Producing exactly when we consume

Frequency, forecasting, flexibility, storage and backup. Theory, formulas, worked examples and applications.

Every second has to balance

Why the grid cannot simply "store the surplus"

Unlike almost any other supply system — water in a reservoir, gas in a tank, goods in a warehouse —, the conventional electrical grid does not store meaningful amounts of energy in itself. This means that, at every instant, the amount of electricity generated by all connected plants, renewable farms and batteries must almost exactly equal the amount consumed by all connected homes, factories and other users at that same moment. If demand exceeds generation, grid frequency tends to fall; if generation exceeds demand, frequency tends to rise. Both cases, if not corrected quickly, trigger automatic protections that can disconnect entire generators or consumers — the mechanism developed in detail in this section's blackout leaf.

A toolkit with very different response speeds

Maintaining that balance requires combining tools with very different reaction speeds. A conventional thermal plant keeps its own stored fuel, but can take minutes or hours to start from shutdown or to adjust its output significantly — it is not an agile source for responding to a sudden change. Hydropower plants and batteries, by contrast, can adjust their output within seconds, making them the preferred tools for correcting fast, unexpected imbalances. Solar and wind depend on the resource available at each moment and cannot generate more than the sun or wind allows at that instant, though they can be forecast with reasonable notice and deliberately curtailed if the grid needs less generation. Flexible demand — shifting in time the consumption of industrial cooling, electric-vehicle charging, or certain manufacturing processes — acts from the consumption side, not generation, and can be as valuable as adding a new plant.

Reserves that are almost never used, but cost just the same

Keeping the system stable requires always having available a certain amount of reserve capacity — generation or storage able to step in within seconds or minutes if something unexpectedly fails. That reserve capacity may operate only a few hours a year in the worst case, but the cost of building it, maintaining it, and the materials it took to manufacture it belong equally to the whole electrical system, even though they almost never show up in a specific plant's generation cost. It is, in a sense, insurance: paid continuously to avoid having to bear the much larger cost of a system collapse.

The grid changes nature with more power electronics

Traditional thermal plants and large hydropower contribute to the grid, almost as a side effect of their mechanical operation, a physical property called inertia: the rotating mass of their turbines resists sudden frequency changes, giving operators more reaction time in the face of a sudden imbalance. Modern solar and wind farms, connected to the grid through power electronics rather than a turbine spinning directly in sync with the grid, do not naturally provide that same inertia. This does not make the grid inherently less safe, but it does require new technical solutions — inverters able to electronically emulate that inertia, ultra-fast-response batteries — to replace a service that used to come almost for free alongside conventional thermal generation.

The right question is not "is there backup?"

Any modern electrical grid has, by design, some kind of contingency prepared for imbalance: the relevant question is never whether a backup plan exists, but how much backup is available, how quickly it can activate, what emissions that backup carries if it is thermal, what cost it has to keep it operational even if barely used, and what simultaneous events — a heatwave that spikes air-conditioning demand right when the wind drops, for example — were accounted for in its design. The more diverse the sources, the more interconnected with neighboring regions, and the more consumption a system can shift in time, the less dedicated reserve capacity it needs to build from scratch.

The question is not «is there backup?»

In favor of a diverse system: combining sources with different response times — fast hydropower and batteries, slower thermal backup, flexible demand — reduces the amount of dedicated reserve that has to be built from scratch for any single event.

To watch with the renewable transition: replacing thermal generation and large hydropower — which provide inertia almost automatically — with solar and wind connected through power electronics reduces the system's natural inertia, requiring new technical solutions (inverters that emulate inertia, ultra-fast-response batteries) to maintain the same robustness against sudden imbalances.

The mistake of assuming "more renewables" automatically means "less stable": it is not renewable technology itself that causes instability, but the speed at which inertia and backup capacity are withdrawn from the system without an equivalent replacement. A system with a lot of renewables and enough synthetic inertia or storage can be as stable as a traditional one; one that withdraws synchronous generation without that replacement cannot.

Useful questions for any grid plan: how much reserve exists, how fast can it respond?, what emissions does that reserve carry if it is thermal?, what simultaneous events — a heatwave and an interconnector failure at once, for example — have been studied?, how much demand can be shifted in time without losing service?

The numbers, with their source

When low inertia makes everything happen much faster: South Australia, 2016

On 28 September 2016, a storm brought down several high-voltage towers in South Australia. The resulting voltage dip triggered automatic protections on several wind farms, which disconnected to protect themselves — a sudden loss of 456 MW of generation —, followed shortly after by the loss of the interconnector to the neighboring state of Victoria, another 613 MW. Up to this point, a severe failure but of a type the grid had absorbed before on other occasions. What turned it into a total collapse was system inertia: at that moment, around 30% of generation came from wind power and close to 50% of the whole system was connected through power electronics, not traditional synchronous generators physically spinning in sync with the grid. With so little inertia available, frequency fell far faster than usual, exceeding the ability of automatic load-shedding systems to arrest the fall before frequency dropped below 47 Hz — well below the 50 Hz target — and the entire state lost power.

Sources: Wikipedia — 2016 South Australian blackout; The Blackout Report.

Why this case is not an argument against renewable energy

The root cause of this blackout was not "too much wind," but a specific combination of two technical factors: wind turbines of a design from that era with limited ability to stay connected through severe voltage dips, and a system with very little synchronous backup inertia at that particular moment. More modern turbine designs, and synthetic-inertia or ultra-fast-response battery solutions — far more widely deployed today than in 2016 — directly address both problems. The case is, in fact, a good example of why grid stability depends on how the transition to more renewables is managed, not on whether it happens.

What this table is missing

There is no universal figure for "how much reserve is needed" that applies to any grid: it depends on system size, its level of interconnection with neighboring regions, the diversity of its generation sources, and the local demand profile — which is why each grid operator calculates its own reserve requirements specifically, revising them as the mix of connected generation changes.

Sources