North Hyde substation fire: report reveals poor maintenance culture

This week, the National Energy System Operator (“NESO”) published its final report into the North Hyde substation fire on 20 March that caused a local power outage which cut supplies to over 71,500 homes and businesses including nearby Heathrow Airport.

The airport was without power for most of the day, disrupting thousands of flights and having a global impact as incoming flights were diverted and outgoing flights were cancelled. Thankfully there were no reports of serious injuries or fatalities.

In this blog I will review the report, explore some of the issues raised around maintenance and asset lives, and consider whether Ofgem should be more proactive in its oversight of the management of network asset condition.

What happened?

According to the report, at 23:21, one of three supergrid transformers at the substation owned and maintained by National Grid Energy Transmission (“NGET”), a subsidiary of National Grid plc, tripped and was later confirmed to have caught fire. Soon after, the adjacent transformer also tripped.

A third transformer at the site was further away and was not affected other than it was unable to be used as long as the substation was de-energised while the fire was tackled. The site layout is described in my previous blog on the incident.

Transformers are some of the most important pieces of equipment on the grid – they allow electricity to be moved in bulk over long distances with minimal losses, by increasing the voltage.

However, as high voltages are dangerous, and are unsuitable at the point of use of the electricity, voltages are stepped down before they reach the consumer. Transformers are used to make these voltage transformations.

Transformers consist of wires wrapped around a metal core. These windings are subject to heating, with heat degradation potentially leading to short circuits. For this reason, they are cooled, through the use of oils which are circulated round the windings. The oils are electrical insulators to prevent shorting between the windings, and are thermally stable ie they don’t freeze in winter. But they are highly flammable, so in the event of arcing, they can explode.

The part of the transformer that failed at North Hyde is called a bushing which is essentially a sleeve that allows cables to enter and exit the metal oil tank without touching the walls of the tank and becoming earthed. Unfortunately if moisture gets inside the sleeve, a short circuit can occur, leading to arcing, and evidence of this was observed in the debris from the fire.

As noted in my previous blog, Heathrow lost power when the fire broke out and closed. Some flights resumed at 7pm on 21 March but the airport did not resume normal operations until Saturday 22 March.

The NESO report tells us a bit more about the Heathrow grid connections and but the substations that actually supply it are simply referred to as A, B and C.

What I didn’t mention in my previous blog as I didn’t discover it until later, was that there had been a very similar substation fire at Atlanta airport in 2017, as a result of which US airports improved their energy resilience. Heathrow could and should have learned from that incident and not referred to the North Hyde fire as being “unprecedented”.

Track record of delayed maintenance

This particular transformer at North Hyde had known issues: moisture had been detected inside the bushings as far back as 2018 at levels NGET’s guidance said indicated “an imminent fault and that the bushing should be replaced”. The technical guidance dictated that following such highly elevated moisture samples, the asset should remain out of service until the bushing can be replaced, or mitigating actions put in place.

“While the reading was recorded, the mitigations appropriate to its severity, as set out in National Grid Electricity Transmission’s relevant policy, were not actioned at the time. The controls in place were not effective and failed to identify subsequently that action had not been taken in relation to the elevated moisture reading. This review finds that there were other opportunities which, had they been pursued, could have caught the issue. Instead, the elevated moisture reading went unaddressed,”
– NESO North Hyde Review – Final Report

A forensic examination of the remains of the bushings after the fire showed evidence of moisture ingress on the C-phase high voltage bushing that suffered the catastrophic failure. Such evidence of moisture ingress would not have been visible externally and therefore only became apparent after disassembly and forensic analysis after the incident.

The forensic analysis also found evidence of arcing erosions within the C-phase high voltage bushing. Arcing erosion is the damage caused by electrical arcs, which normally occurs due to electrical faults or contamination, including by moisture ingress. This damage can lead to the gradual degradation of parts of the bushings, possibly leading to reduced insulation resistance, potential short circuiting and ultimately, bushing failure.

Phases A and B of the high voltage bushings did not show any evidence of arcing erosion, supporting the case that the fault was on the C-phase only and had arisen as a result of moisture ingress. NESO and the forensic engineers that inspected the remains of the bushings concluded that due to the extensive damage to the transformer it is unlikely that any greater certainty about the exact cause of the fire can be achieved.

So there had been evidence in 2018 of moisture within the bushing to such a degree that the transformer should have been taken out of service while the bushing was replaced, but this work was not carried out, and some seven years later, the post-fire examination found evidence that there had indeed been internal damage to the busing consistent with moisture ingress.

In addition to the failure to address this equipment fault, basic maintenance of the transformer was overdue and had been consistently put off, and fire suppression equipment at the substation was found not be to working in 2022. A further inspection in 2024 showed it still didn’t work and repairs had not been carried out at the time of the fire, although London Fire Brigade suggested it would not have made much difference had it been working.

In 2023, NGET decided to extend the interval between basic maintenance from four to five years, and for major maintenance from 11 to 15 years. The most recent basic maintenance of the transformer had been in 2018, so it was well overdue on both the old and new schedules.

NESO’s report sets out the delays to the basic maintenance for the North Hyde transformer, which was overdue by the time of the fire. Multiple attempts had been made to schedule this maintenance, none of which went ahead:

• A planned outage to enable maintenance had been scheduled for 4 October 2021, but was delayed due to NGET’s reprioritisation for other higher risk works
• A further maintenance outage was scheduled for 10 October 2022, but shortly before the outage window, an issue on another transformer at the substation meant it could not go ahead. NGET said it was unable to replan the maintenance within the next outage planning year, which runs from April to March
• From December 2022 onwards, a number of mitigating actions on the transformer were implemented, as it was known that the basic maintenance would be overdue by March 2023 (under policy that was in force at the time). These included carrying out a Radio Frequency Interference condition monitoring survey and a thermovision inspection every three months. Whilst these would not have identified moisture in the C-phase bushing, they may have identified issues caused by moisture ingress
• Maintenance was replanned for 2 September 2024, but did not occur as an equipment defect at the nearby Iver 275 kV substation meant that the North Hyde 275 kV substation was needed to secure the system and could not be taken out of service
• The next planned outage to enable maintenance was scheduled for 19 May 2025, but even this had been moved back to 17 November 2025 to optimise NGET’s overall resource availability.

NGET’s maintenance policy targets 85% of assets being within their normal maintenance cycle – as of May 2025, 697 of 805 transformers (87%) were within the maintenance frequency target, just meeting the standard. However, NESO does not comment on whether there is any further requirement in terms of the length of delays to maintenance one a site is outside the cycle.

In this instance, basic maintenance was put off four times, and at the time of the fire, had not taken place since July 2018, ie in six years and 8 months – 20 months after it should have been carried out under the newly extended policy, which should be a third way through the next maintenance cycle!

While oil sampling of the bushings to detect the presence of moisture is not included as part of basic maintenance and is planned separately, NESO believes that had the basic maintenance been conducted, NGET could have taken the opportunity of the transformer being out of service to conduct oil sampling, which may again have brought attention to the elevated moisture reading on the C-phase bushing.

There are also advanced online continuous condition monitoring solutions which could have enabled detection of partial discharge or arcing prior to the incident. These are deployed in certain circumstances on the transmission system already. In the RIIO-2 price control period Ofgem allowed NGET to allocate £18.65 million, plus indirect costs, to investments in such monitoring. The criteria for deploying this technology are based on an assessment of factors such as asset population, asset age and historical performance data. This technology was not in place in North Hyde, where NGET had determined that the criteria were not met.

NESO did not comment on whether this decision was reasonable given the actual age of the transformer, records that it had been operating at above its rated capacity and the fact that it was located in a highly populated area where the grid was highly congested. Meaning that not only was the asset vulnerable, its failure was likely to have a significant impact on consumers.

The sequence of events at North Hyde suggests a wider, systemic problem with delayed maintenance, and little follow-up in terms of the risks associated with these delays.

Another concern that was not mentioned in the NESO report but that is relevant is that the substation had been identified in a 2022 report for the London Assembly as having peak utilisation in 2021-22 of 106.2%. This is likely to have continued as the grid in west London is highly congested.

NGET’s approach to asset life management

The NESO report identifies that the transformer was installed in 1968, making it 57 years old. NGET believes that transformers can last 60-65 years, but a 2024 report for energy regulator Ofgem by Cambridge Economic Policy Associates (“CEPA”) found that NGET had much wider asset life parameters on average than its peers in the UK (such as the Scottish transmission owners). Despite the North Hyde transformer approaching NGET’s end-of-life window it had not identified the unit as being close to end of life and on a pathway to replacement.

The CEPA report was commissioned to ascertain whether the average asset life of 45 years used for the purposes of depreciation in RIIO-2 remained appropriate. In some ways, this is the first problem – the assessment of asset life should be from a physical performance rather than financial perspective. And unfortunately the question about real life weren’t really asked:

“Given the large number of assumptions inherent in any estimation of technical asset life data and the fact that this data was not part of formal business plan submissions, the average figures resulting from our analysis should be seen as an indication of direction of travel, rather than detailed precise estimates,”
– Economic Lives of Electricity Transmission Network Assets, CEPA

CEPA noted that “Compared to the 2010 analysis, National Grid Electricity Transmission (NGET) and Scottish Hydro Electric Transmission (SHET) both report an increased range in statutory asset lives across their asset portfolio – the range presented by NGET has widened from 15-60 years to 15-100 years, and the SHET values now span 5-80 years (from 10-80 years in the 2010 data set). However, Scottish Power Transmission (SPT) now use a single figure of 40 years, compared to the range of 30-60 years in the 2010 study.” NGET made significant increase to its determination of asset lives since 2010 – has this been subjected to real scrutiny?

CEPA went on the say that “based on the central estimates provided by the network companies, the weighted average technical asset life for GB as a whole was estimated at 55 years”.

NESO’s report into the North Hyde fire stated that NGET has assigned an End of Life (“EoL”) score of 12.7 out of 100 for the defective transformer (0 is for new assets and 100 is for ones that have reached the end of their lives). It is difficult to see how a transformer that was 57 years old and not maintained in line with NGET’s policy, with faults that required it to be immediately removed from service for remedial works could possibly have such a low EoL score unless there had been some major overhaul at some point in the past which was not revealed buy the report.

“On coherence and justification, the plan lacked a clear thread particularly around NARM [Network Asset Risk Metric] and asset management plans. There were contradictions between the load plan and non-load plan which suggested a lack of long term planning and meant that the overall plan lacked coherence in places,”
– NGET RIIO-3 draft determination, Ofgem

Ofgem has identified issues with asset life in the RIIO-3 process. It says that non-load replacement capex plans have poor justifications and high costs, and while non-load refurbishment capex was deemed to be better evidenced, there were criticisms that no schemes were priced below the benchmark. By contrast the load related capex plans were deemed to be comprehensive and acceptable.

(“Non-load” capex refers to capital investment in existing assets — replacing, refurbishing or reinforcing infrastructure such as transformers, switchgear and cables to manage asset health, safety and reliability. “Load” related” capex, by contrast, is driven by changes in demand or new customer connections — for example, building new substations or upgrading existing ones to accommodate additional load or generation. Both types of capex may involve similar physical equipment, but the driver and regulatory treatment differ.)

This suggests a couple of things to me: has NGET prioritised load over non-load projects in response to some perceived regulatory preference, or is Ofgem applying a different standard to these projects reflecting its own priorities relating to renewables connections. It seems strange that a company would be good at assessing cost for one type of capex but not another where the underlying assets are similar (ie network equipment rather than real estate or software).

Was Ofgem asleep at the wheel?

This all raises questions about Ofgem’s role. While it has been critical of NGET’s business plans in respect of aging infrastructure, its response lacked teeth. Why did Ofgem not insist that asset lives be more rigorously analysed and that maintenance of older critical equipment was carried out on time? Was Ofgem’s focus on renewables connections a factor in the approach taken by grid owners to legacy equipment, knowing that consumers can’t pay for everything at once during a cost-of-living crisis?

Ofgem rejects this criticism and says it has never told regulated firms to downgrade investment in legacy assets in order to prioritise renewables connections. But connections are a huge focus for the regulator, and it would be naïve to think that regulated firms would not take this into account when developing their business plans. (Out of curiosity I searched the words “connections” and “asset” on the Ofgem website. “Connections” had 1804 hits while “asset” had 681 hits.)

Ofgem’s job isn’t simply to rubber-stamp investment plans, it’s to challenge asset risk management and resilience plans. But when you dig into it, Ofgem didn’t rubber stamp NGET’s business plan – in RIIO-2, the current price control period, it cut non-load capex by a third from  £2,650.9 million to £1,765.8 million (the draft determination had suggested a 72% reduction!). The load related determination was actually higher NGET’s request – this is money for connecting new generation, primarily renewables.

Ofgem said that NGET’s engineering justifications were insufficient which is why it had proposed a 72% reduction in the draft determination. Following improved disclosures by NGET, the budget was revised upwards, but was still materially lower than NGET had requested. This exposes a fundamental flaw in the price control mechanism: Ofgem wants to endure networks are delivered at the “lowest possible cost to consumers”.

I strongly oppose this objective – a more appropriate measure would be “lowest reasonable cost”: first of all how can you ever be certain that something is the lowest possible cost, and secondly this will bias decision-making against expenditure, possibly inappropriately. It certainly biases towards delayed expenditure due to uncertainty over whether it is the lowest possible cost solution.

So Ofgem’s response to NGET’s insufficiently detailed business plan was to assume that anything that could not be well evidenced should not be done. The bias was towards avoiding cost rather then being prudent with asset replacement. It is probably at the heart of the difference between a nationalised and privatised industry – before privatisation, investment was done when the network owner thought it was appropriate. Sometimes it was premature, excessive or demonstrating poor cost control, but it tended to be an asset-first approach. Now we have a cost-first approach that prioritises short-term cost to consumers.

Specifically in relation to supergrid transformers (“SGTs”), in RIIO-2 Ofgem said:

“in Draft Determinations, we proposed to reduce NGET’s planned SGTs from [redacted] to [redacted] and to reject the rest as we viewed it uneconomic and inefficient to replace healthy assets. NGET provided significant additional asset health and project data to support their request. This included highlighting assets which had rapidly degraded from their originally reported position. We note that NGET’s pessimistic assumption of risk increases are generally driven by asset family issues, which we note are difficult for us to assess. However, we accept and approve the additional 8 units highlighted by NGET as high-risk assets which warrant replacement. We have also decided to approve 3 Static Compensator Transformers, which NGET evidenced as high risk.”

It’s also interesting to consider the costs of the price control itself – Ofgem wants detailed justifications from NGET and other network owners in respect of expenditure. This means NGET and other need to spend significant time and resources to prepare detailed plans relating to the thousands of assets they own, and Ofgem needs to hire engineers qualified to assess these plans and determine whether the works are justified. Industry complains about the regulatory burden – the price control is an example of this, state owned enterprises suffer from poor cost control to the detriment of tax-payers while privatised monopolies have a profit imperative to the detriment of bill-payers. However, Ofgem should not seek to ensure the lowest possible costs as this risks a culture of delaying necessary investments and may result in higher long-term costs anyway.

Ofgem has now announced an investigation saying it expects network companies to “properly maintain their equipment”. The words “horse” and “bolted” spring to mind – the regulator should be monitoring this continually, not just issuing fines after things have gone wrong.

Regulatory reform is needed – simply fining people after a fault is not enough

We need urgent action: Ofgem must demand a comprehensive review of all legacy transformers and other critical grid equipment, enforce swift maintenance or replacement. It also needs to be far more proactive about ensuring grid owners are assessing asset life appropriately and that their maintenance procedures are both rigorous and properly adhered to in practice.

I suggest an approach similar to a controls-based financial audit: Ofgem engineering teams would periodically (eg every 2.5 years being half of the price control period) inspect all asset management policies, review board minutes and other evidence relating to policy changes, and select assets from each major asset class on a random basis for further analysis: what do the records say about these assets, is the maintenance up to date, does the physical condition of the assets correspond to the documented condition etc. They should also review things like site security.

They should also instruct network companies to identify any assets that impact other critical infrastructure, or would expose people or property to high levels of risk were they to fail, and consider whether a stricter maintenance regime would be appropriate. For example, a substation explosion could be more dangerous in some locations than others, for example if located close to a school.

More broadly, I recommend that Ofgem and other regulators create teams whose sole job it is to think about all the ways that disaster could strike their regulated firms. These could be errors, mundane failures or deliberate sabotage or other criminality.

• What if a major supplier goes bankrupt unexpectedly due to audit failures (it happens)?
• What happens if network assets are improperly maintained?
• What happens if NESO control room software or hardware fails or someone hacks the NESO control room?
• What happens if there is corruption within regulated firms so investments Ofgem thinks have happened did not, or were done with inferior materials?
• What happens if there is corruption within Ofgem teams?
• What happens if a third of the CCGT fleet retires by 2030?
• What happens if there is a massive solar flare or someone targets networks with EMP devices?
• What happens if a systemic fault is discovered in any type of network equipment similar to the stress corrosion issues in the French nuclear fleet, eg accelerated thermal aging or water treeing in cross-linked polyethylene power cables or new synthetic ester transformer oils degrade faster than expected?

The impact of solar power on transformers is an under-discussed risk. Rapid load transitions caused by solar PV ramping, particularly at sunset can impose thermal and electrical stresses on transformers, and there’s growing concern that legacy transformer design models may not fully account for the new stress regimes introduced by high-penetration solar.

During the day, large amounts of solar reduce the demand seen by distribution and transmission transformers, even reversing it in some cases (net export). Many transformers run lightly loaded or with reversed power flow. At sunset, solar output drops quickly over ~30–60 minutes. Grid-connected load reappears suddenly as solar generation ceases, and transformers experience a sharp increase in load current, often accompanied by a voltage dip and reactive power swings.

This is especially stressful when air temperatures are still high (eg summer evenings), reducing the transformer’s cooling effectiveness or when there’s a mix of PV and fast-reacting loads (EV charging, air conditioning) causing both electrical and thermal shock.

This impacts transformers in a number of ways:

• Thermal cycling fatigue: repeated heating and cooling stresses insulation, gaskets, and conductor materials. If the evening ramp is steep and daily, it accelerates aging of winding insulation and can promote oil degradation
• Overfluxing risk: voltage dips and high loading can lead to overfluxing if not properly managed, stressing the magnetic core
• Mechanical stress on tap changers: frequent voltage variations at sunset may trigger repeated tap changes, increasing wear
• Harmonic distortion and flicker: in areas with a lot of solar and switching inverters, high-frequency harmonics at sunset may add further heating.

There could be a tipping point of solar penetration connected to substations above which this becomes a problem, and this could vary by location and transformer type.

.

There’s a lot we can and should learn from the North Hyde incident. To its credit, Ofgem is showing itself to be open to this, and to engaging with me and others in how it can improve. The test will be whether it takes a more pro-active approach to managing risk within the sector, or simply continues with the current MO of reacting with fines and new rules only after things have gone wrong.