Wind curtailment

Wind curtailment is costing the UK tens of millions of pounds every year. [1] The full potential of wind energy or even the existing wind generation infrastructure is not being harvested due to grid constraints, spatial distribution and lack of storage. A lot of good analysis has been done on quantifying the costs of curtailing wind generation, and I wanted to show the problem within its context: why and how the infrastructure challenges and other factors are forcing wind farms to not use utilise full capacity.

What is wind curtailment, and why does it happen?

Wind curtailment refers to the reduction of generation for wind farms. This occurs when wind generators could generate more power, but the system operator instructs them to reduce production, because too much electricity being generated would cause the grid to become imbalanced. Due to how the UK's Balancing Mechanism is set up, it actually costs money to turn down wind farms. To balance the grid, it often is the case, that Scottish wind farms are instructed to reduce generation, while other sites (such as gas) are ramping up generation elsewhere to address excess demand in the South.

The Balancing Mechanism

The Balancing Mechanism (BM) is the National Energy System Operator's (NESO) way of ensuring the grid is always in balance (i.e. the supply equals the demand when considering all sources and sinks connected to the grid), by buying and selling energy from and to sites registered to the BM. This buying and selling takes place every 30 minutes, with sites reporting Physical Notifications (the preferred generation or output for the site, e.g. a wind farm), as well as bid-offer pairs which communicate how much it would cost the NESO to curtail the site (sell energy) or ramp up generation (buy energy). At first sight the grid imbalances don't look too large (this is the system operator's success forecasting the demand well - the blue line), however, when looking at the extra generation and curtailment bought by the operator, we get an interesting picture. These are magnitudes larger than the simple imbalance. What happens here is that keeping the grid in balance isn't as simple as summing up generation and demand: location plays a huge role. Due to difficulties of transferring electricity, a unit of generation often needs to be cancelled, and then readded to the system elsewhere. What we see above is the imbalance, curtailment and extra energy bought by the operator in October 2025 (in GWh for every 1-hour time interval). What's outstanding is that the red and black areas both cost money: the operator pays for the extra generation, but also for curtailing a generator. Hence the actual cost of keeping the grid balanced is multiple times that of the simple aggregated imbalance price due to the grid bottlenecks.

Wind generation and curtailment at present

To understand curtailment, Elexon's data on the Balancing Mechanism can be used. We can get the curtailment figures as the difference between the accepted levels and the (final) physical notifications (FPN) - i.e. subtracting the level of generation that was instructed by the system operator (accepted) from the level the site would have preferred running at (FPN). Note that we are looking at negative difference here (=curtailment), we will also look at where the extra energy comes from below. After some data wrangling, this data per BM unit can be linked up with the government's database on renewable energy sources, which gives us richer data including their installed capacity, location, etc. [2] Having this data, we can aggregate curtailment and understand what proportion of wind energy has been cancelled in various parts of the country. It clearly is a problem mostly impacting Scottish wind farms, with some counties (Shetland Islands - Viking Wind Farm, Angus - Seagreen or Aberdeenshire - Moray East) particularly impacted. Note, that I have aggregated all offshore wind farms to the region or county at which they are connected to the grid. This is important because many sources (such as the NESO) separate offshore wind farms into its own category, but for understanding how the grid bottlenecks impact generation, it makes sense to aggregate to onshore counties.

Grid bottlenecks and the problem of location

Two important factors contributing to the curtailment of wind energy are spatial distribution and grid infrastructure. The National Energy System Operator's (NESO) ten year statement outlines the main bottlenecks, and developments to be carried out in the coming years. How do these relate to wind farms, and how do the sites' spatial distribution impact their ability to supply the generated electricity to the grid?

The notorious B6

One of the key boundaries identified by the NESO for upgrade is B6 - the one between England and Scotland, also the boundary between the two grid operators National Grid (in England) and SSE (in the South of Scotland). This boundary has a capability limited to 6.7GW, and is highlighted as the one with the most excess flows in the coming years if no upgrades are made. [3] Toggling the view, and changing the size of the circles raises a few interesting points. Comparing for example Viking (in Shetland) with Triton Knoll (off the coast of East England) we see that even though the nameplate capacity of the latter is about ~2 times higher (857MW vs 443MW / 370MW [4]), Viking has generated slightly more output than Triton Knoll, and that is with Viking's curtailment being one of the highest in the country: ~64%. [5] The impact is not uniform, however, and curtailment clearly can't be explained only by location or grid bottlenecks. A perfect counter-example is the case of Beatrice vs Moray East and West. All three wind farms are in the same location, with comparable capacities, connecting to the grid in the same place, yet Beatrice has a curtailment of 0% for 2025 as opposed to the ~50% of the other two. This might be related to Beatrice's fine last year relating to their pricing and "overcurtailment". [6] Using this partitioning of wind farms (above and below the B6), we can plot the curtailment percentage against the power generated. In the below histogram the power generated is averaged across 15-minute intervals, then summed and binned. E.g., there were 729 different 15-minute intervals so far this year during which the power generation in Scotland from wind farms was between 9.5 - 10.0 GW, and for these cases the average curtailment was at 48.1%. The difference between the aggregates below- and above the B6 is crystal clear: curtailment below B6 is negligible compared to above it. The dashed line gives us a point of reference: it represents the capacity limit of the B6 boundary, which is 6.7GW [7]. This is not taking into account 2 other factors though: Scottish internal demand, the peak of which is at ~4GW [8] and other sources of generation [9]. The dashed line represents the curtailment ratio that would be necessary if all of the wind energy were to be exported southward (which would be the case if internal demand cancels out other sources of internal generation in Scotland). Obviously this is only good as a vague point of reference without exact demand and generation figures, yet it shows that Scottish wind farms are curtailed even above this bottlenecked level.

Wind in the North, Gas in the South

Now that we have looked at curtailment, let's take a look at extra energy bought by the operator. On the imbalance chart we have seen that the seemingly small imbalance is the aggregate of a fairly large curtailment and extra energy bought - where is this latter coming from though? The arrows on the map capture the locations of extra generation (up arrow) and curtailment (down arrow). Coloured by fuel type, and scaled by the difference from the physical level, they show a powerful map of where the curtailed wind energy is replaced: mostly in the South, and mostly by CCGT (i.e. gas turbines).

Conclusion and solutions

Wind curtailment is already costing billpayers tens millions of pounds a year, and given the huge renewable energy projects in the pipeline in Scotland [10], it will remain a burning issue. Steps have already been taken to make sure new sources are connected to the grid faster [11], and other steps have been taken to make the grid upgrades more simple with less legal overhead [12]. Moreover, the NESO treats this as a priority, with significant upgrades being planned to expand capacity on boundaries such as the B6 [13]. There also are other suggestions that do not involve major infrastructural changes, such as nudging people to use their energy flexibility [14]. In part 2 of this series coming soon, I will look at the cost aspect of balancing the grid, and the quirks of bids, offers, and acceptances.