News

Callum MacIver is a Research Fellow in the Department of Electronic and Electrical Engineering at University of Strathclyde. Keith Bell holds the Scottish Power Chair in Future Power Systems at University of Strathclyde. Both Callum and Keith are active in the UKERC research theme on Delivering Energy Infrastructure.

The curtailment of output from wind farms due to lack of network capacity and the associated costs of balancing the system have become a key focal point in the current debate in the electricity sector about the extent to which the system is “broken”. They are also a key driver behind calls being made in some quarters for radical market reform to better align what the market dispatches with the physical realities of the system, i.e. to reduce the need for the system operator to take balancing actions. In this blog, we set out to dig a little deeper into the current state of wind curtailment. In particular, we seek to examine the role of transmission system availability (or rather unavailability), something that is often absent from the discussion. Spoiler: It is very influential! But first, some background is required.

What is wind curtailment?

The National Energy System Operator (NESO) is tasked with balancing generation and demand in real-time in Great Britain (GB) via the Balancing Mechanism (BM). The market provides NESO with a set of half-hourly generator, storage and interconnector ‘physical notifications’ based on those actors’ positions in forward, day-ahead and intra-day trading markets. At ‘gate closure’ – 1 hour ahead of each delivery period – NESO takes over. Their task involves managing any errors in forecasts of supply or demand, or subsequent changes in circumstance. It also increasingly involves unravelling any market dispatches that are not physically feasible within the system, respecting various constraints such as limits to the amount of power transmission lines can carry without becoming too hot, i.e. thermal constraints, voltage limits and a need for a certain minimum amount of system inertia. Costs arise due to each of these factors, but thermal constraints are the single biggest driver of the high and increasing costs for system balancing, in the main due to the need to curtail surplus wind energy in export-constrained areas of the network. This usually means turning wind farm output down in Scotland, because we can’t safely export it south, and replacing that energy in the South, typically with gas generation. This comes at a premium to consumers. The curtailed wind farms still get paid for their original transactions in the market; they might then get paid some more for turning down in the BM, e.g. to cover any lost income from the government-backed Renewables Obligation or Contracts for Difference (CfDs) they miss out on by not being able to generate and which their initial business case – and price that they bid into CfD auctions – assumed they would get. The gas generators also then get paid to generate more and, because this is at short notice, they often charge a premium over their ‘short-run cost’ (the cost of the gas) for doing so. This typically results in the order of a 30% uplift in asking price compared with what they might have charged in the day ahead market.

Constraint costs

System balancing costs caused by thermal constraints are generally stated as the sum of the ‘bids’ to turn down (usually wind) and the ‘offers’ to turn up (usually gas). Each of these are, in principle, competitively priced in the BM. NESO tries to choose the cheapest bids and offers to minimise consumer costs.

Balancing costs have been on a rising trend in recent years. A number of fantastic public data sources developed by curious individuals are now tracking and bringing to life the vast and often otherwise impenetrable market data that is made available by Elexon (the entity that oversees settlement of BM transactions). Below, we make use of one of those resources from Robin Hawkes to show the combined total balancing costs in each year since 2016 alongside the cost per MWh of curtailed energy (turn down + turn-up actions).

Figure 1 – Balancing volumes and costs since 2016 as processed from Elexon data at https://renewables-map.robinhawkes.com/curtailment

Figure 2 – Balancing costs per MWh since 2016 as processed from Elexon data at https://renewables-map.robinhawkes.com/curtailment

Many stories emerge from this data. In one sense, it provides confirmation of one widespread contention: that under a ‘connect and manage’ philosophy (a government policy first introduced in 2009 to allow wind farms to connect before the main network infrastructure had been developed to accommodate them[1]) we’ve built a lot of renewables, particularly in Scotland, without at the same time undertaking sufficient upgrades of the North to South transmission network. Thermal constraint volumes are certainly on an upward trend, but, perhaps surprisingly, it is not a straightforward story of inexorably rising constraint volumes, and therefore rising costs. After a jump in 2020, constraint volumes remained relatively flat through to 2023 (with a noticeable dip in 2021, which was a very low wind output year). However, constraint costs did rise significantly over that period. As Figure 2 shows more clearly, it is the cost of turning up gas that has been big a driving force behind those rising costs. Those with a working memory of the energy price crisis may notice that the sharp rise in costs in 2021 and 2022, followed by a slow reduction, mirrors very closely the cost curve of wholesale gas over the period.

Between 2023 and 2024 we have seen a very large increase in constrained volumes, so even with average balancing costs turning back down towards pre-crisis levels (in part due to lower turn-down prices for wind energy), total costs are on the rise again.

Wasted wind

What this recent rise in constraint volumes means on the ground is some very high levels of curtailment being bought from a number of Scottish wind farms. Somewhat perversely, in the competitive world of the BM, the cheapest – and therefore first – wind farms to be instructed by NESO to turn down output to manage thermal constraints tend to be the biggest, newest and shiniest additions to the fleet. Another new public resource from James Twallin estimates the impact of this for each wind farm in stark detail. Seagreen (with a capacity of 1075 MW), Viking (433 MW) and Moray East (950 MW) – three flagship renewables projects in Scotland delivered in recent years – all show startling levels of curtailment across the past couple of years according to Twallin’s estimates (66%, 60% and 38% respectively) – see, for example, Figure 3.

Figure 3 – Curtailment at Seagreen according to https://windtable.co.uk/data

We’re simply throwing away more than half the energy that could be produced by some new projects. It’s not a good look and it does little for public perception of renewables and the push for net zero. It should of course be noted that, for many other wind farms in the North and other parts of GB, curtailment is close to zero, with the data appearing to show an average fleet curtailment level since the beginning of 2023 of around 11.5% of the available energy.

What is driving all this curtailment?

How did we arrive at this sorry state of affairs? Well, that brings us to the oft-ignored elephant in the room: transmission capacity. There are two parts to this story, one more obvious than the other. As alluded to earlier, the main bottleneck in the network in recent years lies in getting power out of Scotland and into England – the main constraint boundary on the border is known to those in the business as the ‘B6 boundary’. There are two high-voltage transmission corridors on the onshore network across B6 with a nominal ‘secure’ power transfer limit across the border of around 4.5 GW. (‘Secure’ means that a single fault event wouldn’t lead to overloads on the network or any system instability). In 2018 the Western HVDC Link was added – a 420 km, 2.2 GW subsea cable running from Hunterston in Ayrshire to Connah’s Quay in North Wales bringing us to an existing theoretical secure transfer capability on the B6 boundary of around 6.7 GW. Another key bottleneck is the ‘B4 boundary’ which delineates the two transmission network areas within Scotland (run by SSEN Transmission in the North and ScottishPower Transmission in the South) – this has a present-day nominal secure transfer capacity of 3.4 GW. This is important because lots of the wind in Scotland is located in the far North, including all of that new capacity from Seagreen, Viking and Moray East, totalling around 2.5 GW. The B4 boundary is therefore often the primary pinch point on the system.

A number of new subsea HVDC links are planned along the east coast of GB, first proposed the best part of 15 years ago, one with a capacity of 2 GW running from Peterhead to Drax and another from Torness to Hawthorn Pit, the combination of which would considerably capacity to transfer power across the B4 and B6 boundaries. Ten years ago in the 2015 NOA assessment the system operator (now NESO) suggested that the Peterhead project should proceed with an expected required delivery date of 2023. We are now in 2025 and the latest version of that Peterhead project has only recently been given final sign off and put into construction (Eastern Green Link (EGL) 2) with delivery expected in 2029. Like buses, we might end up with a long delay but two coming along at once with the Torness to Hawthorn Pit project also now due for delivery by April 2029 (Eastern Green Link (EGL) 1). The original timeline for the shorter EGL 1 project was for delivery in late 2027 but the project is now subject to a 16-month delay relating to procurement issues (including a dispute with Ofgem about financial penalties relating to the delay). Three further EGL projects have also been announced for delivery in the mid 2030s to help accommodate the vast amount of new wind capacity expected to connect in Scotland to meet Clean Power goals. The required upgrades are belatedly on their way, it would seem, but the first of these will be more than half a decade late. No wonder we find ourselves in this world of high constraint volumes. It could be a long four years with little hope for short term improvement before the cavalry starts to arrive.

It seems clear we haven’t built out enough North to South transmission capacity quickly enough, and that lies at the root of our current issues… but is there more to the story?

Operational Unavailability

As avid attendees of NESO’s Operational Transparency Forum (OTF) on a Wednesday morning (along with a couple of hundred other energy geeks), one thing has been niggling at us for a while. Each week they show a chart of the operational availability of the main boundaries on the network. Through this, we start to see an additional problem. It is not the maximum secure transfer capability that matters, but how much of that you can utilise in real-time. Figure 4 shows that, very often, the real-time capacity on the B4 and B6 boundaries is well below the maximum level, often even below 50%.

Figure 4 – Operational transmission capacities – from NESO OTF 14/5/25

We looked into these data for 2024 to give an idea of the pattern across the year and found that average available transfer capability on the two key boundaries in the North was in the region of 60%, with significant periods in the vicinity of 40%. The data are a little messy, but translating them into a MW reduction on the maximum possible capacity gives the chart in Figure 5. It is quite illuminating – not only have we, up to now, failed to add a 2 GW link across the congested Scottish boundaries, but B4 spent more than half of 2024 with an additional equivalent scale 2 GW reduction in operating capacity. The B6 reduction was even higher. Why?

Figure 5 – Estimated reduction in capacity on B4 and B6 boundaries in 2024

Well, ironically, it is due to the implementation of network upgrades. Unfortunately, to upgrade the existing network or make a new connection into it, circuits must be taken out of service to perform the work safely. In one recent transparency forum, we asked NESO how long this work would last and whether upgrades in capacity could help to reduce constraint costs while we wait for the subsea links in 2029. The reply came: “Due to ongoing project work for increased power flow from North to South across two Transmission Owner (TO) regions and the interaction of the outage plans, increased capacity across the boundary will be limited and intermittent till 2029. The works include installing 2 phase shifting transformers to control the power flow, and upgrading an existing major power carrying corridor from 275kV to 400kV, benefits of which will be realised after all works are completed.”

The cost of network outages

All this begs the obvious question: how important has this operational reduction in capacity been to the curtailment volumes and costs we started out by discussing, particularly the step change in curtailed volumes from 2024? One of the authors of this blog, Callum MacIver, undertook a simplified analysis to give an idea of what the impact of additional network capacity across the B4 and B6 boundaries would have been since the start of 2024 up to the present. Using data, provided again by Robin Hawkes and his Wasted Wind website, the timeseries of combined wind curtailment was constructed from the beginning of 2024 to present. The peak half-hourly curtailed power averaged as much 7.4 GW. Assuming that the overwhelming majority of this constrained volume was in Scottish territory, Callum was able to look at three hypothetical cases in which available capacity over the period was a fixed 500 MW, 1000 MW and 2000 MW higher across both of the B4 and B6 boundaries than the reality. He then estimated the impact that each of these hypotheticals would have had on curtailed volumes and associated constraint costs over the period.

What we find is that even a modest increase in the capacity across these boundaries of 500 MW (well inside the amount by which the real capacity was reduced for the majority of the year) could have reduced curtailment costs by as much as 25% from the £1.65bn total in the 15 month period from the start of 2024 to the end of April 2025. A 1000 MW uplift in capacity would have tackled nearly half of the problem (45%) while if we’d delivered a 2000 MW uplift, in line with delivering the Peterhead Eastern Link project to its original schedule of 2023, then a full 73% of the thermal constraint costs could potentially have been avoided.

Figure 6 – Curtailment levels: real vs theoretical scenarios with increased available capacity on B4 and B6 boundaries

Figure 7 – Potential cost savings on thermal constraints via increased capacity on B4 and B6 boundaries

For 2024 only, we also ran a scenario where the B4 link was available at full capacity (with a matching MW uprating of B6) – this indicated a roughly 60% reduction in curtailment volumes and costs. It would seem that the issue of reduced operational capacity on the existing network – due largely to outages to facilitate increases in capacity – is at least as important a driving force behind present-day constraint levels as the issue of simply not having built enough new transmission capacity. The cavalry might be due in 2029 but it’s as if part of it is already here but taking up resources while it does its work. When both parts have done their work, we can expect dramatic reductions in wind curtailment.

Conclusions

A number of thoughts flow from these findings:

1. Constraint volumes are highly sensitive to transmission capacity, and even small improvements (or deteriorations) can have strong impacts on volumes.
2. If we’d been able to utilise our existing transmission infrastructure on key boundaries more fully compared with the real level of outage-induced reductions in capacity, then sizeable reductions in constraint volumes and costs could have been made.
3. A sensible amount of transmission capacity accepts some level of curtailment. Current levels of curtailment across the B4 and B6 boundaries seem excessive. However, the current level of installed transmission capacity might actually be closer than we think to what a sensible trade-off may look like for the current generation background, if only we could make fuller use of it.
4. Is it conceivable that the work facilitated by outages could be arranged by the transmission owners and NESO such that key circuits were more available in periods of high wind and high expected constraint costs? (This might not be feasible for construction-related planned outages, as these are typically long and can be quite complex, but might it be possible to design maintenance programmes so that work can be paused and then resumed?).
5. Why have the reinforcements identified in 2015 taken so long to be delivered? Is the effect of planned outages fully accounted for in the cost-benefit analysis sitting behind the development of transmission reinforcement plans and the timing of when works should be delivered?
6. As we build out more renewables in pursuit of Clean Power by 2030 and beyond, the provision of transmission network capacity needs to keep pace.
7. As we embark on the ‘great grid upgrade’ with upwards of £58bn investment by 2035 in new transmission assets[1], it seems inevitable that we will be entering a period of great flux in the availability of transmission capacity with likelihood of high planned outage rates and uncertain delivery times in the name of increased long term network capacity.
8. Major market reform and the introduction of zonal pricing would mean reducing generators’ revenues in constrained periods and placing the risk of transmission unavailability on generators. Our analysis shows quite how sensitive constraint volumes, and therefore generators’ business cases, are to even relatively small changes in transmission availability, something that the generators themselves have no control over. In a reformed market, not only would higher constraint volumes mean lower revenues, but quite how high those volumes will be is hard to forecast. That entails risk and, in all likelihood, a risk premium on bids in future CfD auctions with higher strike prices paid for, in the end, by consumers. Is this preferable to the risk of high system balancing costs, also paid for by consumers? What is the best way of managing risk associated with lack of network capacity, who is best placed to do it, and how long can we expect significant network-related uncertainty to last? Answering these questions presents a challenge for a government looking to secure unprecedented levels of investment in new renewables at low cost.

[1] See https://www.neso.energy/publications/beyond-2030

[1] See https://www.gov.uk/guidance/electricity-network-delivery-and-access