[hfkajshfaks]

It isn't really though. A cascading failure can knock out an entire grid. The Northeast blackout is an example.

  A line goes out. 
  while (there some lines are still live):
    The current redistributes itself among neighboring lines.
    Neighboring lines trip
  end
  BlackStart()

BlackStart() is fun. Having your plants and grid down is like having a dead battery (plants use ~30% of the electricity they gen on themselves) and no one to give you a jump.

Usually the hydroelectric plants are started first. Arg. is lucky, they have a lot of those. Otherwise you need to fire up diesel generators (which hopefully have been maintained).

You give power to the grid first so you can start another plant.

Then you feed hospitals and continue firing the thermal plants.

Throughout all of this you have to keep phase stability (no grid, no phase reference). I'd assume this would limit renewable's use-fullness for re-start (a thousand 5Mw turbines out of phase are best left idling)

As a last step you turn on your nuclear plants (Arg. has a few). These are last because of nuclear poisons that accumulate from decaying wastes that weren't burned while the reactor was off.

It's a bloody hard problem.

[6thaccount2]

Technically there is a last step you didn't mention. At some point after your island is ready, you'll want to reconnect with other islands. I've never done it in real life, but have done so in many simulations.

I forget the exact calculations involved (it's been years and I'm no longer involved with that area of operations), but you want to make sure your frequency matches the grid you're connecting with to within a certain amount. I think the stations which connect neighboring areas usually have extra monitoring equipment for a variety of energy accounting reasons, but also to measure the frequency on each side to see if it's a go or no go. I believe they use synchroscopes (https://en.wikipedia.org/wiki/Synchroscope) to do this.

Edit: it looks like y'all started talking about this below and I missed it, but the information is good, so I'm leaving it up.

[sunkensheep]

To connect two regions, we align the phase of the two islands. As the voltage across our switch bank falls to zero, we close the switches. It's not a big difference from between synchronizing a generator. The phase is aligned by requesting small changes to generator output, and measurement is done by synchrophasor.

[nnq]

    > Neighboring lines trip
    > As a last step you turn on your nuclear plants

...don't get it: why would you turn off your nuclear power plants, even if the grid failed?!

Won't it make more sense to keep it running as usual maybe just using the power to generate steam that you throw into the atmosphere? ...mainly because stopping and restarting a nuclear power plant is expensive af.?

(It still boggles my mind why are electric grids so focused on efficiency at the cost of resiliency... with abundant nuclear + hydro you could easily over-produce a bit, I'm 99% sure that making electricity dirt cheap for consumers by gov controlled over-production would help fight climate change even if it appears wasteful: there will be huge incentive to replace petrol burning trucks, buses, cars etc. when electricity it dirt cheap... and then prices can be gradually ramped up after everyone made the conversion.)

[pjc50]

I'm not an expert, but there's an interesting discussion on Quora: https://www.quora.com/Can-an-American-nuclear-power-plant-bl...

The key factors seem to be:

1) the plant must be capable of running as an island. It seems not all of them are; I suspect the problem is the plant consuming a variable amount of self-power e.g. 1MW being connected to a 100MW generator results in a very unstable system.

2) whatever safety event causes the problem must not trip at the generators themselves but disconnect the plant cleanly from the grid. Since these events are by definition exceptional, this may be very hard to avoid.

> with abundant nuclear + hydro you could easily over-produce a bit

Both of these are exorbitantly expensive to build, and hydro requires specific geology.

[snowwindwaves]

Most of the plants I've worked on can't island themselves by design. It costs more to do it and sometimes requires additional or more expensive equipment, more attention to details in the design, and there is no point for a lot of generators that just get paid for MWh and they probably aren't going to generate many MWh islanded over the life time of the plant. Some generators even get paid for the energy they would have generated if the grid is down.

How many people have a manual transmission car just so they can pop the clutch if the batter is dead.

[hfkajshfaks]

I'm not in this field (university friends are), but reactors require triple redundancy for their power supply, including the grid.

A better question is, should we have nuclear power at all? I took classes in nuclear engineering, by friends are nuclear engineers, ect. but it's not obvious to me that cheap electricity is worth it.

[nnq]

> not obvious to me that cheap electricity is worth it.

Why would you say that? I mean sure, if you can't manage to get cost-effectiveness right then it's not worth it. But otherwise dirt-cheap and readily available electricity is absolutely needed to solve the world's problems! Think about what will happen when large swaths of land loose access to drinking water after climate + pollution wreck things up and the only alternative is ocean water filtration and desalination plants that will need huuuge amounts of power! Also think about massive irrigation and civil engineering projects that will be needed to fight desertifications etc. And to keep heavily populated costal cities above (the new, higher) sea level. That's terrawats upon terrawats of energy that will be needed to constantly shuffle earth and water around, and maybe even to power huge city-scale air filters etc.

We have and still are wrecking up the planet to such a high extent that sooner or later the bill will arrive and we'll need to put heavy effort into re-terraforming Earth to keep it inhabitable... And there really is no other answer than nuclear here, while cities filled with LED-lights and low powered devices and highly thermally efficient buildings can run on renewables, heavy civil engineering projects and land decontamination and producing drinking water etc. will need waaay more power than our blanket of communication equipment and drones.

We could decide not to, but then we'll end up fighting for resources and instead of nuclear power with it's inherent risks we'll get... nuclear wars!

[segfaultbuserr]

> >...don't get it: why would you turn off your nuclear power plants, even if the grid failed?!

> reactors require triple redundancy for their power supply, including the grid.

Thanks for posting. I learned something interesting today.

I can imagine the consequences of a total loss of power in nuclear reactors...

[mulmen]

Thanks for this, it is very informative and not something I had considered before.

I'm curious about the statement about renewables though because it seems that with wind or solar the production is very irregular even in normal operation. Doesn't that mean there has to be some technology to balance individual generator differences which would make re-starting those installations easier?

Re: Nuclear, how much energy is required to start up a nuke plant? If decaying toxins are a risk shouldn't those be prioritized? I might misunderstand what you mean by toxins, does that reduce production for some time?

[Filligree]

"Nuclear poisons" refers to neutron-absorbing substances produced through transmutation. When the reactor shuts down, it will continue to produce these for some time after the shutdown -- Xenon-135, particularly.

This proves problematic during reactor startup. Since it's absorbing neutrons, the controllers would need to retract the control rods further out than during steady state in order to achieve a replication rate of 1.0. However, xenon-135 stops being a nuclear poison after absorbing its neutrons... which means the replication rate will increase, and the reactor might run away.

This is part of what happened in Chernobyl, though they'd also disabled most of their safeties. (And dismantled the others.)

Regardless, nuclear poisons are something to keep a very careful eye on.

[dredmorbius]

True.

Xenon poisoning was discovered at the Hanford, Washington, B reactor during the WWII Manhatten Project.

https://www.osti.gov/opennet/manhattan-project-history/Event...

[sunkensheep]

There can be issues with a line-commutated inverter or asynchronous generation (wind-farms might have both) if the fault level is too low (synchronous plant too small or too distant). More recent research has shown the benefits of adding grid support (reactive power control, fault response) to the control system of renewable generation.

In future I would expect renewable and storage elements to provide black-start frequency reference.

[apthnz]

What do you do for a living, and how can I learn more about it? This stuff fascinates me and you've explained it really well.

[6thaccount2]

The user is probably a transmission level grid operator. These exist at most decent sized utilities (Entergy, AEP, Duke), although it is possible there is a TOP (transmission operator company) that is not the same as the utility.

There are also organizations like the RTO/ISO that also act like air traffic controllers for the grid by keeping it reliable. They also are usually responsible for committing and dispatching generation for dozens of utilities collectively to save money (by banding together, you need to carry less reserves or backup power and can leverage economies of scale). In North America these organizations are (CAISO, SPP, ERCOT, MISO, PJM, NYISO, and ISO-NE). All of these organizations practice black-start drills (I used to help run them) and have NERC certified operators.

[hfkajshfaks]

"The user is probably a transmission level grid operator. "

Close. My dad's a power engineer, specifically grid protection.

[6thaccount2]

If he's doing grid protection he's probably in distribution engineering and they would be pretty involved with a black start drill from as in they or the line crews with their boots on the ground as not all of that procedure can be initiated remotely I'd imagine.

The transmission level engineers and operators will call the shots as far as how the high level stuff goes down. In other words they'll issue a directive to get unit x online and then close in breaker b and clear a path to unit y while picking up load at z. The actual process from the end of the people doing that work is likely to be significantly more detailed.

[hfkajshfaks]

"If he's doing grid protection he's probably in distribution engineering and they would be pretty involved with a black start drill from as in they or the line crews with their boots on the ground as not all of that procedure can be initiated remotely I'd imagine."

And has a nosy son ;)

[MisterTea]

Honestly, I know that entire procedure myself as it's listed on a few different sites, one being the wikipedia blackout article. Then again this kind of stuff interests me so I read up on it.

[6thaccount2]

I'm not sure you fully understand, or maybe it's me that misunderstands what you're saying :)

Each utility generally has its own black-start procedures. One utility might start two diesel generators (only really used for black-start) and then slowly start energizing the path to various loads and other generators. That utility and it's parent RTO/ISO train on that scenario several times a year and have the procedure printed in a binder within reach.

Something I've found interesting is that many of these plans are ~35 years old and we're created before computers were super common place for these kinds of things. The point being people weren't doing graph and optimization algorithms to determine the optimal paths. There has been a little university research and national lab research into this recently and I've heard they've helped certain utilities that they worked with determine some improvements, which is always exciting.

[sboctrl]

The EPRI Power System Dynamics tutorial is a free resource that is really solid https://www.epri.com/#/pages/product/1016042/?lang=en-US.

[6thaccount2]

Can confirm. This is one of THE best references out there for explaining a lot of key power systems topics. I didn't think it was publically available though?

[sboctrl]

It is one of the few products that is available to everyone. And frankly one of best deliverables from all EPRI programs. Period.

[hfkajshfaks]

Lol. I'm a computational material scientist, so take everything w/ a grain of salt.

My dad's a power (line protection) engineer, everything I know is from talking to him.

[davps]

Right now I'm reading this book about this topic: Power System Stability and Control by Prabha Kundur https://www.amazon.com/System-Stability-Control-Prabha-Kundu...

[mjevans]

The renewables depends on how they're connected to the grid. If they're DC charging a battery bank then presumably said bank can be treated as a single source so they'd be useful. Though it might be more efficient if they were also able to synchronize and direct-connect to the grid as AC.

[segfaultbuserr]

> BlackStart()

I love how the people coming from different fields invented their own word for "the process of starting a system through external intervention because it cannot support itself at this point".

Kickstart. Jumpstart. Bootstrap. And now I learned a new one: blackstart.

[lostlogin]

How is grid phase synchronised?

[niftich]

The following videos from BTCInstrumentation demonstrate this succinctly.

[1] Synchronizing AC generators -- Part 1 (introduction and sync lamps)

[2] Synchronizing AC generators -- Part 2 (strobe light view)

[3] Synchronizing AC generators -- Part 3 (sync and unsync)

[4] Synchronizing AC generators -- Part 4 (phase deviation)

The first video explains all the concepts and shows a sync, but the later videos bring additional conditions and detail.

[1] https://www.youtube.com/watch?v=RGPCIypib5Q [2] https://www.youtube.com/watch?v=sFohkp2aaU4 [3] https://www.youtube.com/watch?v=GRk_qJxaxh8 [4] https://www.youtube.com/watch?v=RT1ySBc-Bls

[toomuchtodo]

https://hackaday.com/2017/07/05/how-do-they-synchronize-powe...

Phase is monitored on both sides of the connection, and high voltage contactors complete the circuit when tolerances are met.

[amelius]

But if you have a grid with a large number of nodes, how do you impose which node is running at the correct (reference) frequency and phase? Do the edges of the grid impose a master-slave relationship?

[amaterasu]

I'm trivialising the problem, but whichever station energises the grid first becomes the reference. An operating grid has what may as well be infinite inertia (it isn't infinite of course), so new stations coming online must match it (or try and change the phase of the grid to match theirs). I visited a power station in Melbourne many years ago, and they have the shattered remains of a turbine and generator that was connected with a phase mismatch.

[hfkajshfaks]

"In the event that there's a phase mismatch, I assume that gets physically transmitted back to the generating device?"

In case of phase mismatch, you have a short circuit and the weakest element along the short gets fried.

[ethbro]

In the event that there's a phase mismatch, I assume that gets physically transmitted back to the generating device? (Mode subject to type of device) Or am I over-trivializing it?

E.g., will an out of phase turbine have a physical resistance against it, as a consequence of the rest of the grid?

[Filligree]

Of a sort. If it's a phase mismatch of a few degrees, the generator may attempt to physically rotate to match the phase.

This kills the generator. Also, less amusingly, anyone nearby at the time.

If it's a larger mismatch, then that's more likely to fry the wiring before too much other damage happens. On rare occasions generators get connected while they are (exactly) 180 degrees out of phase; this is a dead short, and its internals will instantly vaporize. The vaporized copper may then proceed to explore the outer limits of the generator's physical integrity.

(Disclaimer, etc: I've only worked with DC generation, AC would be somewhat different especially if it's three-phase.)

[sunkensheep]

If the phase is not quite matched, you get power swing and mechanical oscillation. If you get it very wrong expect pole slip http://supachaisabua.blogspot.com/2013/09/2.html?m=1

[amelius]

Ok but imagine a circular grid of N nodes. Following the circle, the stations are connected to the grid. Of course there is a small, inevitable mismatch at every connection. Then, when connecting the last node, you might get a problem which can only be solved by adjusting the settings at the other nodes.

[pjc50]

It turns out that this can be a substantial problem, resulting in wasted loop flows: https://iopscience.iop.org/article/10.1088/1367-2630/18/10/1...

Generally systems try to avoid too much loop power in the first place for this reason, setting up a directed but acyclic graph with some redundancy links carrying zero or near-zero power as standby.

[rtkwe]

The grid doesn't really pass through each of the points though so you don't really get the frequency mismatch between one side of a node on the grid and another. Instead of a ring of nodes it's closer to a ring with nodes hanging off of it so you never really get two frequencies observed at one power station because it's only connected to the rest of the grid at a single point.

[sboctrl]

The reference frequency is interconnection wide. To bring a three phase ac generator online the terminal voltage, phase rotation, frequency, and phase angle must be within some tolerance.

[snowwindwaves]

What type of power plant has 30% of nameplate rating on station service loads?

I've been to about 30 stations under 50 MW and the biggest station load might have been 300 kW if they had all the heaters, fans, pumps, crane, welders, air compressors, and gates operating at the same time.