What Is an Electricity Grid? Supply and Demand Balanced Every Second

Across a synchronised grid, every large generator spins in lockstep, and their combined rotating mass acts as a mechanical buffer. When demand rises, the generators are loaded more heavily and slow slightly, which lowers the frequency; when demand falls, they speed up. That makes frequency a direct, instantaneous, grid-wide measure of whether supply matches demand, available to every operator and every connected device without any communication. Control systems respond at different speeds: inertia from spinning mass absorbs the first fraction of a second, governors on generators adjust output within seconds, automatic reserves respond within minutes, and dispatch decisions handle the rest. In Europe the nominal frequency is fifty hertz and in much of the Americas sixty, and operators aim to hold it within about a tenth of a hertz. If it falls far enough, automatic load shedding disconnects blocks of customers to save the system, and if that fails, generators disconnect to protect themselves and the area goes dark.

The chain from generator to socket has distinct stages, each with a purpose:

• •Generation, from thermal plants, hydro, wind, solar and storage, connected at various voltages depending on size
• •Step-up transformers raising voltage to hundreds of kilovolts, because power lost as heat in a conductor depends on the square of the current, so transmitting at high voltage and low current cuts losses dramatically
• •Transmission lines carrying bulk power over long distances, which is the part that looks like pylons
• •Substations stepping voltage down progressively, with protection equipment that isolates faults within cycles
• •Distribution networks at medium and low voltage reaching streets and buildings
• •A control room running state estimation, contingency analysis and dispatch, deciding which plants run and holding reserves against the largest single failure the system must survive

The competition between direct and alternating current in the 1880s, conducted with considerable public theatre, was settled by transformers. Direct current at the time could not easily change voltage, which meant it had to be generated near where it was used and could not be transmitted far without unacceptable losses, restricting a power station to a radius of about a mile. Alternating current can be transformed up for transmission and down for use with a simple static device, which made central generation at scale possible and decided the outcome. The irony is that high-voltage direct current has returned for specific purposes, since it has lower losses over very long distances, requires no synchronisation between the systems at each end, and can run undersea where the capacitance of a cable makes alternating current impractical, which is why interconnectors between countries and long links from remote generation are increasingly direct current with conversion stations at each end.

Wind and solar alter the operating problem in three specific ways rather than simply adding supply. They are variable, producing according to weather rather than to demand, which increases the need for flexible resources that can fill gaps. They are uncertain, forecastable well but not perfectly, which increases reserve requirements. And they are non-synchronous, connected through power electronics rather than as spinning machines, so they contribute no inertia, which means the grid's natural buffer against sudden imbalance shrinks as their share grows and frequency moves faster after a fault. The responses are well identified: batteries, which respond in milliseconds and now provide a large share of fast frequency response in several markets; synthetic inertia from inverters programmed to imitate the behaviour of a spinning machine; synchronous condensers, which are essentially generators spinning without producing power, added purely for inertia; demand response, paying consumers to reduce load; and interconnection to neighbouring areas, which shares both variability and reserves.

Large blackouts follow a consistent pattern: an initial fault, a failure to contain it, and then cascading as each subsequent element becomes overloaded and disconnects to protect itself, all within minutes. The north-eastern American blackout of 2003 began with a tree contacting a line in Ohio and a software fault that suppressed the alarms in a control room, and left fifty-five million people without power. The Indian blackouts of 2012 affected over six hundred million. Restarting is the hardest part, since most power stations need electricity to start, so recovery relies on black start units, typically hydro or gas turbines able to start unaided, which then energise the network in stages while carefully matching generation to the load being reconnected, a process that can take days. The system is designed to survive the loss of its largest single element without customer disconnection, which is the criterion the reserves are sized against, and the failures occur when several things go wrong at once or when the operators cannot see what is happening.

A grid stores almost nothing, so generation must match consumption continuously, and the frequency of the alternating current is the instantaneous grid-wide indicator of whether it does. Spinning generator mass buffers the first moments, governors and reserves respond over seconds and minutes, and automatic load shedding protects the system if frequency falls far enough. Alternating current won because transformers allowed high-voltage transmission, and renewables complicate operation by being variable, uncertain and contributing no rotational inertia, which batteries and synchronous condensers now supply.