v0.7

Start with the evening it nearly went wrong.

On a Friday afternoon in August 2019, a lightning strike knocked two large generators off the grid within seconds. More than a million people across England and Wales lost power: trains stranded, traffic lights dark, hospitals on backup. None of the physics was new that day — but the event is the clearest way to watch that physics at work.

It all comes back to one number. Every socket in Britain hums at the same rhythm. Fifty times a second, the current flips direction. Fifty hertz. It's the heartbeat of the grid.

? What is frequency?

Grid frequency is the rate at which the alternating current flips direction: 50 times per second in Britain. Every synchronous generator on the system spins in lockstep with this rate, so any mismatch between supply and demand pushes the whole grid faster or slower at once.

Read more on NESO →

Except it isn't really fifty. Look closer.

Frequency moves constantly. Every kettle switching on, every cloud passing over a solar farm, every gust through a wind farm: all of them push and pull at this number. The grid operator's job is to keep it close to 50.

? What drives the frequency back to 50 Hz?

Grid frequency stays near 50 Hz because the operator continuously matches generation to demand. When demand creeps up, frequency drifts down, and contracted generators are paid to ramp up automatically: a service called frequency response.

Read more on NESO →

But close has limits.

Inside the green band, life is normal. Cross into amber and the operator scrambles. Drop to 48.8 Hz, the red line, and the grid starts cutting customers off automatically to save itself — exactly the cliff edge it raced toward that August evening in 2019.

? Why does over-frequency look different from under-frequency?

The two failures are physically different. Under-frequency means generation is short and the system is slowing. If it slows too far, generators trip off and the shortfall grows, cascading toward blackout.

Read NESO's Frequency Risk and Control Report →

Want to see how easily this happens?

Try breaking the grid. Trip a power station and a thousand megawatts of generation vanish in an instant. Trip a major customer and a thousand megawatts of demand vanish the same way.

? How big are real disturbances?

A thousand megawatts is roughly what a single large gas plant or one of Britain's HVDC interconnector cables produces. Britain plans for losing up to 1 800 MW at once, the largest secured infeed loss.

Read NESO's Security and Quality of Supply Standard →

Something is resisting the fall.

Across Britain, hundreds of enormous spinning machines are turning in perfect lockstep with that 50 Hz heartbeat. When something trips, they don't stop. Their momentum keeps the grid going for crucial seconds while help arrives.

This is inertia.

Think of all that spinning steel as one giant flywheel bolted to the grid. A flywheel is just a heavy wheel that, once it's turning, resists any sudden change to its speed: load it in an instant and it gives up its momentum gradually instead of stopping dead. Britain's generators, turning together, are exactly that — the flywheel that holds 50 Hz steady.

That momentum comes from synchronous machines — gas, coal, nuclear. As wind and solar take over generation, fewer of them are online to lend their weight, and the flywheel grows lighter.

? Why doesn't the spinning just stop instantly?

A large generator's rotor weighs hundreds of tonnes and spins at 3 000 RPM. The kinetic energy stored in that motion is enormous. When the grid loses generation suddenly, that energy briefly flows out into the network as the rotor slows down.

Read EDF's explainer →

So what happens if the flywheel gets lighter?

Drag the slider. The flywheel shrinks. The grid is still calm for now. Push the slider and try the disturbance again. Same trip, same lost megawatts. Watch what changes.

? Why are wind turbines and solar panels not part of the flywheel?

Wind turbines spin, but their rotors are not electrically locked to the grid. Modern turbines connect through inverters: power electronics that translate blade speed into clean 50 Hz output.

Same disturbance. Different grid.

A heavy flywheel gives a comfortable dip. A light flywheel plunges straight through amber into red, possibly clipping the blackout line. The number on the right, RoCoF, is how fast frequency moves when something goes wrong. Rate of Change of Frequency.

A heavy grid gives you time to react. A light grid doesn't.

? How fast is too fast for RoCoF?

NESO currently allows the system to experience a RoCoF of up to 1 Hz per second. Above that, some generators may start disconnecting automatically because they interpret the fast movement as a sign that they are no longer on the main grid.

Read the Ofgem investigation →

Which leaves one question hanging: why is the grid getting lighter in the first place? You won't find the answer on the frequency trace — it's in what's feeding the grid.

So why did the grid get lighter?

Pull up everything feeding the grid on an ordinary day and it looks reassuringly busy: gas, wind, nuclear, a slab of solar at midday, a little hydro and biomass, imports sliding in over the interconnectors. Add it up and supply meets demand. By that measure, nothing is wrong.

But a tally of megawatts only tells you what is making energy. It says nothing about what is holding the frequency steady. For that, the labels on the chart matter far more than the total at the bottom.

Collapse the legend to two colours.

Strip the fuel names away and only one distinction really counts. Some of these machines are synchronous — their rotors are physically locked to the 50 Hz system, all turning as one body. The rest are non-synchronous, joined to the grid through inverters that rebuild a clean waveform from whatever the wind and sun happen to be doing.

? What counts as synchronous here?

Following the dataset's own grouping: gas, coal, nuclear, biomass and hydro are counted as synchronous; wind, solar, storage and the interconnectors are treated as non-synchronous or inverter-based.

A gigawatt is not always a gigawatt.

A gas turbine and a wind farm can each deliver the same gigawatt, and on the energy chart they sit at exactly the same height. Underneath they are nothing alike. The turbine's rotor is part of the grid's spinning body; lose it and you feel the weight go. The wind farm lives behind an inverter — quick and precise, but holding none of that physical momentum in reserve.

So as the old thermal fleet retires and inverters take its place, the grid keeps making power but quietly stops storing it as motion. The line is still there. The flywheel behind it is thinner.

The weight has a unit.

That stored motion is something engineers actually measure, and it carries an awkward name: gigavolt-ampere-seconds, or GVA·s. It is the kinetic energy held in every synchronised machine turning together. More GVA·s means a heavier grid, and a heavier grid sags more slowly when a generator or an interconnector trips. That slowness is time — time for controls to catch the fall before anything has to be switched off.

? Is GVA·s the same as generation?

No. Megawatts describe the power being produced right now; GVA·s describes the rotating energy sitting in reserve in synchronised machines. A system can be flush with megawatts and still be dangerously short of inertia.

Two versions of the same number.

This is also where the inertia figure quietly splits in two — worth holding onto if you wander over to the data afterwards. There is the market inertia: whatever the system happens to be carrying because those machines were economic to run anyway. And there is the outturn inertia: what the grid actually ends up with once the operator has stepped in. The gap between the two is the cost of stability, made visible.

Because when the market alone falls short, NESO buys the difference — and increasingly it buys it from machines built for nothing else. Synchronous condensers, essentially power stations with the fire taken out, spin in step with the grid and lend their weight without burning a thing.

? Market vs outturn — what's the difference?

Market inertia is the pre-intervention picture: what the running plant would provide on its own. Outturn inertia is the realised total after NESO's actions. When outturn sits above market, the operator has paid to put extra weight on the system.

Browse NESO's inertia data →

Running closer to the edge, on purpose.

The operator keeps an operational floor for inertia — a line the system is not meant to drop below. What is striking is that the line keeps moving down: around 140 GVA·s, then 130 in early 2024, then 120 that summer. That is not physics turning kinder. It is NESO admitting it can run a lighter grid because everything around the flywheel has improved.

Some of that improvement is bought outright — the synchronous condensers, and Scotland's first grid-forming batteries like Blackhillock, standing in for the missing steel. Some of it is just rules: after the 2019 blackout, embedded generators had their trip settings relaxed so they would stop bailing out at the first wobble. And some of it is reflex — battery-backed services that throw their full weight in within a second of a fault, catching the drop a heavy grid used to absorb on its own.

? How does Britain replace the missing inertia?

There is no single fix. NESO contracts dedicated stability services (inertia and short-circuit strength), grid-forming inverters set a voltage and frequency reference rather than just following the grid, and faster frequency-response products replace reflexes the old spinning fleet provided for free.

Read NESO's Frequency Risk and Control Report →

Held together on purpose.

Lowering the floor saves money and makes room for more wind and sun. It also leaves less time when something goes wrong — which is the whole reason the apparatus of condensers, contracts, faster response and new HVDC links down the east coast exists at all: to manufacture, at cost, the stability the old grid threw in for nothing. None of it quite replaces what coal gave away for free, so Britain is doing all of it at once.

When it works, you don't notice. The dial barely moves, the kettle boils, the lights stay on. That isn't luck — it's a premium the system now pays every second so a disturbance never becomes the crash.

For the numbers underneath all of this, head to the analytics page: the falling inertia, the market-versus-outturn gap, and what low inertia ends up costing.