Should NESO be allowed to lower its minimum inertia requirement?

In my latest column for the Daily Telegraph I discuss the risks associated with NESO’s ambition to run the power grid without gas by the end of this year and to reduce its minimum inertia requirement. It has boasted already of running “95% carbon-free’” although this has only been possible due to a quirk of carbon accounting rules which designates the burning of wood at an industrial scale as zero carbon. That is despite the carbon dioxide emissions from wood-pellet biomass being higher than for the coal it replaced.

Whether this ambition will be possible even with the dodgy carbon accounting rules is unclear. In the summer months, gas power stations are often turned up by the system operator in order to stabilise the grid, providing important inertia to control voltage across the network. Throughout the summer, NESO has observed unexplained voltage oscillations (which I will discuss in another post – this is being held up by junk frequency data being reported to Elexon which has yet to be corrected, despite Elexon requesting re-submission), including on the day before the Iberian blackout. These oscillations can, in extreme circumstances, cause the type of grid failure that led to the loss of eleven lives on 28 April in Spain and Portugal.

Despite the risks, NESO wants to reduce inertia limits

Last month, NESO published its Frequency Control and Risk Report (“FRCR”) for 2025 in which it proposed reducing the amount of inertia it is required to hold from 120 GVA.s to 102 GVA.s. It claims such a move would save up to £96 million, a drop in the ocean of the total system balancing costs which were last week reported at £2.7 billion for 2024/25. NESO cites the use of batteries and other assets such as synchronous condensers procured through its Stability Pathfinders as providing alternative forms of inertia to conventional generators.

NESO’s claims about inertia management in the report and elsewhere don’t stack up. In the Operability and operations analysis annex to its Clean Power 2030 advice to the Government, NESO states that “we currently ensure system inertia is always above 120 GVAs.”

This chart (from the FRCR) implies that the 120 GVA.s limit is never breached during 2024, but in fact, analysis of the raw data show that in calendar year 2024 in outturn inertia was below this level in 9.6% of settlement periods, rising to 10.7% for the IFA year 2024/25 (1 April 2024 – 31 March 2025). If we compare with the actual inertia limits in place at the time of 140 GVA.s up to 28 February 2024 and 130 GVA.s from 28 February to 18 June 2024, the lower limit was breached in 15.4% of settlement periods in 2024.

While some people may take comfort saying this indicates the system can be safely run at lower levels of inertia, the fact that NESO fails to meet its obligations with such regularity is far from comforting. And the reason for this failure lies in the fact that it doesn’t actually know either in real time or afterwards, what inertia actually is or was at any time.

While NESO has invested in some real-time measurement systems, these are currently only deployed in Scotland. Elsewhere, the control room relies on estimates. So it is hardly surprising that when it calculates actual inertia after the fact it turns out to have missed the mark so often. In fact, according to statements made in the NESO Operational Transparency Forum (“OTF”) (specifically on 29 January 2025), even the outturn inertia data are estimates rather than actual measurements:

“System outturn inertia is estimated as the sum of the inertia provided by Generators, the contribution from Demand Side and the contribution from Pathfinder Units,”
– NESO, Operational Transparency Forum slides, 29 January 2025

These estimates are based on which conventional power stations are synchronised at any moment, using pre-agreed inertia values published for each generator type. The control room uses these static figures combined with real time SCADA data to model total system inertia. However, this method does not account for fast-changing conditions or inertia contributions from newer assets such as batteries, inverter-based generation such as wind and solar, or embedded generation.

While post-event analysis may offer a more accurate picture, in real time, the control room is essentially working off educated guesses that have increasingly been proven wrong. Even NESO acknowledges that its estimation tools are still being developed, and real-time inertia measurement developed by Reactive Technologies (which I wrote about back in 2017) remains limited to parts of Scotland.

Ironically, the push for gas-free operation makes it even harder for NESO to know whether the system is truly safe. With few conventional generators online, the control room loses visibility over most real-time inertia inputs — meaning the estimates they rely on become less reliable just when accuracy matters most.

Another key input into system management is demand forecasting. In the past few years, NESO’s demand forecasts have demonstrated significant errors when compared with actual outturn data – up to 4.7 GW over a half-hourly interval (on a day-ahead basis), with an average of 609 MW. NESO does not publish more granular data, but if this is the average error the day before delivery, what sort of errors is it making within-day? And what is the error over the sub-1-minute intervals which are relevant for grid stability? On 22 January NESO announced an audit of its demand forecasting, with the engineer presenting this to the OTF commenting that the forecast methodology had not been looked at in “10-20 years”!

Of course, this raises concerns about what would happen if the inertia target was reduced. Were NESO to continue to undershoot at the same level, this could put the system at real risk of failure. The financial savings that could be realised by reducing the target could be quickly swamped by the economic costs of a blackout – early estimates of the cost of the recent Iberian blackout begin at €150 million, and could rise to €450 million depending on the degree to which heavy industries were able to re-start their process in line with the restoration of power to the grid.

Official report into Iberian blackout is damning

On 16 April, Spain’s system operator, Red Eléctrica de España (“REE”), boasted that it had generated 100% of its electricity with wind, solar and hydro power, for a few minutes. Two weeks later on 28 April, a prolonged blackout sent all of Iberia dark. The outage affected parts of France and even Belgium reported issues in what was Europe’s biggest ever blackout. A week later, another blackout hit Spain’s Canary Islands.

On 28 April, at around 12:30 pm local time in Spain, just before the grid collapsed, renewable sources accounted for 78% of electricity generation on the Iberian system, with solar accounting for almost 60%. By contrast, conventional generation, such as gas and nuclear power plants, comprised only around 15% of the total generation mix.

According to Raúl Bajo Buenestado of the Baker Institute, two consecutive generation loss events occurred in southwestern Spain, likely involving large solar installations. “In just five seconds, Spain lost approximately 15 GW of capacity, equivalent to 60% of its national electricity demand. The remaining generation was insufficient to meet demand, thus triggering a cascading failure across the entire grid. Various generating units were automatically disconnected to protect infrastructure, and nuclear plants were shut down in accordance with safety protocols.”

Now, according to reporting by Reuters, the first official report by the Spanish authorities has blamed REE for miscalculating its power capacity needs on 28 April, meaning that a surge in voltage led to a massive blackout. REE did not have enough thermal power stations switched on during peak hours according to Spain’s Energy Minister Sara Aagesen in a news briefing in Madrid.

“The system did not have sufficient dynamic voltage control capacity…[REE] told us that they made their calculations and estimated that (switching on more thermal plants) was not necessary at this time. They only set it for the early hours of the day, not the central hours”
– Sara Aagesen, Spanish Energy Minister

The report describes the blackout as being “caused by an overvoltage problem with a multifactorial origin: the system had insufficient voltage control capacity, there were oscillations that conditioned the operation of the system and disconnected generation facilities, in some cases in a seemingly undue way”. There is an English version here. A longer version of the report (only in Spanish) is available and I have started to look through a translation, but ChatGPT tells me it does not add anything material. I’ll come back and update if that turns out not to be the case. The charts are from this longer report.

It described the chronology of the outage as follows:

Phase 0: Voltage instability
During the days prior to the incident there voltage oscillations and on the morning of the 28th the voltage varied more intensely than normal.

Phase1: Oscillations in the system (12:00 – 12:30)

At 12.03 pm an atypical oscillation of 0.6 Hz was recorded, which caused large voltage fluctuations for 4.42 minutes. This oscillation forced the System Operator to apply measures to cushion it, such as increasing the grid meshing – restricted by low demand – or reducing the flow of interconnection with France. All these actions dampened the oscillation, but had the side effect of increasing voltage. At 12.16 pm the same oscillation was recorded again, this time smaller, and 3 minutes later there was a further oscillation of 0.2 Hz. The System Operator applied the same measures to cushion it, which also contributed to increasing the voltage.

PHASE 2: Generation losses (12:32:57– 12:33:18)

Voltage began to rise rapidly and steadily, and numerous and progressive disconnections of generation facilities were recorded in Granada, Badajoz, Segovia, Huelva, Seville, Cáceres and other provinces.

PHASE 3: Collapse (12:33:18 – 12:33:30)

The progressive increase in voltage caused a chain reaction of overvoltage disconnections that could not be contained, as each disconnection contributed to further increases in voltages. There was also a drop in frequency that led to the loss of synchronisation with France. The interconnectors with France tripped, and the Iberian peninsular saw its power grid fall to zero.

.

The report goes on the identify three main causes of the blackout:

1. Insufficient voltage control

The system showed insufficient voltage control capacity for two reasons. Firstly, on the day before the incident, the System Operator scheduled the activity of 10 synchronous power plants with the capacity to regulate voltage on the 28th. The final number of coupled synchronous power plants was the lowest since the start of the year. Secondly, several of the plants capable of regulating voltage – and specifically paid for it as they were scheduled for this purpose – did not respond adequately to the instructions of the System Operator. Some even produced reactive power – the opposite of what was required – exacerbating the problem.

2.Frequency oscillations

The oscillations – the first of which was atypical and had its origin in an installation in the Iberian Peninsula – forced the configuration of the system to be modified, making it harder to stabilise voltage. After the second oscillation, the REE demanded the availability of a plant capable of regulating voltage, but it was technically impossible to instruct them before the collapse.

3. Generation trips

Some of generators that tripped would have done so before the voltage thresholds established by the regulations were exceeded (between 380 kV and 435 kV in the transmission network) ie they failed to meet their fault ride-through obligations. Other trips occurred after the limits were exceeded, as would be expected to protect equipment.

Once the cascade started the usual grid protections were unable to stop or contain the failures. Some of the protections mechanisms, such as load shedding, may even have contributed to the overvoltage by discharging the lines even more, contributing to the rise in voltages, because they acted to compensate for the drop in generation rather than to manage the voltage.

In conclusion, there was a lack of voltage control resources, either because they were not sufficiently scheduled, or because some of the units that were scheduled did not provide the service adequately, or because of a combination of both.

The Government report contains the following recommendations:

• Increased supervision and compliance monitoring of regulations such as fault ride through
• Technical measures to increase voltage control and protections against frequency oscillations. The National Commission on Markets and Competition is currently considering reforms that will allow asynchronous installations to use power electronics to manage voltage variations
• Increase the flexibility of the electricity system. The Electricity Plan 2025-2030 will prioritise industrial consumption and an increase in storage capacity
• Continues with planned increases in interconnection with neighbouring countries

Although cybersecurity was determined not to be a factor in the blackout, the Government also proposes to accelerate the implementation of European regulations and network controls that will increase detection and monitoring systems providing a higher level of grid surveillance.

Since the incident, there has been talk that REE has increased the use of gas on the grid to support inertia, however that is only part of the story.

While the amount of gas on the Spanish grid has increased by 25% since the blackout, compared with the four weeks immediately before, overall synchronous generation was about 2.7% lower due to seasonal reductions in nuclear for maintenance, and unexplained reductions in hydro. This suggests REE is still not being as conservative with the system as the blackout might warrant, not least because the genesis of the blackout has yet to be made public, suggesting that rather than a mundane grid fault, it may well have been related to the behaviour of inverters.

Like NG ESO as it was back in 2019, REE is trying to pin the blame for the blackout on non-compliance with grid codes by generators. At the time I said this was weak: why was NG ESO not monitoring Grid Code compliance? And why did windfarms not have to be compliant beyond a certain level of capacity – historically generators did not have to fully comply with the Grid Code until they were fully commissioned, but since windfarms can have over 1 GW connected to the grid, as I think Hornsea 1 did at the time, surely this rule needs to be changed. Unlike conventional generators. Windfarms commission incrementally so beyond some agreed level of capacity they should be required to meet materially all of the obligations any other generator must meet. (Not least because Hornsea also failed to make timely REMIT notices so the market thought the CCGT at Little Barford tripped first after the lightening strike but it didn’t, Hornsea did!)

But what is striking from this initial report into the Iberian blackout is that there is still no explanation for the initial cause. Three generators didn’t just trip off at the same time for no reason – the frequency oscillations described triggered their trips. But the report is silent on the origin of those oscillations. I suspect that if there had been some mundane grid fault, this would have been made public, which leads me to suspect we’re dealing with inverter-induced oscillations, something they would be reluctant to admit. We shall have to see if more comprehensive subsequent reports uncover more detail.

Ideology not money driving NESO’s agenda

The real reason NESO wants to drop the inertia target isn’t cost but ideology – to enable it to claim zero carbon running (despite the use of wood). The burning of wood isn’t the only source of smoke and mirrors – NESO also relies on embedded generation to help out. This generation, like interconnectors, is counted as zero carbon for NESO’s carbon accounting purposes regardless of their actual fuel. Some of these units are diesel and gas engines, many of which were built in the early years of the capacity market that is costing some £3.6 billion per year.

The Spanish government’s official report into the 28 April blackout confirms that the collapse was not caused by a cyberattack or discrete fault, but by a “multifactorial” sequence triggered by a miscalculation by Red Eléctrica . Investigators found that REE failed to ensure enough synchronous generation was online to maintain voltage control, leaving the system exposed when three major generating units disconnected unexpectedly – but three units don’t all trip unexpectedly at the same time – something caused them to trip.

While voltage regulation systems failed to stabilise the system, this was a consequence of the underlying planning error: too much reliance on inverter-based generation in low-demand, low-inertia conditions. Reports of voltage overshoot – rather than collapse – strongly point to uncoordinated reactive power injection by inverters, a recognised risk in weak grids with insufficient damping.

This is uncomfortably relevant to Great Britain, where NESO is seeking to reduce the system’s minimum inertia floor from 120 GVA.s to 100 GVA.s. NESO maintains that frequency response and voltage services can compensate, but Spain’s experience suggests this assumption may be fragile. Crucially, in 2024 NESO failed to meet its minimum requirement in over 15% of settlement periods – despite this being the existing safety threshold. Some have argued this proves the system is stable even at lower inertia, but this misreads the situation: the fact that no faults occurred during those periods is not proof of resilience. Rather, it highlights a more troubling truth—NESO is already struggling to enforce its current minimum, and the measurement and forecasting of inertia remain uncertain. Under these conditions, lowering the threshold looks more like a gamble than a controlled evolution. If Spain’s blackout teaches anything, it is that when planning assumptions meet physical reality, the latter always wins.

.

The question is why, when  the risks have so recently been illustrated, is NESO pushing in the opposite direction? In practical terms, a few symbolic minutes of “zero carbon” or “gas-free” running mean little when the use of gas is still so critical to the power grid (in winter months we could not contemplate running without gas if we want to be able to meet demand at all times). So why take the risk, just for a brief moment of virtue signalling?

Spain tried that, and it cost eleven people their lives.