In the aftermath of high profile – and highly impactful – power system failures in recent months, most notably affecting Heathrow airport and the entire Iberian peninsula, UKERC Co-Director Keith Bell reflects on those events and what lessons they hold for ensuring that electricity supplies remain sufficiently resilient as we transition to a low carbon power system.
In this first part of a two-part extended blog, Keith discusses the need for good engineering and what we’ve learned about what happened in Spain and Portugal on 28 April.
I met the House of Lords Science and Technology Committee a few years ago, in a private discussion about electricity supply resilience. One of the members at the time had – still has, I think – a newspaper column that he’s regularly used to try to disparage climate scientists and advocates of reduction of greenhouse gas emissions. My meeting with the Committee was not long after the power system in South Australia had collapsed. This particular member suggested to me that it happened because of the use of wind farms.
I don’t remember exactly what I said in reply. It was true that weather had played a part – there was a severe storm – but mainly because of lightning strikes that had caused transmission lines to suffer from short circuits. These were short-lived but caused voltages on the network to dip temporarily. Grid connection rules had required wind farms to be capable of continuing to operate when there was a voltage dip. However, when a number of wind farms disconnected themselves – that is, they failed to ‘ride through’ the low voltages – it turned out no one had defined what wind farms should do when there are multiple dips in quick succession. In the absence of any specification, the manufacturers of these particular wind turbines had written the control software to ride-through one voltage dip but not a number of them. It also turned out that, in spite of the storm conditions, the system operator had failed to set the system up to remain stable even if the overhead line to the neighbouring state of Victoria tripped. It did trip. The combination of loss of power from wind farms – partly due to failure to ‘ride through’ successive voltage dips and partly due to predictable shutdowns of wind turbines if they experience very high wind speeds – and the loss of the ability to import power from Victoria left the system in South Australia with an inability to support demand, leading very quickly to a complete shutdown.
The main point that I made to the Committee was that the problem in the South Australia event was not the use of electricity from wind farms but a failure to get the engineering right.
I made the same point at the end of April when responding to various media enquiries about what was likely to have caused the collapse of the power system in Spain and Portugal. It doesn’t matter where the energy comes from, system operators are still supposed to operate the system in a secure and stable manner (and the evidence suggests that, for the most part, they succeed). In other words, regardless of the energy source, you’ve got to get the engineering right. Even when good practice has been followed, experience from around the world over many years shows bad things can and do happen.
There tend to be, on average, one or two major power system outages somewhere in the world every year. These affect large regions, or entire countries, impacting systems deriving their energy from all sorts of sources – be it fossil fuels, nuclear power, hydro power or wind and solar. These events all involve a multitude of factors with often very complex interactions. This was true of what happened on the Iberian peninsula on 28 April, something that has been confirmed in reports published in June by the Spanish government and the system operator in Spain, Red Eléctrica.
The blackout in Spain and Portugal shows dark cityscape with electricity pylon. Source: Shutterstock
The first problem on 28 April was that two slow oscillations of power and voltage started happening (the first taking between a bit more than 1.5 seconds and the second around 5 seconds to go through each complete cycle). One of them was something that has been observed many times before, with some procedures defined to guide the system operator on what to do about it. One of the effects of that action was to increase average voltages across the network. However, a number of the large, conventional, thermal power units (that use rotating ‘synchronous machines’ to generate electricity) apparently failed to take action (that the owners had accepted contracts to carry out) to correct voltages back to lower levels.
Equipment connected to a power network cannot operate if the voltage is very low or very high. Voltages naturally vary. In order to be allowed to connect and operate, generation of all types must be capable of operating within a particular, defined range. However, some generators – it seems mostly solar PV – disconnected themselves when voltages were rising, albeit still within the range in which they were supposed to still operate.
The effect of these disconnections was to cause voltages to rise still further, in many places, one supposes, above the statutory maximum value. This led to further disconnections, big power swings and the loss of the alternating current lines connecting Spain to France. Very quickly – in around 20 seconds – with the loss of so much generation, voltages across the system in Spain and Portugal collapsed to zero and all motors stopped, lights went out, IT equipment lacking batteries stopped working, air conditioning failed and so on.
Control of voltage – keeping it low enough to stay within safe bounds for electrical insulation and loading of equipment, but high enough for voltage stability and for energy users’ equipment to work correctly – is a basic duty of a system operator. It depends on resources connected to the network being able to modify the relationship between voltage and current and, in a well-established engineering shorthand, inject or consume what’s referred to as ‘reactive power’.
Figure 1; Variation of voltages on the 220 kV network in Spain between 09:00 and 12:00 CEST on 28 April, 2025. Voltages should normally be permitted to vary only between 205 kV and 245 kV. Figure: ENTSO-E based on measurements from Red Eléctrica.
https://www.entsoe.eu/publications/blackout/28-april-2025-iberian-blackout/
A mix of resources can do this: synchronous machines (either as the means of generating electricity at conventional power stations or as ‘synchronous compensators’); transformers with an ability to automatically change the point of connection to one of the windings inside them; or big banks of capacitance or inductance that can be switched in or out. However, the power electronic ‘inverters’ that connect solar PV, many wind turbines and high voltage, direct current (HVDC) interconnectors to the network can also do it.
‘STATCOMs’ use power electronics and have a similar capability. These, synchronous compensators and the banks of capacitance or inductance are collectively known as reactive compensation and are often installed with the specific purpose of controlling voltage. In Britain, sometimes they’re part of a wind farm, installed in order that the wind farm as whole can comply with Grid Code obligations at the point of connection to the network. More usually, they’re built and maintained by the network owner or, more recently, by a third party following a competitive procurement process run by the National Energy system Operator (NESO).
One of the consequences of liberalisation of electricity supply industries around the world has been that different physical assets on which stable operation of an electricity system depends are owned by different parties. All these different assets have to have the right capability which, in turn, needs to be properly maintained with any changes being communicated to the system operator. From when the industry in Britain was first privatised in 1990, a key document in governing that has been the Grid Code (there are similar documents in other countries). It’s also essential that compliance with the Grid Code is adequately monitored and enforced, something that has always been a challenge, regardless of the technology. (I remember discussing it with colleagues from the US, Canada, Australia and Europe in a CIGRE Working Group in 2010. I suspect that it was also a challenge to get accurate, up-to-date information out of power station managers in vertically integrated days, as well).
Figure 2: Structure of the electricity industry in Britain showing main equipment and voltages (left) and commercial entities (right). Figure: Keith Bell
According to the Spanish system operator, Red Eléctrica, one of the problems on 28 April was that a number of conventional, thermal power stations didn’t do what they were supposed to do in respect of voltage regulation.
Another of the issues in Spain seems to have been a failure to make use of the full voltage control capabilities of inverter-based resources (IBRs). They seem to have been used in only a crude way, leaving it to the system operator to decide on a fixed amount of reactive power instead of allowing the control system to modify it to suit system conditions as they change second-by-second. One of the reasons for that seems to have been a failure to implement a change to statutory operating procedures – equivalent, I think, to Britain’s Grid Code – obliging all owners of generation, including that interfaced using inverters, to provide a continuous voltage regulation service. That change was, as I understand it, first proposed by Red Eléctrica in 2020 but was only approved by the regulator in late June this year and won’t take effect until January 2026. Approval of that change had been top of the list of Red Eléctrica’s recommendations following the 28 April event.
In Britain, the Grid Code requires that any generators connected to the transmission network or ‘large’ generators connected to distribution are capable of controlling voltage through continuous variation of reactive power. The Grid Code also stipulates the maximum positive and negative reactive power that generators must be capable of producing. However, that reactive power range does not reflect what every technology is capable of.
Voltages need to be controlled everywhere on the system and sources and sinks of reactive power properly coordinated – location matters as reactive power only has a localised effect. This was one of the issues in Spain. Both here and in Spain, in a worst case, fossil-fuelled generation might need to be run for its voltage control capability rather than the energy it produces but it needs to be in the right place to be useful for that. If a system planner carries out the right kind of system modelling and sees a need for extra voltage control capability in a certain place far enough in advance, they can invest in reactive compensation. However, in places where there are wind farms, the theoretical capability of IBRs is such that they might not to need to, provided that capability can be fully utilised. This is something that owners of IBRs might attempt to exploit, seeking extra payment for access to enhanced reactive power capability. If provision of more reactive power means a restriction on the amount of energy that the IBR owner can sell, payment of the opportunity cost would seem reasonable. Otherwise, perhaps not.
Another problem in Spain seemed to be a lack of coordination of fast, continuous and slow, ‘lumpy’ voltage control. Banks of capacitance or inductance are useful but they are switched in and out only in large blocks, only when voltages deviate by quite a lot from a target value and only after, typically, tens of seconds. They also need to be available when they’re needed – Britain’s transmission connected reactive compensation only had an average availability of 75% in 2023-24. Transformers can provide a finer-grained control but they also respond only slowly. In principle, we’re better off in Britain because the main rules for design and operation of the transmission network – the Security and Quality of Supply Standard (SQSS) – stipulate that ‘step changes’ to voltages should be no bigger than a certain value. This tends to drive use of at least a certain amount of fast, continuous voltage control.
In other ways, though, we face similar challenges: make sure that the Grid Code (or its equivalent) is kept up to date and fit for purpose, and that it is complied with; and make sure that the impacts of smaller scale generation – connected into the lower voltage distribution network rather than at higher voltages on the transmission network – are well understood, something that’s a challenge when little of it is directly monitored or controlled. As I will discuss in part 2, those impacts include the behaviour of the generation equipment’s own self-protection.
Also in part 2, I will reflect on the substation fire at North Hyde that resulted in Heathrow airport being closed. I’ll describe some of the principles that guide design and operation of power systems and how we can ensure continued resilience of electricity supply.
Keith Bell holds the Scottish Power Chair in Future Power Systems at the University of Strathclyde and leads the UKERC research theme on delivering energy infrastructure.