Now suppose some of these incandescent lamps are turned off. The power being consumed is reduced, so the power being taken from the shaft rotation is reduced, so you need less effort to spin the shaft. If the mechanical input is the same, the shaft will spin faster.
And here is the trick: there's a mechanism which senses the shaft increasing its speed, and reduces the mechanical input, for instance by partially closing a water gate, until the speed returns to normal. The same happens in the opposite direction: if more of the incandescent lamps are turned on, the shaft slows its speed, the mechanism senses this and opens the water a little more, until it's back to the correct speed.
What we have on modern continent-wide power networks is basically a scaling up of that principle. So the answer to your question is: the extra power is not produced, the production automatically reduces until only 95 units of energy are being produced, matching generation and demand.
What happens if that's not the case? Simple, if production and demand are not within a few percent of each other, protective relays cut the generators and/or loads until it matches, to prevent damage to the generators, power distribution devices, and loads.
What you suggest turns out to be impossible. The amount of energy that goes in and out must be identical. So this means:
* Either less goes in: The required mechanical effort to make the turbine turn goes down. It will have to dissipate the extra energy as heat and catch fire, or it will become less efficient, and have stronger winds behind it as less wind energy has been used.
* Or more goes out: The transparting cables are not perfect, they are a kind of very low resistor. They do convert electricity to heat. A short is an extreme example of this. In this case, a smaller version will happen: cables will get slightly hotter.
As I understand it, monitoring and balancing the electrical grid is a very complicated problem.
Recently, and discussed on hacker news, the opposite happened in Europe: Not enough power was available on the net because of political circumstances. All the generators had to push a little harder, so they all slowed down a tiny bit. Result was the 50HZ of the european grid went a very tiny bit down to 49.999...Hz, and some clocks lost 6 minutes of time as a result.
This news had an interesting discussion recently:
Continuing frequency deviation in the Continental European Power System (entsoe.eu)
478 points by Aissen 80 days ago | 323 comments
https://news.ycombinator.com/item?id=16535067
https://www.entsoe.eu/news-events/announcements/announcement...
https://arstechnica.com/tech-policy/2018/04/european-grid-di...
So if you have a gas turbine, and you see it slowing down, you feed it a little more fuel and vice versa.
Unfortunately, solar and wind energy make this impossible. So for wind, the turbine has brakes, but there are videos on youtube where they fail or get overloaded, resulting in burning turbines. I guess solar does something similar (Edit: the braking not the burning of course).
In practice, the non-renewable energy turbines like gas are the easiest to control so these are used to balance the grid.
On top of all of this, running a turbine costs a lot, so there is a lot of grid management going on to decide who can generate which amount of electricity. Decisions made at this level better match the actual load on the grid, or the catastrophic consequences of the post above happen.
Brning turbine: https://www.youtube.com/watch?v=6y1Ib0GM5x8
Most wind turbines built today with a capacity of over a few hundred kW are not synchronised to the grid frequency.
They run at whatever speed they can run at with the current wind speed, which is converted to DC, goes down to the root of the wind mill and gets converted back to regular 50Hz AC when it's put into the grid. They only thing that could happen there is that this converter station blows up.
Also, those modern turbines always have turnable rotors that can just be moved into a longitudinal position relative to the airflow, stopping immediately.
As I understand it:. It used to be that every turbine on the grid was synchronized and people found out how the grid communicated its energy behaviour by changing frequency. This happened as emergent behaviour, almost by accident.
As the grid operators came to rely on it, newer technology has to emulate it.
But today, most of Europe shares the same grid, which was impossible at the time. There have to exist technological advances to make this possible. And solar doesn't even have an inherent frequency, it's DC. So what you say sounds very reasonable.
But I hope never to see from close by what happens if modern energy equipment fails catastrophically.
You can use energy from the renewable sources to charge up storage systems, which in turn can be controlled more easily. That way you achieve flexibility without relying on non-renewable energy sources.
For gas it is a few minutes at least.
Nuclear and coal mostly can be used just as baseload.
EDF were talking about possibly needing to change this in France in the future, since otherwise their heavily nuclear-based power grid couldn't cope with the increased levels of highly variable renewable power they're predicting. The nuclear plants just can't adjust their output power fast enough or deep enough.
Also, many wind turbines are variable pitch: the angle of the blades can vary, so the controller can put them in a less efficient position to reduce the turbine power, or "feather" them to produce no power (the brake is also used in this case to completely stop the rotation).
Most electricity networks have a "system operator" whose job is to keep electricity generated track electricity consumed very closely. (An organization, with many automated systems, not a single person).
In practice, on a large network, the most noticeable side effect would be AC frequency fluctuations. In a severe case of imbalance, sensors/switches on the network would disconnect part of the network to protect the underlying equipment (probably causing blackouts, if not cascading blackouts).
The operator predicts demand and tells each of the many electricity generators how much electricity to generate. Some types of generation are easier to ramp-up o or throttle-down: for example, hydro and natural gas are very easy, while nuclear is not. Residential roof-based solar makes things complicated because it is now a new part of the system that is no longer under the operators control.
Some networks also have access to sinks of energy, the most common being pumped storage (usually in connection to a hydro facility), or even, industrial consumers who are willing to "ramp up" on demand.
On top of all this is usually a market layer (depending if the electricity generation is privatized or not). For example, in Ontario, the IESO (independent electricity system operator) asks for bids every 5 minutes from all the providers, and based on a modified dutch-auction, selects which providers are allowed to produce and sends them the corresponding instructions. Of course, taking into account various other technical requirements to decide who gets to power or not. In this kind of market, nuclear power, because it cannot throttle up or down, typically bids 0 or negative (because the actual price set is based on the last most expensive bid to get in). During some hours of the year, the market price of electricity is sometimes negative, because nuclear is the only bidder. In Ontario, the government phased out all coal production (yay, no more smog!), but... had to build a lot of new natural gas plants to be able to do the fast ramp-up/throttling the network requires on some days of the year. Most of the year, these plants sit idle.
Grid balancing is a tough act.
Imagine you are driving a car downhill and the car starts speeding up. You let off the accelerator in order to not go too fast. Worst case, you apply the brakes and dissipate the energy as heat.
Suppose you have a generator somewhere operating at 1MW: 1MW of mechanical shaft power being put in (torque x speed) and 1MW of electrical power (current x voltage) being taken out. Then demand drops slightly to 990kW. The "braking force" of the generator coils reduces, but the input torque does not, so the generator accelerates.
This usually feeds back into the rest of the system - throttles in steam and gas turbines.
If too much power is being produced the system will eventually fail and everything will go dark. The entity responsible for the system, the system operator, will pay someone to produce less in order to prevent that from happening. As you can imagine most producers are happy to receive more money for producing less!
A single hydro plant can only do so much though - it can not reduce the flow of water below zero and the system will fail if it ever runs out of primary regulation reserves. The system operator will then have to pay an additional party to produce less in order to prevent that from happening. This is called "secondary" or "tertiary" regulation and is not real-time.
In Germany, there are 3 different steps to handle this (I guess it's pretty much the same in other countries too).
1. The primary reserve power is short-term fine-grained capacity that takes effect automatically (is not requested by the TSO[1] explicitly) if the frequency leaves the +-0.01 Hz band around 50 Hz. The capacity is then provided proportionally to the distance, full capacity is supposed to be provided when the frequency goes under 49.8 or over 50.2 Hz. The tricky part is that the capacity needs to be supplied in both directions which can for example be done by modulating the frequency differences on a power plant running slightly under its full capacity. In Europe, this capacity is provided jointly spread over the whole grid and totals roughly 3 GW in both directions (e.g. simultaneous full outages of two large power plants).
Broadly speaking, primary reserve power is supposed to stabilise the grid, i.e. to stop a frequency in/decrease.
2. Secondary reserve power is supposed to kick in after 5min, latest 15min of primary reserve power usage. This is done automatically by the individual TSOs (in Germany we have 4 because we can :)) if the reason for the deviation is in their grid, and sends a request to a controller in the providing power plants, which have to provide 1MW 1min after the request and full (sold) capacity after 5min. It's not as immediate as primary power as the to-be-produced capacity is not directly calculated from the frequency, but the request messages still change quite frequently, usually a few times per second.
This part is supposed to move the frequency back to 50 Hz.
3. At last, if secondary reserve is drawn for too long, minute reserve power is activated. This is just a straight "you have 15min to provide 50 MW, 30min onwards after you start". It's essentially just an economical optimisation on the process, since due to the higher lead time, conventional power plants can actually start up to provide it, while for both secondary and primary reserve power they need to react so quickly, that a full ramp-up wouldn't be efficient.
Back to your main question, on how negative reserve power is provided then concretely, this could be:
- Downmodulating running power plants (primary, secondary and minute reserve)
- Charging batteries (likely just primary and secondary reserve since minute reserve would exceed capacity)
- Starting up additional consumers (minute reserve)
- Running turbines without actually producing any power (worst case minute reserve)
- Shutting of wind turbines by turning their rotor blades (likely just secondary, can't be provided guaranteed for long due to changes in the wind, would be fast enough for primary reserve but can only work in one direction)
[1]: Transmission System Operator
The power grid is already a giant distributed control system and that‘s only going to increase in the future. Really interesting stuff.
that lets get people get paid for taking load in exactly these situations! You should probably apply to get paid for operating that emergency siren system.