[huhnmonster]

I was wondering if someone could explain the countermeasures for such an event. Obviously, as the article states, producers are being shut off in the regions with surplus, while drains, who can afford to shut off, are shut off in the deficit regions.

Is this an automatic process? Or is it more like someone from the company's energy provider calls them and tells them to shut off some devices? And is there not a potential problem, that if too many shut of at once, you now have a surplus again? Or is it coordinated by one single entity?

[paulmac_ie]

The company I work for provides Demand Response to the Irish grid operator.

In cases like this, our systems would detect the frequency deviation, and shut off loads within 100 milliseonds to reduce the demand on the grid. This helps in cases where demand is greater than supply.

The entire system is automated - the required time frames are so quick that you don't have time for humans to be involved. By the time we're aware that an event has occurred, we've already reduced demand on the grid.

Handling high frequency events where supply is greater than demand is tricker. Sites that have long running generation can be instructed to shut down their generation, but large-scale batteries are probably the best solution in these cases. They can be switched quickly to start charging (if they have spare capacity).

As you've identified, one potential issue is that you can end up over-responding to the event and move from a low frequency event to a high frequency event.

The way we do in in Ireland is that our response is proportional to the frequency nadir. Not everything is tripped off at the same time.

As other posters have noted, the actual frequency deviations that occurred are not that big. 49.7 Hz is not that low compared to normal grid frequency. In fact, some of our systems wouldn't even activate at this level. They would see it, but wouldn't trip off any loads.

[xoob]

I’m curious about this. What kinds of load is your company is running? How do you take them offline and online so quickly without affecting production or leading to long restart cycles?

[paulmac_ie]

We control a mixture of loads, but for Fast Frequency Response (FFR), it's mainly compressor/ motors, with the occassional mains breaker thrown in. Large cement factories, refrigeration companies etc.

As for how we control them so quickly, it's usually via a direct connection to a breaker of some description, or where there's a SCADA system that can trip out the loads in the required time frames. Before the site can participate in these services, we perform extensive testing to ensure they meet the required time constraints.

Our system monitors the grid voltage and current at 8 kHz, and we down sample that to 50 Hz (average grid frequency). We can detect a frequency event across 3 phases within 60 milliseconds. (We check all three phases for multiple cycles to reduce any false positives.)

When we trip out the loads, production stops. We'll notify the client why their loads have been turned off, but the SMS message will usually arrive a few seconds after the trip.

The clients that participate in these services get paid quite well to make their loads available and they're aware of the process. They agree to turn off demand without notice when there's a frequency event.

It's not suitable for every site. We work quite a lot with pharma companies for "regular" demand response, but very few can do FFR. Shutting down a pharma plant with no notice can cost a lot of money in wasted product and down time as they clear the wasted product off the production lines.

[jnsaff2]

The grid frequency is a measure of energy balance in the grid. So anyone can measure anywhere in the frequency domain whether there is a surplus or deficit of generation. This is due to the fact that the grid itself can't store any energy, it has to be balanced all the time.

Imagine the massive spinning generators as a big mass that slow down just a little bit when you switch on a light and then that generator has to add more steam (or open hydro valve or whatever).

So anyone with an accurate enough measuring device can exactly monitor the state of the grid. We use this device [0] for example.

There are generally frequency containment reserves (FCR) that consist of different ways of generation and have their different reaction times, power and energy capacities.

Hydro for example can react in about 15 seconds, battery inverters in milliseconds. Gas turbines in minutes, coal fired plants in hours.

You can also shed energy by switching off loads (Demand side response).

The system operator is responsible for grid balancing in the short term, they have direct facilities under their control and they have contracts with generators and consumers. And there are markets to bid your generation and flexibility.

The markets in the Nordics for example are:

- FCR-N (Frequency containment Reserve - Normal operations)

- between 49.90 - 49.99 and 50.01 - 50.10 (reaction time up to 20s)

- FCR-D (Disturbance) - between 49.7-49.90 and 50.10 - 50.30 (reaction time up to 2 seconds IIRC)

- FFR (Fast Frequency response) - below 49.7 - reaction time 0.6s IIRC

Once a day you bid your capacity for the next 24h (for each hour) and then you measure the grid frequency yourself and when you detect a deviation you activate your response. You get paid for availability and activation separately. There is a ton of qualification and logging you need to do to be able to participate but the activation message is the grid frequency itself, no further communication needed.

Outside frequency regulation there is energy markets where generation and consumption is agreed 24h ahead.

[0] https://www.gobmaier.de/

[bennofs]

> Imagine the massive spinning generators as a big mass that slow down just a little bit when you switch on a light and then that generator has to add more steam (or open hydro valve or whatever).

I have always wondered, would that still hold true if the grid was fully solar-based? There would be no rotating mass in that case.

[netflixandkill]

Yes, although a significant issue with that is there is no spinning mass to "borrow" inertia from, so while modern inverters at solar sites are good at shaping output phases, they have very little capability to absorb significant frequency deviations.

Grid scale battery systems are often used for voltage or frequency stability as opposed to deep discharging as generation offsets, although that will change eventually if batteries get better enough or really cheap LNG stops being a thing.

[jnsaff2]

Yes indeed, inverter based generators (solar and batteries) don't have any inertia. But as long as the grid is still AC the frequency will be the same and the inverters would need to compensate internally. Especially with solar where a cloud going over a solar farm can easily knock off a few MW in seconds the volatility of the grid will increase and need for storage (batteries, hydro, etc) alongside Demand Side Response is getting much bigger.

[brandmeyer]

It does hold true for some systems. Wind turbine generator controls can provide virtual inertia over very short time scales (a second or few) by exchanging energy with the turbine rotor. You can also provide primary frequency reserve in the negative direction (load step-off) using exactly the same ramp control that steam plants use. Typical ramp rate is 100% of rated power for a 5% change in frequency.

However, in order to provide primary frequency reserve in the other direction, you do need additional local storage. You don't need very much. Just 10% of rated power for 15 minutes gets you to very deep renewable penetration.

The trouble isn't with the technology, its with the economics. Once you set a sufficiently high price for frequency support and primary frequency reserve, suppliers will show up.

[Faaak]

These are called ancillary services. There are 3 levels: frequency control (for example dams), primary reverse (fast acting power, in less than 10s), and secondary reserve (slow acting: < 15min).

Participants can either be positive (they consume more energy, for example PHES systems), or negative (they inject or consume less energy).

All these ancillary services are paid (annual auction + per case). Nowadays, it's getting bigger with VPP: virtual power plants, which aggregate small loads (i.e. small ~1MW generators) in order to propose a bigger load to TSOs.

It's all automatic

If you're interested: https://www.swissgrid.ch/en/home/customers/ancillary-service...

[Faaak]

On top of that, newer solar inverters MUST (at least in the EU) reduce power if frequency rises in order to automatically shed power in cases of extreme supply

[Reason077]

"Shutting off drains" is known as demand side response or DSR. It's often cheaper, faster, and more environmentally friendly to pay industrial customers to cut their demand by x MW than to fire up x MW of ancillary generation. And as far as grid frequency/balance is concerned, has exactly the same effect.

Often such customers will have flexible demand in the form of non-critical heating or cooling, pumps that only need to run some of the time, etc, which they are very happy to turn off temporarily in return for extra income.

This is indeed an automatic process, triggered in near-real time in response to signals from the grid.

[bo0tzz]

It is an automatic process, coordinated - as far as I'm aware - by two national energy grid providers (Switzerland and Germany) who have been selected for this role.

[fisian]

The response is automated and every system operator provides some ressources to react: https://en.wikipedia.org/wiki/Operating_reserve

[rtkwe]

They talk about it in the article as contracted interruptible services that got shut down in France and Italy to reduce the draw on the grid.