Inertia, Voltage and Grid Strength
Spinning mass, fault current and reactive power, and what changes when generation sits behind inverters. The Iberian blackout of April 2025, and why the popular explanation for it was wrong twice over.
Words the industry uses are marked like this at the sentence that defines them, so you can tell a term you will hear on a desk from a phrase this book happens to be using.
At 12:33 on the afternoon of Monday 28 April 2025, the Iberian Peninsula stopped. Three minutes earlier the Spanish system had been running normally, roughly 32 gigawatts of generation against about 25 gigawatts of demand on a mild spring day. Then, over a few seconds, something close to 15 gigawatts of generation disconnected itself. That is around 60% of what Spain was producing. Spain and Portugal went dark together, and 99% of Spanish demand was not restored until six the following morning.
Within hours the explanation had already been settled by people who had no data. There was a lot of solar on the system that lunchtime, and the grid had gone down, so the solar had done it. The more technical version of the same claim, which sounded better and was repeated by people who should have been more careful, was that the system had too little inertia: too few large spinning machines, too many solar inverters, not enough mechanical ballast to ride through a disturbance.
The final report of the expert panel convened by ENTSO-E, the association of Europe's transmission system operators, was published in March 2026. Drawn from 49 people across transmission operators, regional coordination centres, ACER (the EU agency coordinating national energy regulators) and national regulators, and chaired from two unaffected systems, found that neither of those was the cause. What it did find is more useful than the popular account, and the failure it describes is a market failure as much as an engineering one.
Voltage, and the odd thing called reactive power
Two terms carry the whole of this chapter, and neither is common knowledge outside the industry.
Voltage is electrical pressure, the force pushing current along a wire. Where frequency, from Chapter 1, is a property of the entire synchronous area and reads the same everywhere at once, voltage is local. It differs from one substation to the next, and every piece of equipment attached to the system is built to tolerate only a narrow band around its rated figure. Too low and motors stall and equipment misbehaves. Too high and insulation fails, and protective devices start disconnecting machines to save them.
Reactive power is the quantity used to control it. In an alternating current system, part of the current flowing does no useful work at all. It sloshes back and forth each cycle, building and collapsing the magnetic and electric fields inside motors, transformers and the lines themselves. That component is reactive power, measured in volt-amperes reactive rather than in watts, and it delivers no energy to anybody. What it does is set voltage. Equipment that supplies reactive power pushes local voltage up; equipment that absorbs it pulls voltage down.
Generators do either on command, by adjusting excitation, which means changing the strength of the magnetic field inside the machine. A generator asked to absorb reactive power is being asked to hold the voltage around it down. That instruction, and whether it was followed, is what the rest of this chapter is about.
What actually happened
The panel identified at least fifteen contributing factors, which is itself the first lesson. What it describes is a combination: oscillations, both a local mode near 0.63 hertz and an inter-area mode near 0.2 hertz, meaning a slow swing of whole regions against one another, together with gaps in the control of voltage and of reactive power, inconsistent voltage regulation practice between operators, and rapid output reductions and generator disconnections inside Spain.
The mechanism that turned a manageable disturbance into a blackout was overvoltage, and it ran away from the system. Voltage began to rise. Generators, protecting themselves, disconnected. Each disconnection removed a machine that had been helping to hold voltage down, so the voltage rose further, which tripped more machines. Several of the actions taken to address what began as a small instability made it worse. That is a self-reinforcing loop, and once it is running there are seconds, not minutes, in which to stop it.
The plants that failed to hold voltage down were the conventional thermal ones. The panel found non-compliant reactive power control by exactly the machines whose job it was. Wind and solar did not cause the blackout, and the panel found they helped restore supply afterwards.
The president of ENTSO-E put the conclusion in one sentence: renewables were not the problem, voltage control was. And on the second, more sophisticated theory, the assessment of ENTSO-E and the Spanish system operator was that a lack of inertia was not a root cause either. The popular explanation was wrong twice, and the second error is the more instructive, because it was made by people who knew what inertia is.
Two different things the grid needs
The confusion is understandable, because a grid does need both of the things being discussed, and they are easy to run together. They are not the same, they fail on different timescales, and they are supplied by different equipment.
Table 6-1: Inertia and voltage support are different problems
| Inertia | Voltage and reactive power | |
|---|---|---|
| What it defends | Frequency | Voltage |
| Against what | A sudden mismatch between supply and demand | Too much or too little reactive power on the network |
| Timescale | The first seconds, before any control system acts | Continuous, and local rather than system-wide |
| Where it comes from | Rotating mass, or grid-forming inverters emulating it | Generator excitation, synchronous condensers, capacitors, reactors, inverters |
| Travels? | Shared across the synchronous area | Barely. It has to be supplied near where it is needed |
That last row is the one to hold on to. Frequency is a property of the whole synchronous area, so every machine in Europe sees the same frequency at the same moment and inertia anywhere helps everywhere. Voltage is local. Reactive power, the current that sloshes back and forth each cycle without delivering any energy, does not travel usefully over long distances, which means voltage support is not a commodity you can import from a neighbouring country when you need it. It has to exist near the problem, which means somebody has to have built it, and somebody has to be obliged to operate it correctly on an ordinary Monday when nothing appears to be wrong.
What inverters do change
None of this means the concern about inverter-based generation, meaning solar, batteries and most modern wind turbines, is invented. It is real, and Chapter 4 introduced the equipment. A synchronous machine is physically locked to the grid: it turns in step, and its stored kinetic energy is available to the system whether anyone planned for it or not. An inverter has no such property by default. It measures the grid and follows it.
A grid built mostly of followers has two genuine weaknesses. It has less inherent inertia, so frequency moves faster after a disturbance. And it supplies less fault current, which matters because protective relays were designed to detect a fault by the large current that flows into it. The current a given point on the network can push into a fault is itsshort-circuit level, and how firmly that point holds its voltage when something changes is what engineers mean bygrid strength. The two go together, because a point that can deliver a lot of current into a fault is also one whose voltage barely moves when a machine trips. A synchronous generator delivers several times its rated current into a fault for the first cycles, because the physics of a spinning machine makes it do so whether or not anyone wanted it. An inverter delivers barely more than its rating, because it is protecting semiconductors that would fail in milliseconds. So a network with ample generation on paper can still be weak at the instant a fault has to be cleared. The engineering answers exist and are being deployed: synchronous condensers, which are essentially generators spun up with no fuel, purely to provide inertia and reactive power, and grid-forming inverters, which impose a voltage waveform rather than following one. The name has nothing to do with condensing anything: condenser was the older English word for what is now called a capacitor, and a synchronous machine spun with no load and over-excited draws a leading current, which is what a capacitor does.
Illustrative traces on a 50 hertz system, shaped to the behaviour described in the text rather than taken from any particular event. The dashed line marks a level at which automatic disconnection of customers typically begins. Note that inertia does not fix the imbalance, it only slows the fall.
So the fragility argument survives the report. What does not survive is the specific claim that this event proves it. Iberia was not a demonstration that a renewable grid cannot hold together. It was a demonstration that a service which every textbook lists, which every grid code requires, and which barely anybody is paid properly to provide, will eventually not be provided.
Why this is a market chapter in disguise
Everything above is physics, which is why it sits in Part One. But the conclusion is economic, and it sets up Chapter 13.
Energy is easy to buy. It is measurable, it is fungible within a grid, and an exchange can clear it in five-minute intervals. Stability is a set of capabilities that must exist before they are needed, that produce nothing on the vast majority of days, and whose absence is invisible right up to the moment it is catastrophic. A market that pays only for megawatt-hours is paying for the one thing that is easy to measure and none of the things that keep the measurement possible.
Note also the two halves of what went wrong, because they fail independently. One is procurement: whether enough voltage support and reactive capability was contracted to exist at all. The other is compliance: whether the plants that had agreed to provide it actually did so on the day. The Iberian panel found a problem in the second, which no amount of additional procurement would have fixed. Chapter 13 takes up both, and Chapter 19 returns to the regulatory aftermath, because the durable consequence of 28 April 2025 is a set of new obligations.