Electricity pricing

When talking about wind curtailment and the UK electricity markets, there is one key aspect we can't get around, and that is pricing. In the first part of the series, I have looked at the curtailment of wind farms, and how the balancing mechanism replaces them with (mostly) gas in the south. Now we will look at the role pricing has in this, and what it costs overall. Navigating the site: most of the visualisations are dynamic, so you can get the most out of them, if you hover, click around, zoom and interact with the data being presented. This post also contains a considerable amount of jargon, and there are

explainers

that you can hover over for each term that is underlined by a dotted line.

We curtail wind. How much does it cost?

Let's take a look at what each of the wind farms are paid for being constrained (i.e. not generating energy). The map is familiar from the previous part, but instead of energy volume, now it's the cost of curtailment that the sizes show, aggregated yearly for the past five years. Some of these numbers have been reported on widely, Seagreen in particular [1], but not many of the others, such as Moray East and West that are outpacing the former in revenues made through not operating. We can also see Beatrice topping the chart in 2021 and coming second in 2022: something that we will see in more detail in this section. Figure 1: Yearly aggregate curtailment cost by wind farm These outstanding numbers are just one side of the story though - when wind farms are curtailed, this is often done to avoid overloading parts of the transmission system (note, how most of the large values are in Scotland, for reference, read part 1). When this is the case, the electricity that's missing due to the curtailment, needs to be replaced. This usually happens in the South, where generators are not constrained by the

transmission bottlenecks

. This replacement of electricity is creating another bucket of extra costs for the billpayers. For the

upward

costs we can only have an estimate: we won't know which bits of electricity generated were to replace curtailed wind, and which weren't. What we can do, however is a pretty educated guess: by taking the average price of

extra generations

in a

settlement period

, and multiplying by the amount of wind energy curtailed in that period we get an approximate value for what it cost to replace the electricity lost. (See more details in the appendix) As we will see later, most of the replacement comes from gas, which explains the dip in 2023: as gas prices collapsed, it heavily swayed the total cost of curtailment + replacement in that year, despite the amount of wasted (and then regenerated) energy growing steadily. Some bittersweet news on the other hand: even though the wasted wind energy increased to 10.12TWh in 2025, the direct cost of it to the consumers has been less (£383M compared with £394M in the previous year). This gain evaporates however, when we consider the cost of replacement, which pushed the total figure to a whopping £1.55B. Figure 2: Yearly total curtailment and extra generation cost

The cost of replacement and skip rates

The balancing mechanism (BM) requires every site to submit prices for every half an hour: these are 'bid' and 'offer' prices submitted in pairs, reflecting the 'price of turning generation down' and 'price of turning generation up' for that period respectively. These are the prices that are paid by the

System Operator (SO)

to the site when they are instructed to increase or reduce generation. This is key in understanding how the costs of balancing and curtailment are made up, for more explanation on these, see the appendix. Balancing the grid is a very delicate task: the location, capacity, ramp up time of sites need to be taken into account as well as the price on their bids and offers and

transmission constraints

to only name a few aspects. Skip rate refers to the rate of offers that could have been accepted by the SO based on their

"merit"

, but weren't - essentially the missed opportunities to get electricity on the grid that's cheaper and cleaner [2]. In particular, battery storage (BESS) in the UK has been suffering from this, whereby many BESS sites are being ignored even though their offers hold more merit. Plotting the distribution of fuel types in the extra generation in the periods when wind was curtailed can be used as a proxy for the upward balancing energy that's replacing wind. Let's start with the good news: coal is gone by now,

BESS

has been picking up, especially since the end of 2023, and the summer months can mean increased, greener alternatives to ramp up generation. Yet, the overwhelming majority (between 80-90%) of the generation is still gas, Closed Cycle Gas Turbines (CCGT) in particular. Figure 3: Distribution of fuels used to replace curtailed wind Now, looking at the same as we did with the wind curtailment costs on Figure 1, but from the perspective of "up instructions", i.e. all the extra electricity generated, we can plot the same two charts. Note that from now on I'm describing all positive balancing energy, i.e. not restricting it to the electricity that was generated to replace the curtailed wind. Looking at the top 20 sites that made the largest cashflow from their offers being accepted, we see 18

CCGT

sites in 2025, with only one biomass and pumped storage each. What's clear, is that the leaderboard is very stable across the years - it is mostly the same sites that are instructed to ramp generation up in the South and the Midlands when the grid requires so. Furthermore, as expected (and discussed in part 1), all of the top 20 extra generators are located in the Midlands, South and Wales - there isn't a single one in the North of England or Scotland. This is due to the imbalanced demand in the UK exacerbated by the transmission bottlenecks. Figure 4: Yearly aggregate cost of upward balancing generation by site

Skip rates and batteries

To visualise the skip rates, I've selected 3

BESS

sites from the 20 biggest "most ignored" sites in 2025: those sites that have consistently submitted offers priced in the range where the SO accepted others, but were rejected nonetheless. In addition, I've picked these three because they are in the South and Midlands, i.e. no boundary constraints apply to them (e.g. Lockleaze is just next to Seabank Power Station near Bristol, which is one of the most frequently utilised generators for balancing upwards). The circles correspond to offers submitted by sites - those that are grey, have been accepted, while the coloured ones belong to the 3 BESS sites. The sizes of the circles correspond to the level of the offer (i.e. how much energy is offered at this level). What's clear, is that these sites get ignored a lot. This is down to a number of reasons (mostly generally true for BESS sites), and some are already being tackled by the NESO [3]. If we look at Lockleaze for example, we can see the same pattern every day: out of its 17376 offers in 2025, 0 have been accepted. Its offers were on average £106 cheaper, than the maximum accepted price in the corresponding settlement period. Figure 5: Accepted offers and the offers of the 3 BESS sites for each period It's also clear, that the units can be provisioned - such as Hunningley Stairfoot

BESS

between periods 20 and 27 on the 1st of January. There doesn't seem to be a generic, fundamental reason for ignoring these sites. Of course, it isn't this simple: batteries cannot be used for generation continuously, they need to be recharged. So clearly, the 17376 offers being rejected are well above the theoretical ceiling - there is no way the BESS site could have generated for the whole year non-stop. Regardless, skip-rates show an existing problem that can be improved. So far we've only looked at the number of acceptances, but at least as important is the dispatch time: i.e. even if accepted, for how long does the SO provision the site? This is traditionally a lot shorter for BESS than other types of generators. The left plot describes the distribution of the dispatch length for each month since 2021, whilst the right hand side aggregates these dispatches into a total number of minutes. As we can see, the launch of the Open Balancing Platform (OBP) in 2023 [2] significantly increased the provisioning of BESS sites. What we are still seeing though is that these are very short periods: in the last month of 2025, the average battery dispatch was only 4 minutes. Figure 6: Dispatch time distrubution of BESS sites and total dispatch time per month

The cost of curtailment

To talk about curtailment costs in detail, we first need to understand how electricity is traded. Electricity is usually traded between Generators and Suppliers. [4] The latter are companies that distribute electricity to their customers. [5] The parties contract by signing a Power Purchase Agreement (PPA) and agree on a price and volume for each settlement period. What's crucial is that this is independent of the balancing mechanism: when turned down, a station can still sell power as if it were generating. [6] This is why the NESO will pay other generators to turn up if needed, to keep the grid in balance. The

bid price

a generator submits is reflecting the costs of turning down. Now, because of the above, the bid price is not equal to the wholesale price. The generator was still able to sell the energy it contracted despite being turned down, so there is no cost occurred by missing revenue from the PPA. Since this is the case, some generators will actually pay the System Operator to turn down: sites such as gas turbines will have 'negative cost', i.e. will save by being turned down, because they are not burning fuel. Then why is the SO paying wind generators to turn down? Because there are other costs to the site: in case of wind farms, an example of this is the foregone subsidy on the generated electricity. [7] Unlike the above, the subsidy is paid on the electricity generated. One type of subsidy comes from a scheme called Contracts for Difference, or CfD. [8] The Contracts for Difference scheme is meant to incentivise renewable generation projects, by fixing a strike price at which the site will be able to sell its energy after opening. It is a smart scheme, that aims to derisk the development of renewable sites. To make things fair, if the wholesale price of electricity goes above the strike price, the site will still only be entitled to the strike price. This is a classic case of hedging, the aim is to take the risk of electricity price volatility off the sholders of wind farm developers and owners. In practice the strike price is implemented through a top-up from or payback to the Low Carbon Contracts Company (LCCC) that is managing CfDs: if the site was able to sell at a price higher than the wholesale price, it pays the excess back to the LCCC and vice versa, if they could only sell at a lower price, they get a top-up from the LCCC. [9]

A case study on Beatrice

Beatrice has been one of those wind farms that have made excessive profit on being curtailed, according to the Ofgem investigation launched into its bid pricing practices. [10] The main claim of the report is that Beatrice has placed a cap on its bid prices, and charged more for turning generation down, than its actual costs of doing so would have suggested. Beatrice is one of the wind farms under a CfD and its strike price in 2022 was £175.47 [11] What we are expecting from its bid prices then is that they will follow the shape of the wholesale price for the period: the higher the wholesale price, the lower the subsidy. In the extreme case of very high wholesale prices it might even happen, that it is better for Beatrice to be turned down: they have already sold the electricity at this high price, and were they to actually generate it, they would have to pay the difference back to the LCCC. What seems to be the case, however is that in 2022, bid prices of Beatrice didn't continue to follow the wholesale prices, instead they are capped at the -56.48 £/MWh level. If we put a ceiling on the wholesale price chart at £175.47, the two become very similar. This means, that Beatrice priced its bids without taking into acount the repayment to LCCC: were they to generate in these periods of high electricity prices, they would have had to make payments to LCCC. This is one of the extreme cases, where a generator is better off not generating: this however wasn't factored into the bid prices, hence Beatrice making a sizable profit through being curtailed. Figure 7: Beatrice bid prices and the system price As we can see, this behaviour was indeed restricted to 2022; the wholesale price peak of early 2025 is followed on the bid price chart by a peak of (positive!) bid prices well above the 2022 ceiling of -56.48 £/MWh. Beatrice has agreed to make a payment of £33.4 million to Ofgem's consumer redress fund, and we can see on figure 1, that it has dropped from 2nd place in 2022 with £33 million made through being curtailed to 70th place with only £0.6 million made the same way in 2025.

Is it only Contracts for Difference then?

Clearly not - currently only 29 wind farms fall under Contracts for Difference [12], and there are other schemes, such as Renewable Obligation, and wind farms can also be unsubsidised. Renewable Obligations: even though the scheme was closed in 2017 and replaced by CfD, a large part of existing renewable generators (including wind) still fall under it. The RO was a more complex scheme that involved both suppliers and generators. Under the RO, generators are given certificates (or ROCs) and a 'multiplier' on their generation based on the technology (e.g. offshore wind gets 2.0 ROCs, onshore wind gets 0.9 ROC). Suppliers that get their energy from these generators, need to present a specified number of ROCs to Ofgem. If they do not meet their obligation, they need to pay into buy-out fund at a specified price. ROCs can be traded as well, creating a market. Eventually these costs are indirectly passed onto the consumers of suppliers that could not meet their obligations. [13] Given their complexity, and the fact that these contracts have been replaced in 2017, they are outside the scope of this post. Wind farms can also be unsibsidised, meaning they behave as regular generators would on the electricity market. There is also a special case: Moray East having won a CfD in Allocation Round 2, didn't seem to have implement it after coming online in 2021. [14] As we have seen with Beatrice above, wholesale prices in 2022 meant that wind generators under CfD would have had to pay back considerable sums of money to the operator (if the wholesale price was above their strike price). By not implementing the CfD, Moray East ensured that they get paid the wholesale price and that when curtailed they don't need to take into account the difference between the market price and strike price. As we have seen above, Moray East has made the biggest profit on being curtailed between 2022-2025. Ofgem has opened an enforcement case in 2025, so we will soon find out whether the Transmission Constraint Licence Condition was broken here, or not. [15]

Conclusions

CfDs and similar schemes have been set up to create incentives and reduce risk in a highly volatile and complex market. They have been very successful to secure developments, and increase renewable capacity. As a byproduct, through pricing and costs, a new incentive has been created, to improve the grid and reduce the bottlenecks: the savings we can achieve through them are in the billion pound order of magnitude at this point. Batteries on the grid are an increasingly viable option for balancing; they are quick to dispatch, green, cheaper and more and more available. As we have seen the NESO has already made huge steps towards the better utilisation of this technology, but more can be done: by improving skip rates on batteries, batching small orders instead of preferring large provisioning will further reduce the CO2 impact of balancing, reduce energy bills and make the grid more robust. Co-location battery storage is a great solution for the two issues we have looked at above: it can reduce the need for curtailing green generation by loading batteries when the system is long (i.e. there is too much electricity being generated), and they can act as balancing whenever there is less generation. Of course, co-location BESS cannot replace grid improvements: in cases where the electricity flow is practically one-directional (as is the case on the

B6 boundary

), the co-located batteries will struggle with the same bottlenecks as the generators. Transmission upgrades will be crucial, and need to be sped up in the coming years: the generation infrastructure has far outpaced grid improvements in the past decade.

Technical bits

Bids, offers, acceptances - untangling electricity pricing in the BM

Units that are part of the Balancing Mechanism (BM), i.e. those that can be directed to turn down or ramp up generation, or demand, need to follow a structured bidding process. On a high level, this involves submitting pairs of bid-offer prices that describe what the System Operator would have to pay for extra generation or curtailment of the unit. The format is quite particular, which requires a bit of wrangling: Figure 8: Bid-offer pairs submitted for a settlement period After visualising this, it becomes clear how the charge is calculated for bids and offers: it's the area between the step function and zero. I.e. if the requested

level

is above the first price, the second is activated for the remaining levels and so on. As we can see, below 0 the bid price is used, above 0 the offer - the point of providing these as pairs is that it provides an "undo" mechanism for the System Operator (SO), meaning that if for some reason the SO wants to cancel an acceptance they have already confirmed, they can do so by buying the other side of the bid-offer pair.

System prices

As mentioned above, to estimate the cost of replacing the curtailed wind, it isn't the systemBuyPrice that I have used, but rather the average extra generation cost. This I have calculated by taking the cashflows of all the sites that were instructed to generate extra energy in the period, and divide it by the total extra energy added to the grid. Since the system price calculation is rather complicated, and takes into account a bunch of different types of generators too [16], this gives an estimate anchored more in the actual generator earnings, than does the system price. Figure 9: Electricity replacement price and system price It is clear, that the two move together (as the replacement cost is a factor in the system price, that is expected), but the replacement price seems to be generally higher than the system price. Balancing does in general come with a premium, which is paid for the short notice of the instructions, but it seems like the up instructions are pulling this weighted average upwards.