an auditor (one of my roles) I found a number of issues as above but the most significant was inadequate funding by senior positions because of the ‘reprioritisation’ of safety goals. As auditors we had the power to put a stop to that and we did.
What a great point. Which could potentially secure some more transparency throughout the whole sector not only in the UK.
Great thought! Thanks
I can only hope that they get struck off before they kill their patient. The National Energy System Operator is clearly rattled by your independent examination of the health of the electrical power system. There are a few very revealing sentences in their “Clean Power 2030” report that show puzzlement at warning signs from their modelling that their treatment plan is not working. They will continue with treatment until near death. Then expect face saving and resignations.
Kathryn surely has this one wrong, because Modo identified batteries still charging at peak time. There were 700 MW of batteries charging from 17:00-17:30, and 500 MW from 17:30 to 18:00. See https://www.linkedin.com/feed/update/urn:li:activity:7283074377959497728/
The peak demand was expected to occur around 17:30.
Presumably Modo doesn’t have access to any more information than Kathryn has, but Kathryn has not mentioned this.
The largest possible loss of load at the time was a 1,400 MW interconnector. Reversing the charge/discharge state of 700 MW of batteries on charge would immediately have injected enough power to replace the interconnector flow. Batteries have a technical ability to respond within a few cycles (0.2 seconds).
Trying to guard against a 1,400 MW loss with gas plants would require them to run at inefficient partial load, and can ramp up only in minutes, not fractions of a second.
Yet Kathryn just doesn’t mention any of this.
Further, Kathryn is going on the formal figures, but batteries are derated based on limited duration. However, most of them are at least half an hour, which gives a gas plant time to start up to make a significant contribution. So the formal balancing mechanism statistics presumably would not be allowed to count all the battery power available. And Kathryn hasn’t taken this into consideration – likely another 300 MW
There are currently 5 GW of UK grid batteries installed, and presumably a bit more behind meters.
If I was the NESO, knowing that it needed to cope with the possibility of a 1,400 MW load failing, I would do precisely what the NESO seems to have done – have 700 MW of batteries on charge, waiting to reverse to discharge if that interconnector failed. And if the overall margin was tight, I would go for this as it minimises the possibility of a frequency slowdown for a short period of time taking any more generation offline, as it did in August 2019.
That would mean operating a few very expensive (at the time) gas plants at SEL or MEL while charging batteries, ready to reverse the batteries rapidly, so nothing else would notice the disruption.
Presumably the 700 MW of charging counts in Kathryn’s figures as “demand”, rather than under some other heading as negative “balancing supply”. So that means the demand figures reported by Kathryn would have been too high.
The question is, if Modo can identify these situations, why can’t Amira Technologies also identify them?
While the NESO does know what happened in gory detail, the rules may not permit it to disclose full details. In particular, some of the stuff which would have happened “behind the meter” may have to remain confidential. However, the batteries on charge presumably is identifiable from the BMRS data, so Modo, and Amira can analyse this.
I checked the PNs of all BMUs in SP25v and found some batteries were charging. Yes, they could reverse, but how much charge would they have?
The total generation net of charging is in my generation figures in the blog.
It’s hard to get data for non-BMU units but, for example, I know a lot of the BMU STOR assets were running in the market so I guess that would also be tryue for the non-BMU ones. So I’m confident both my margin and reserve figures are right. (Also DM/DR/DC is included in the margin data)
It’s hard to ramp up gas plant when it’s alraedy running. There was spare MEL at Connah’s Quay and Sutton (700 MW available in 15 mins) and Rye House could come back in c 25 mins, but it would have been tough to meet the loss of NSL or Viking. Normally you’d want Dinorwig to bridge the gap but that was also already running.
Also those batteriers would only switch it they have a contract with NESO to do so, otherwise they would carry on collecting the £££ available in the market
We should have kept maybe 4 – 8 coal powered stations open until we had more Nuclear capability
These would be continous generation and would have given much needed breathing space, given that the new Nuclear plants would have taken 10 – 15 years to go live.
And of course, just when we need it, wind generation is still feeble at just 2.58GW
I have always found Modo Energy to be extremely capable in analysing the contributions of batteries to the system, and I read their publicly available content regularly. As a business they make their money by providing tools to battery owners that help them to optimise their returns and to assess how they price and bid into the various markets now open to batteries for ancillary services, as well as offering longer terms forecasts to aid investment decisions and benchmark comparisons of different operating strategies. The nature of their business gives them access to client data at a level that is simply not available in the control room at present. They hire some of the smartest cookies in the electricity system. By the same token you need to be very careful to make sure you have grasped what is being said and what it implies.
Firstly, the data analysed by Kathryn includes the Maximum Export Limits of some 3.3GW of nominal battery capacity as advised by their operators: batteries cannot export more than the maximum they have available which may be reduced from the installed capacity for a variety of operational reasons. Anything not included in that total would be connected in the distribution system, and any discharge would be seen as a reduction in transmission system demand. Secondly battery operators are smart enough to try to ensure that where they take advantage of energy arbitrage opportunities they charge up when prices are low, and importantly aim to discharge as much as possible during price peaks without compromising their overall operation and other ancillary service commitments they may have stacked to the point where they either damage the battery or incur stiff penalties for breach of contract. Any batteries not being dispatched via the new Open Balancing Platform that is designed to ensure that batteries are dispatched through a more automated system that can handle multiple small instructions and reduce battery skip rates (times when batteries might have offered a cheaper way to balance) will have been sure to have been discharging to secure peak prices, reducing apparent demand.
I believe you have misinterpreted the chart from Zachary. He shows the volumes of contractual commitments under various ancillary services and under trades registered with Elexon ahead of gate closure. The purple bars do not cover actual rates of charge and discharge: the other colours show the contingent maximum charge or discharge for ancillary services for large and/or persistent frequency deviations, but these typically require minimal actual battery operation unless there is a frequency event, leaving the battery free mostly to do other things. Some batteries were requested to moderate their rate of discharge, which is not the same thing as charging up. That was indeed creating a bit of headroom offsetting e.g. the use of STOR. Remember that batteries will need to consume about 25% more energy to charge than they will be able to discharge – more as they age, because of round trip losses. In fact, batteries stack their operations: here’s a nice explanation from Modo:
https://modoenergy.com/research/gb-bess-outlook-q3-2024-battery-energy-storage-revenue-dispatch-cycling-duration-optimization
I expect if you were a paying client the article would touch on the opportunities afforded by extra operation in real time where the cost of charging or income from discharging is set by the system price for the settlement period: these swings can be huge, and there were several periods that paid out at a System cashout price of £2,900/MWh for discharging if you weren’t lucky enough to get dispatched in the BM at an even higher price.
By way of example, consider the 49.9MW battery at the Richborough Energy Park next to the substation for the NEMO interconnector
https://bmrs.elexon.co.uk/balancing-mechanism-bmu-view?bmuId=T_RICHB-1&activeTab=Physical&startDate=2025-01-08T00%3A00%3A00.000Z&endDate=2025-01-08T23%3A59%3A59.999Z
You will see that its MEL varied enormously over the course of the afternoon.
IDAU said “The purple bars do not cover actual rates of charge and discharge: ”
Certainly there are no storage BMUs I can identify showing negative FPN (i.e. charging) between 17:00 and 17:30 on 8th Jan. The statement from Porter of Modo was “As the system hit the lowest de-rated margin, some battery positions were reversed to balance the grid, which suggests that Wednesday may not be quite as tight as it first seems.”
I haven’t investigated all the Elexon files I’ve downloaded yet, to check the leeway on all the storage units.
There is certainly a lot of battery activity missing one way or another. UK is supposed to have 5 GW / 7 GWh of batteries installed now, and it is difficult to imagine most of that would not have made itself available between the predicted 17:00 to 18:00 critical period, when prices would have been at a maximum.
IDAU said “Remember that batteries will need to consume about 25% more energy to charge than they will be able to discharge – more as they age, because of round trip losses.”
~25% round trip losses are a more appropriate figure for pumped hydro than for batteries.
Some BESS round trip efficiency figures are available in spec sheets, for instance for the latest Tesla Megapacks. See https://marketingdocuments.energytoolbase.com/marketing/TeslaMegapack2Datasheet.pdf
The spec sheet says the 3 MWh Megapack comes in a 2 hour version with round trip system efficiency of 92% and a 4 hour version with round trip system efficiency of 94%, at 25 deg C, both including thermal management loads. A Modo 2020 document https://www.energy-storage.news/what-is-dynamic-containment-and-what-does-it-mean-for-battery-energy-storage-in-the-uk/ also contains the text “Assumptions: 1MW/1MWh (1h system), 88% efficiency (RTE)” which is in line with the Megapack specs, given the losses from internal resistance will be significantly higher with 1 hour charge and discharge.
Of course, if charging or discharging can extend over even longer periods than the specified duration, then BESS efficiency will be higher still. It is rather more likely that charging can be slowed down, rather than discharging.
The implication is that the batteries can trade profitably at a much lower percentage margin between charge and discharge wholesale prices than can a pumped hydro storage plant, though pumped hydro cycle life is higher. CATL and EV LFP cells typically offer 20 years of 1 cycle per day (7,300 cycles) down to 67% of original capacity.
however the megapack datasheet you linked also says:
Round-trip
efficiency is specified for a full-depth cycle and includes all
power conversion and thermal system losses during the cycle.
and then qualifies the RTE figures with:
1 Nominal energy and RTE at 25°C (77°F) including thermal management
loads, Day 1
Batteries typically don’t perform as well in the cold. So what was the temperature on 8th Jan in the location you have it installed and how did that affect the efficiency? What’s the RTE with a partial charge/discharge as would seem likely to bridge a short term problem like the 8th Jan peak?
Working out actual efficiency vs datasheets is likely to be complex as there are lots of factors whereas datasheets can sugar coat things by quoting the theoretical best cases. Real world will always be different.
Looking at your Megapack datasheet again and what strikes me is the inverter capability, one of these couldn’t support the building where I work as the inverter is too low power vs the total energy storage. It’s output also appears to be 480v 3ph, so that would need conversion to something useful for the grid. Our feed is at 11kV, so how many conversions until it’s back to something useful and what cumulative losses through all of the conversion steps?
Things could be different if you had a megapack in the carpark and could somehow use it’s 480v 3ph output directly, but even then a single 3MWh battery, in context, isn’t particularly useful for us – hence the 11kV genset in the carpark instead.
Iain said “Batteries typically don’t perform as well in the cold.”
While true, this is completely irrelevant. If you think about it, it is grossly misleading in the case of grid batteries!!
Grid batteries are a large investment, and the cycle life is a critical cost parameter. The environmental controls in the container will try to keep them at a roughly constant temperature to maximise cycle life.
Assume a 90% battery round trip efficiency, just to make the maths easier. Assume, the battery is cycled between once a day and once every 3 days, depending on how windy it is and when the wind blows.
The Megapacks are 3 MWh, and the 10% of losses will manifest themselves as heat in the container, so that is 300 kWh generated within the container every 1 to 3 days for a complete cycle. Our lounge is very roughly twice container sized (7m long x 1.6m wide x 3m high), and certainly takes no more than 20 kWh of boiler natural gas fuel per day to keep warm, even in sub-zero temperatures. The Megapack has a huge thermal mass, unlike our lounge, so might take a week or more to naturally cool to ambient temperature after charging or discharging.
The natural conclusion is that most of the thermal management energy is going to be the removal of the 300 kWh of heat around charging and discharging. It would only be necessary to heat the battery if it was charged but hadn’t been used for a week or so – and this is not likely to be the case in winter.
Thus, the lower the temperature, the less energy will likely have to be expended by the environmental control system for heat removal, and the higher the round trip efficiency is likely to be. By contrast, higher ambient temperatures than 25 deg C, such as in summer in Florida or Australia, will require significantly more cooling system effort, reducing the efficiency.
This is the reverse of what happens with electric cars, where the battery is more likely to reach thermal equilibrium with the environment.
Iain said “It’s output also appears to be 480v 3ph, so that would need conversion to something useful for the grid. Our feed is at 11kV, so how many conversions until it’s back to something useful and what cumulative losses through all of the conversion steps?”
I’ve just googled some decent power 450V/11kV transformers, and they are certainly capable of >99% efficiency at a power factor around unity, nnd maybe just under 99% at a PF of 0.8. Any transformer losses will be incurred twice, of course – once on charge and once on discharge. The same probably applies to 450V/22kV transformers. There might need to be two stages of voltage step up, but both could be at 99% efficiency (i.e. 98% through the pair, or 96% for the two-way journey – at least).
There are potential efficiency gains on power through transmission lines if there are batteries involved. If the battery is co-sited with generation at one end, then there are no gains, as the power through the line is not affected. If the battery is sited at the load end, then, although there is a little more power transmitted (because of the 6, 8 or 10% battery losses – already counted), the I^2 x R losses will be lower as the power transmitted is more evenly spread by the BESS.
During that period some 599MW of STOR was utilised, so it was no longer a reserve (as it should have been unless there was a frequency event). However, STOR offered prices may have been cheaper than alternatives so creating some reserve by cutting battery exports may have made at least some financial sense. It did not create additional reserve, because it merely replaced STOR.
I have downloaded the complete metered numbers for batteries for 8th January with any kind of BMUid, which add to the nominal 3.3GW of capacity I mentioned. That is (more or less) what they actually did: more or less because the data do not split out charging and discharging within a settlement period – only the net position is recorded. However, that does show the net power flow in and out of storage, including any activity associated with fulfilling ancillary service contracts. You can access the resulting chart here
https://datawrapper.dwcdn.net/ESxAe/1/
Bear in mind that the data are MWh per half hour Settlement Period, and therefore the average flows in MW are twice as big. You will see that some batteries with activity dominated by ancillary services end up with a net position that goes against the flow of the wholesale arbitrage action. Also, the highest rate of discharge was in SP19, not during the afternoon peak, with discharge spread out into the early evening, as described by Ed Porter, who pointed out that almost all the capacity was in action providing ancillary services or wholesale arbitrage. I did spot that the large complex at Blackhillock appears to have been undergoing commissioning work, mostly drawing a small amount of power, and just running a small scale test of charge and discharge in the afternoon. In any event, some batteries were not able to make all their discharge capacity available, as the MEL chart for the Richborough Energy Park unit I pointed to previously demonstrates.
Please remember that batteries may be installed but not yet grid connected, especially at the current pace of new projects: I’d trust Ed Porter’s figure of 70% of operational batteries being under BMUs. That will increase as more of them come under aggregators operating virtual power plant BMUs because that will pay better. As to losses, I prefer to estimate from measured data rather that a manufacturer Day 1 spec sheet. Needing 25% more energy to charge up implies a round trip efficiency of 80%, which is in the ballpark. I have been monitoring real world grid batteries ever since Hornsdale Power Reserve came on line in 2017: that project has seen its efficiency drop below 75% after they did deep discharges into a market paying A$14,000/MWh, and requiring the replacement and addition of capacity, since when degradation to below 80% has been observed again. The long run average is around 80%. HPR data show charging and discharging separately at 5 minute resolution, so a pretty true figure can be calculated for efficiency. B1610 data mask within period losses where the battery is both charging and discharging e.g. because it is providing ancillary services. Over the 24 hours the effective efficiency across the fleet was 83% as measured by total net discharge divided by total net charge summed across batteries for each settlement period: at some point I will attempt a much longer data run to analyse the UK fleet so there can be no significant influence from initial and final battery state of charge, but it will overestimate the efficiency because it won’t show in period losses. I can’t immediately re-find the analysis I saw that had actual data for a number of GB grid batteries, some of which were showing an alarming rate of deterioration. If I do, I will add a link.
The round trip efficiency of pumped storage can be measured more precisely, since the nature of operation mean that even when providing ancillary services, that will be by variation of speed of charge or discharge, rather than oscillating between charging and discharging. Before the refurbishment at Dinorwig its annual efficiency had dropped to around 73%: since, it is back above 75%. I will be looking at the whole grid performance in 2024 soon, but to begin with I have to cope with missing and glitchy data. That involves looking at the 100,000 + 5 minute periods for clues, so it will take some time.
Great analysis Kathryn. I thought it would be helpful to compare NESO’s Winter Outlook with reality so far.
Their de-rated capacity calculations are the stuff of fantasy and the peak demand estimates are a work of fiction.
More here:
https://open.substack.com/pub/davidturver/p/neso-margin-call?r=nhgn1&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false
Can you predict when the system will fail ? You could assign a probability of failure of the single largest in-feed loss and multiply this with the probability that the reserves are overstated. example 50% chance of loss of largest feed x reserves overstated by 50%. If you plotted this over the next 5 years you could show how soon the system is likely to fail. Rather than analysing MWs I think you [or someone] needs to be independently modelling risk. The risk of failure of each part of the system will be well understood.
Kathryn, can I be impertinent and ask you to clarify something for me, please?
Cannot load be shed in a controlled way? – e.g. by prior consent of major industrial users -thereby averting blackouts in more sensitive areas such as domestic, street lighting, hospitals etc.
You may have addressed this somewhere in your recent (and, I must say, excellent) output, if so I have missed it.
Thanks, Steve
It can be controlled regionally and crtitical infrastructure is protected, but otherwise not. If they decide to turn off N Wales, everything in N Wales will be cut off except anything critical (if possible)
Thank you Kathryn, that’s helpful.
The popular narrative (echoed by NESO) holds out batteries and interconnectors as the cure for the building feast and famine of electricity production. In reality they are difficult to control and optimise. They act in their own commercial interest which does not necessarily align with system need. Why were batteries charging over a very tight winter peak?
the proliferation of players in the market presents plenty of opportunities for mercenary behaviour – the good of the grid, the country, or consumers isn’t a factor, only their own interests and ultimately their bottom line.
This post by NESO is pretty detailed – doesn’t it answer all the questions raised?
https://www.neso.energy/news/what-happened-margins-8-january
No it doesn’t. It’s designed to deflect anyone without enough detailed understanding of the realities. It’s a cover-up. That is why Kathryn wrote this blog, unmasking the questions it doesn’t address, some of which we still have no answers for..
Can anyone point me to a guide on determining what size of stand alone generator I need for a domestic property? Think I am going to need one next winter.
Also what happens when the grid does go down? Do we have enough large, spinning units to produce the “inertia” to get the grid going again?