Challenging the myths of the renewables and interconnectors model

In the past few years, we have been repeatedly told by the system operator, now known as NESO (National Energy System Operator) that the GB power grid can be underpinned by a combination of renewables (wind and solar) and imports from other countries. NESO and its predecessor organisations have been a huge cheerleader for interconnectors. The Department for Energy Security and Net Zero (“DESNZ”) is on the same page. The idea is that when wind and solar output are low in Britain, the shortfall can be largely made up with imports.

As I have described previously, there are some significant problems with this assumption, primarily the high weather correlation we have with our connected markets which means that they may also experience shortages of weather-based renewable generation at the same time that we do – this was the case in early November when Britain and many of its neighbours experienced dunkelflaute – dull, still weather which is terrible for wind and solar generation.

Another key problem is that exporting electricity in general causes electricity prices in the exporting country to rise. Ofgem has said that once GB becomes a net exporter of electricity, there will be a consumer dis-benefit and as a result it has rejected almost all of the proposed new interconnector projects with Continental Europe in Window 3 of the Cap and Floor regime.

The problems go further than this. The contribution of both renewable generation and interconnectors to meeting demand in GB is often lower than advertised.

Wind load factors are low and are not rising

In its Energy Generation Costs 2023 report, DESNZ claims that the load factors for offshore wind will be 61% in 2025 and higher in future years. This is in direct conflict with other Government figures quoted in the Digest of UK Energy (“DUKES”) reports which show that offshore wind has a load factor of around 40%.

However a lot depends on what is understood by “load factor”. Some analysts look at the output of the windfarm at the asset level and define the load factor as the amount of electricity generated per unit of capacity. So if a 10 MW wind turbine produces electricity 35% of the time it would have a load factor of 35% and generate, on average, 3.5 MWh/h. However, what matters to consumers is not how much electricity a wind turbine generates, but how much of that electricity is delivered to them (or to the grid more generally). This is a lower figure because Britain lacks the grid infrastructure to efficiently utilise this electricity, so often wind turbines are forced to curtail – ie reduce – their output. Analysis of the actual contribution of wind to demand illustrates this.

I analysed the contribution of wind to meeting demand from the start of 2024 to the end of November, using data from BMRS. Assuming 6.5 GW of embedded wind (as set out in NESO’s Winter Outlook), and a total amount of installed offshore wind of 30.163 GW (which was the year-end capacity quoted in DUKES 2023 – clearly more has been added during 2024 but using the figure for the end of 2023 would be conservative for the purposes of this calculation), I found that the amount of transmission-connected wind meeting demand in 2024 was just 30% of nameplate capacity.  

In September, Net Zero Watch wrote to DESNZ requesting an explanation for the 61% figure used in its Energy Generation Costs 2023 report. A response was finally received in mid-November after the matter was raised in a House of Lords debate, in which the following explanation was offered:

“To enable comparison across technology classes it is standard for LCOE estimates to be calculated assuming that they operate at their technical maximum. This differs from actual annual operation that accounts for all reasons for wind plants to be operating at less than maximum capacity. This will include periods of maintenance and curtailment for example, which varies across years and across projects. Another factor is that the Generation Costs Report considers new turbines where improvements in turbine design and larger turbines (higher hub height) enable increases in load factors. Hence, we expect a higher load factor for the newer models assumed in the Generation Costs Report than for the range of turbines in the existing fleet that you refer to in your letter,”
– Jenny Inwood, Energy Infrastructure and Markets Analysis Team, DESNZ

Oh dear! There is so much wrong with the explanation.

Firstly, suggesting that the technical maximum must be used to allow for technological comparisons is nonsense – what matters, particularly in the context of costs to consumers, is what the technology ACTUALLY produces, not its technical maximum.

Secondly, as Net Zero Watch points out, Levelised Cost is supposed to denote the lifetime cost of the generation divided by the lifetime amount of generation. By using the technical maximum, DESNZ assumes the windfarms always produce at their maximum level which is clearly nonsense and counter to the definition of “levelised cost” in the  first place.

Thirdly, the idea that larger wind turbines will be developed producing greater load factors is highly speculative – the trend for larger turbines has stalled in the face of significant warranty issues faced by turbine manufacturers as larger turbines fail to perform. This difficulty is easy to understand – the larger a turbine, the greater the distance between the tips of the blades and the higher the chance they will experience different wind speeds. These different speeds impose significant stresses on the blades causing them and their mountings to fail. Although some larger turbines are being developed in China, Western OEMs are shying away from increasing turbine size, so a conservative approach would be to wait to see signs of this changing and not simply assuming that it will.

Net Zero Watch has pointed some of these errors out to DESNZ and awaits a response.

Interconnector availability is lower than expected and falls with age

The problems I have previously described regarding reliance on interconnectors relate to the availability of electricity for exports to GB, or the desire of exporting countries to continue to export. But there’s another factor in play and that is the reliability of the interconnectors themselves. And they are also interesting to consider.

The table below shows the de-rating factors for interconnectors in the next capacity auction. There is the range which ESO (now NESO) suggests, and the values chosen by DESNZ for the auction. The Capacity Market de-rating factors for interconnectors take account of both the physical capacity of the cables and the availability of spare generation in the connected markets for export to GB.

It’s particularly interesting to look at the values for IFA and IFA2 which are virtually identical despite IFA being many decades older than its newer cousin, and being actually available a lot less. It’s reasonable to assume that the spare generation parameters for both are the same since they originate in the exact same place in France, so the difference in de-rating factors should reflect the difference in physical availability, but this does not appear to be the case.

I analysed REMIT availability data for all GB-connected interconnectors (excluding Ireland which I consider to be an additional source of demand on the GB grid) over the past three years. These data are difficult to find since the interconnectors can choose where to report their REMIT notices and can also change as Eleclink has done in the past year. So some of these interconnectors report to Elexon (BMRS) and some to NordPool. ENSTOE also has data but they are not always consistent with the other two which are the ones I relied on for this analysis.

As an aside, I have asked Ofgem to consider requiring all GB-connected interconnectors to report their availability to Elexon, a request I understand Elexon has also made, in order to improve the transparency of market data relating to the GB market. It would also be good if Elexon would report availability data in percentage terms, since extracting the data for my analysis was very time-consuming. I would have liked to go back further but the data gathering simply takes too long. If any of my readers have these data and are willing to share, I would be happy to update this post to include them.

The table shows the percentage of capacity that was available ie not reported as unavailable on REMIT, for each interconnector.

Of course, selected years can be affected by specific improvement works for example, a synchronous condenser is being built at the Sellindge connection point for the French interconnectors, which has led to reduced availability over the past year as various cables must be disconnected while the equipment is tested. And in its first months of operation, the capacity on Viking was restricted by half due to constraints on the Danish grid.

However, it is clear that IFA has much lower availability than IFA2, which is to be expected, given its greater age, so it is difficult to understand why its Capacity Market de-rating factor is just 1% lower than that of the newer IFA2.

Two other things jump out from these data. The first is just how variable some of the interconnectors have been over the past three years. While some of this may be due to projects such as the synchronous condenser at Sellindge which should improve overall availability, it seems that there are always ad hoc “issues” affecting flows.

In 2016 a ship’s anchor severed half of the eight cables making up IFA, putting them out of action for over a year. Around the same time a large part of the French nuclear fleet was offline for inspections as a result of the discovery of the steel carbonisation fraud. In 2022 large parts of the French nuclear fleet went offline again as a result of the stress-corrosion problem. In each case, France went from being a net power exporter, to an importer, changing the balance of flows to and from GB. This suggests it would be unreasonable to discount apparent “one-off” disruptions, since these “one-offs” occur with some regularity.

Another thing which jumps out is the narrow range of de-rating factors determined by DESNZ for the Capacity Market. Despite the differing ages of the links and the market differences, and the very wide range supplied by ESO, DESNZ determined all interconnectors should be de-rated at roughly two thirds of their nameplate capacity. This feels more like a rule of thumb than something scientific. Of course, the Capacity Market should indicate availability in times of system stress and not average availability. It is quite likely that system stress would coincide with both ad hoc availability problems and limited spare generation capacity in the connected markets eg a widespread dunkelflaute.

Trying to model spare generation capacity across the European markets is beyond what I have time for, but I have looked at the extent to which GB exports during times of high GB system demand.

When I looked at this in 2019 I found that GB exports during 12% of the hours with the top 5% of demand. In 2022-23 this rose to 23%, reflecting the extensive French nuclear outages which saw France switch from being a net electricity exporter to a net importer. So far in 2024 (using 5-minute data rather than hourly), the frequency of exports during the periods of highest demand has been more consistent with Winter 2019 at 13%. (2020 and 2021 have not been analysed due to the impact of covid on demand.)

In order for interconnectors to genuinely support security of supply it would be better if exports during periods of high demand were minimal, but 13% of the time is almost the equivalent of one day per week (14%) which is not insignificant.

.

So what does all of this mean? Clearly both renewables and interconnectors contribute less than the Government would like us to believe, and in the case of interconnectors, even Ofgem has warned that being an electricity exporter is not good for consumers. Perhaps the countries that currently export to us will realise this as well and reduce their enthusiasm for cross-border trading.

The approach to renewables is arguably worse. Overstating load factors benefits no-one since the contribution to the grid cannot be faked – either windfarms are generating electricity or they are not. If they are not, other generation or imports need to fill the gap. Expecting windfarms to contribute twice as much as they actually are means these other supplies may not be procured in the required quantities, threatening security of supply.

It also means that the cost of windfarms is higher per unit of electricity generated than the Government would have us believe. If they are only running half as much as advertised, the capital cost of the electricity they generate doubles. This translates into poor value for consumers, undermining the “cheap renewables” narrative still further.

It’s ironic that the term for the psychological manipulation of people to make them believe something they would otherwise think is false is known as “gaslighting”, since these failures mean we will continue to rely on actual gas for lighting for many years to come.