There's literally a single person who adjusts the whole power grid according to second-by-second fluctuations in power which result from half the nation boiling water for tea after the most popular TV shows finish: https://www.youtube.com/watch?v=slDAvewWfrA

That was actually really interesting

I don't know how it works in the UK, but in the US, there are three major interconnections within the country, and each interconnection has system operators. The system operators are responsible for managing the grid, including the frequency. I've been in several US-based power control rooms, and the current system frequency always gets pride of place on the instrument panels (along with the current system load and sometimes the grid time). System operators are also very concerned about maintaining the balance between their generation and their load.

The grid time is itself an interesting concept. Because the grid runs at a known frequency, it's possible to use it as a time base for a clock. The grid time is considered to be the current time for such a clock. It basically integrates the grid frequency error over time.

Regarding generators that lag the grid, it's not quite as simple as generators turning into motors. The generators all have real time governors that adjust their throttles based on their current operating speed. (The operating speed is directly linked to their operating frequency, for the vast majority of units.) In the case of a generator operating too slowly, the governor immediately requests more input power to compensate. Because this happens grid-wide, this gives the entire grid the ability to quickly respond to short-term fluctuations in operating frequency. Longer term frequency response is handled by dispatching individual units up and down to balance power and load.

In the event that a generator cannot adequately respond, it will trip (shut down) and disconnect from the grid entirely. This is a protective measure that keeps the generator from skipping cycles. (You can think of this as the electrical equivalent of gears skipping teeth, and can be hugely destructive to the equipment.)

The US Grid is split into three relatively unconnected grids, east coast, west coast, and texas. Each section has one plant designated to keep the frequency, sort of a "master plant" that is used to balance instant load. All of these are synchronized across the US using phaser measurement units that are timed to GPS.

The other cool thing about AC power is that you can send power based on the relative phase of a node on the grid. This means that a node at a lower AC voltage can send power to another node at a higher voltage by adjusting the relative phase. Mostly this is done by switching capacitor banks at various transmission nodes to adjust the reactive power component at that node.

They have isolators to stop that from happening - the generators can't draw a current.

I don't necessarily think this is what he meant, but generators can sometimes be operated in synchronous condensing mode. In this mode, the generator is run as a motor, which consumes real power but can help with reactive power issues important to grid stability.

No, electrically generators and motors are the same circuitry. If you have two generators hooked to the same output line, and one gains a few degrees of phase angle on the other, the one lagging draws power from the leader in proportion to the phase angle difference.

Yes but with a bit of electronics you can stop this from happening. Source: Institute of Physics.

Curious what electronics will prevent a flow of AC electricity.

This is what is used in the UK: http://en.wikipedia.org/wiki/Synchronization_%28alternating_...

49.960 is pretty exact; the frequency can reasonably deviate much further. Some systems try (tried?) to keep a 'daily average' of 50.000 hz, but the grid will run quite okay at 49.960 hz, at 49.900 hz and at 49.000 hz; not that it's recommended but it's an option if there's a supply/demand disbalance.

What would you expect the tolerances to be on grid frequency?

Well within a fraction of a cycle/second. One of the big reasons is that a generator's mechanical speed is directly tied to the electrical frequency. This includes the prime-movers that actually drive the generators. Many of these large machines don't react well to operating far from their design speed, which can cause huge issues.

.01 cycle was my memory of a threshold from some years ago.

Part of the problem is lower frequencies don't meet as much impedence in inductors (e.g., transformers), and another part is that interconnects can have control system stability problems keeping them synced.