UFLS is mostly intended to buffer transient loss of power supply (for instance, tripping of several large generators). If the grid operator knows they are facing an inadequacy of generation supply and all resources have already been called in, they will start to shed load under manual operator action to avoid UFLS activation. Before doing this they will declare an emergency which, generally speaking, requires all generators to make best efforts to supply as much energy as they can to the system.
Think of it like automated emergency braking on a car vs driver braking. AEB is great, but if you can already see that you're gonna hit something, just hit the brakes right now instead of waiting for AEB. By the time UFLS kicks in, you're already in dire straits and have only moments before reaching an unrecoverable state.
This exists on a few different timescales, and the faster-responding contracts trigger at higher frequency deviations.
Primary operating reserve triggers at +-200 mHz and needs to be fully available within 30 s and for up to 15 min, secondary operating reserve has to take over (and be fully available within 5 min), and the tertiary operating reserve has to be fully available within 15 min.
The entso-e map: https://www.entsoe.eu/data/map/
The problem in ERCOT is demand massively exceeding available supply over a period of days. You need to carry your primary operating reserve all the time, because you never know when a generator or transmission line may trip.
When there's simply a lack of energy supply to meet demand on a steady state basis, you still have to preserve your primary reserves for other unexpected events and cannot call them in to meet longer term (30+ minutes) demand. So load shedding happens once available reserve dips below a certain threshold (generally around the single of the largest single contingency on the grid which may be the largest single generator or a line that feeds multiple large generators).
ERCOT hasn't run out of primary reserve, but they are going to have to shed load in order to maintain adequate reserves. You always have to shed load before exhausting reserves, because one moves you closer to stability while the other moves you further away (consuming operating reserve).
At this point, with power prices spiking to thousands of dollars per MWh, any plant that can run is running, including some that have come out of planned outage earlier to help out (and be paid handsomely). Even if you simply invoked some emergency order that "all generators must run until further notice", there simply isn't enough capacity due to the large number of generators on forced outage. There's no way out of that situation except temporary load shedding.
Also, going beyond this, charging spot prices to consumers (both commercial and residential) serves to align incentives. Such as dis-incentivizing the use of resistive heating for extreme cold, as the prices are correlated with the extreme cold, making alternative heating setups for extreme weather more affordable in comparison.
Or, building houses so that extreme cold can be tolerated without damage by allowing pipes to be drained and the remaining sections to be heated separately, so the house as a whole can be unheated (and the people stay with friends).
Longer-term predicted loads can be billed at a fixed rate, but spiky loads, especially those correlated with ones at other customers of the utility, can't be.
You'd raise the prices until demand falls below the limits, instead of needing rolling blackouts. If people can't afford to take a hot shower right now, they just wait. At least in the absence of certain diseases, the human body is fine without showering for months. And as long as your building doesn't take damage, you'd gladly be payed a few thousand $ to put on really warm clothes for half a week.
And depending on the details, you'd even conclude it more economical to have to replace some pipes after water damage, than to keep that area heated.
ERCOT (and others) also implement peak demand charges, where some portion of a large user's overal yearly rate is determined by their usage on days when the grid is under the most stress. Usage for that particular day ends up setting rates for the entire year, which makes it not only risky but also very costly.
https://en.wikipedia.org/wiki/National_Grid_(Great_Britain)#...
A small industry popped up predicting when the 3 half-hour snapshots would be taken so their customers could minimize their use during those periods to minimize their costs for the year.
https://www.havenpower.com/news/what-you-need-to-know-about-...
I know the Datacenter we were in years ago would be asked to go to generators during CA power crisis.
https://en.wikipedia.org/wiki/Ultra-high-voltage_electricity...
https://github.com/pirate/quebec-power-grid-talk
It's one of the only multi-terminal HVDC lines in the world, most of the others are just point-to-point.
They also literally short out the power lines with DC to warm them up and melt the ice off during ice storms, pretty crazy stuff.
https://en.wikipedia.org/wiki/Pacific_DC_Intertie
More interestingly--it dates to the Kennedy Administration.
HVDC has better marginal loss , but you pay a loss up front for conversion
They could receive more money to power down than to work, so they do that when it's interesting enough..
The electricity net is really fragile. You need to guess averages to the system as a whole, so there's not much flunctuation. Renewable energy is an extra variable too.
Some additional reserves can be powered on/off to make the system stable.
Ps. If you cross those averages by 5% for example, you get a hefty fine, so monitoring your power outage is really important.
Source: worked at a company monitoring this for multiple countries.
In the UK, there is a system of load shedding that is tied to the grid frequency. My understanding is that some industrial users agree terms where they get cheaper rates but if the demand outstrips supply, they are first to be cut. This load shedding is implemented by components that monitor the grid frequency, and if it drops below a fixed value, it disconnects. It seemed a nice system to me as in theory it operates just based on the principles of an AC power supply.
Secondly, the UK has its own grid but is linked up to the EU grid with asynchronous HVDC connectors. We get a lot of our energy though those. Is this inefficient?