undefined | Better HN

0 pointsadrianmonk4y ago0 comments

It gets complicated quickly, but I think interconnections (if big enough) could have helped with the February 2021 power outages.

At first, you'd think maybe not. If the neighboring states don't even have enough power for their own needs, how could it help?

The thing is, the outage in Texas was severe enough that it went beyond rolling blackouts. My city's utility completely lost the ability to rotate at all, so while some sections of town never lost power (those with essentials like hospitals), others had no power for ~48 hours straight.

Neighboring states did have problems, but they didn't have widespread outages. What could have been done is to have rotating outages in a much greater geographic area in order to prevent Texas from having continuous outages.

Not that the other states would have liked that, but on a technical level, I think it could have done something.

0 comments

2 comments · 1 top-level

salawat4y ago· 1 in thread

That's completely because of failure to Network partition.

Think about it. Power Delivery is the OG wired network. You will have a wire. Unlike with signaling, power delivery is default on (cycling to deliver AC power) instead of default off (line clear for signaling).

If you really want to discriminate priority branches and be able to fine-grained load shed, that means, no building off of high priority trunks, and you'll have to string some extra wire to accomodate being able to centrally isolate the high priority loads from the low priority loads short of doing something like inline smart meters connected to mains coming into a home (i.e, teleconnected at-the-pole termination).

The Texas fiasco, as I understand it, came from a combo of poor isolation of high-priority trunks from low priority loads. It's common for cities to just wire nearby businesses off the hospital's high priority trunk, so the lights stay on in empty buisnesses instead of being in the "potentially sheddable" pool.

But no. Nobody wants more wiring. Too expensive. This is what you buy when you do that.

adrianmonkOP4y ago

That's part of it. In their report[1] after the incident our utility called that "sectionalizing". From their status update[2], I think they have done some already and will do more.

It sounds like there's some low-hanging fruit like adding switches at key points without running more wire. They may be planning to do more than that too. I'm not sure.

But it gets more interesting than that! There are actually two things that can make it so that the utility can't shut off a particular circuit. We've already covered critical loads like hospitals. The other is circuits that are configured for what that report calls UFLS.

This bit from their report (p. 22) explains it:

> Austin Energy’s portfolio for Manual Load Shed is limited, as many circuits are typically protected from Manual Load Shed due to the existence of Critical Load Customers or the UFLS status of the circuit.

> ...

> ERCOT also requires each electric distribution operator, including Austin Energy, to designate circuits for UFLS. Austin Energy must designate 25 percent of its load in three blocks of circuits (5 percent, 10 percent and 10 percent) that will automatically drop offline if the ERCOT system frequency drops to certain thresholds. Insufficient power supply, high demand or a combination of both can cause drops in system frequency. If frequency deviates too far from the tight system requirements, the physical reality of electricity can cause system instability and grid collapse. UFLS is designed to prevent these frequency changes from causing an uncontrolled cascading blackout of the ERCOT grid. Due to the quick actions that are needed, UFLS is automated without operator involvement. The UFLS settings are programmed into under–frequency relays (a method by which electrical devices known as relays sense frequency drops in the power system and cause the circuit breakers to trip). In accordance with ERCOT requirements, UFLS–designated circuits are generally not available for Manual Load Shed, as UFLS acts as circuits of last resort for Load Shed to avoid this system collapse.

In other words, if load gets too high, normally they respond by manually turning off circuits. But, as a last resort fail-safe, some circuits are set up to automatically shut off. They're not allowed to include those circuits in rolling blackouts either because they'd lose their fail-safe.

So if critical load is X% and UFLS is 25%, then you can only do rotating outages among 100% - (25% + X%) of circuits. And sectionalization can make sure X% isn't higher than necessary, but it can't do anything about 25%.

So sectionalization helps, but it's limited.

In theory, I think they could break that limit if they did something tricky. I assume they just chose 25% of circuits and installed UFLS relays only on those. Instead, they could have UFLS relays on all circuits and have a way to disable (bypass) or enable those relays. Then they could dynamically rearrange which circuits are UFLS or not as they do rotating outages. Power up a non-UFLS circuit, enable UFLS on it, disable an equal amount of UFLS on another circuit, shut that circuit off. Then repeat.

But in practice, I think that may just be too error-prone and risky. Too much intricate juggling things around during the chaos of an emergency. It goes against the idea of having a fail-safe.

Anyway, at the bigger picture level, you can really do any/all of these things. Add more generation capacity and you are less likely to hit the "we can't even rotate anymore" limit. Tie larger grids together and you can make more people less miserable. Make local grids more granular and you can rotate better. It just depends on how many layers of defense you want to have.

If you got this far, thanks for reading my long, rambling comment!

----

[1] https://austinenergy.com/wcm/connect/482f26ba-7c94-465a-8a00...

[2] https://austinenergy.com/wcm/connect/440e5f7f-6806-47ed-b5e8...

j / k navigate · click thread line to collapse