[londons_explore]

> But the biggest long-term risk may be the unexpected tripping of solar resources during grid disturbances,

This is entirely the electricity grids own doing. They told everyone who wanted solar that they must use inverters that are super safe and self disconnect if anything unexpected happens. The regulations were written super tight, with frequency limits, harmonic limits, reactive power limits, rate of change of frequency limits, residual current limits, DC injection limits and vector shift detection requirements, presumably because rulemakers were hoping that having a massive thick rulebook would make rooftop solar too expensive. Unfortunately for them, all the complexity is pushed to software, and when that's written once it can be deployed lots almost free.

Fast forward 10 years, and this stuff is widely deployed, meeting all the very strict criteria. Except the grid only stays within those strict limits during the good times. When the grid is stressed, for example low frequency due to insufficient generation, it will go outside the frequency limit, and all rooftop solar generation statewide will disconnect all at once. And now the grid has lost gigawatts of solar production, making a minor incident far far worse. The end result is a cascading failure and within a matter of seconds half the state has a blackout.

Smart rulemakers back in the 2010's would have dropped islanding protection, and vector shift protection (they actively destabilise the network). They would also remove underfrequency protection (again - destabilises the network) and slacken substantially undervoltage protection.

The only protections important to leave in are rate of change of frequency (upwards only), overfrequency, and overvoltage.

What's ironic is there is a more sensible requirement they could have required to make rooftop solar cost more and add substantially to grid stability. They could have required solar systems have a load line equivalent to coal. That pretty much means if the grid frequency decreases, the solar systems must output more power. But solar can't output more power than the sun provides. So in turn, to meet this requirement all solar systems would have to be constantly throwing away say 10% of capacity, so that it could be delivered if and only if the frequency suddenly dropped.

[toomuchtodo]

Perfect comment on the topic. With that said, newer inverters do have “grid fault ride through” capabilities where they will continue to operate during excursions from nominal operating windows, and some regulators are more hip to it than others.

https://www.iso-ne.com/static-assets/documents/committees/co...

https://www.solarpowerworldonline.com/2019/03/smart-inverter...

Grid forming inverters (versus the traditional grid following) are at the finish line of being proven, and paired with distributed storage providing ancillary services (including frequency response), regulators simply need to be dragged to the present to solve for this (distributed energy orchestration, rapid system segment isolation and recovery during transient events [“self healing” for the marketing folks], etc). This is one of those tug of wars between very conservative regulators and a quickly innovating industry (DER power controls).

https://www.energy.gov/eere/solar/articles/powering-grid-for...

https://www.nrel.gov/news/features/2020/renewables-rescue-st...

https://www.nrel.gov/docs/fy21osti/73476.pdf

https://www.eia.gov/todayinenergy/detail.php?id=50176

[malchow]

The Enphase IQ8 is now in general release, and is grid-forming.

During sun, a site can island with perfect 60Hz AC power from the ASIC-driven IQ8. No battery –– and certainly no grid –– needed.

[toomuchtodo]

I spec’d my 21kw system with IQ8+s for this reason. It’s future proofing distributed generation by being software defined (Enphase’s Ensemble system). Highly recommend them to anyone considering residential solar, and go for as big of a system as you can (and your utility will allow, if they limit based on usage history; tell them you’re buying an EV if necessary).

[bitexploder]

We ended up with 16kW and solar edge stack. 16kw was about all we could fit on the roof. Was a nightmare getting Xcel to approve it. Took them 2 months. We have 20 kWh of battery to buffer and back us up in the case of extended outages. It is a nice system. Anything over 10k in Colorado requires a lengthier approval process.

We also moved the house entirely to heatpumps. They practically sip energy for cooling.

[toomuchtodo]

Excellent choices!

[bitexploder]

It is kind of fun :)

[kkfx]

Consider the peak power of the panel vs the maximum power of the microinverter... My very small p.v. if converted to microinverters from string inverters would be FAR LESS productive due to the maximum power limit of the microinverters, surely they allow "free" panels, mixed with any orientations, many different numbers, even different kind of panels, but they cut much the output to a point they are IMVHO interested only if you can't make chains due do shadows / physical location issues...

[malchow]

No, these days microinverters aren't the bottleneck. The IQ7 can handle up to a 460 W module, with 366 VA output. The IQ8 can handle up to a 540 W module, with 384 VA output. 97.6 efficiency.

[kkfx]

Mh, from Enphase catalogue in my country (France) I do not see such data, the IQ8 are not on sale here and IQ7 variants arrive at maximum at 315 VA for the IQ7X-96-2-INT witch is the top performer on sale here. Anyway, it's still less than classic string inverters even if they give "resilience" and "installation freedom"...

[sbierwagen]

Do you have any battery storage? How well do islanded microinverters handle startup surge on electric motors, or is it the case that you need a 20kw+ array to handle a refrigerator cycling on?

[toomuchtodo]

No battery storage. 1:1 net metering with my utility. 21kw array is about 30% over what we’d need for annual consumption, but (Q Cell) panels were the smallest component of total system cost and they help keep the attic 5 degrees F cooler. Also, I’m required to carry $1M in liability insurance for anything over 10kw. Even if I have credits forever I want to maximize production to offset utility natural gas generation during daylight hours. Not needed but installed a soft start on the 4 ton AC condensing unit for longevity and to reduce start surge. I will likely add battery storage in the future when it declines in cost for outages.

The system is not yet configured to support solar only utility disconnected operation due to supply chain issues (additional components, rush to PTO), so I cannot speak to surge start capability of the IQ8+ micros in island mode.

[malchow]

Every IQ micro and battery you add adds more ability to deal with current surge requirements. On my 8.5Kw system with 30KwHr of Enphase LFP batteries, I can handle turning on AC when I have full sun and partially charged batteries without going to the grid.

AC is a difficult load. Get the inrush current here: https://www4.enphase.com/en-us/support/faq/how-support-hvac-... and you will be able to figure it out.

Your refrigerator will be fine. Its locked-rotor amperage requirements are small compared to AC.

[ahartmetz]

This made me think "What about SMA Solar, they are supposed to be a technology leader?"... and yes, they do have all the grid stabilization features (and an island feature in at least some models, not mentioned there):

https://www.sma.de/en/partners/knowledgebase/sma-inverters-a...

[elric]

Interesting. Can it sense when the grid is present, and can it match phase?

[amluto]

Last year, I had the opportunity to talk to some PG&E-associated contractors [0] who were replacing the distribution transformer serving me. I asked them whether they trusted the anti-islanding protection in everyone’s inverters and whether they would like me to turn off my main breaker. They laughed and said that they couldn’t care less. They were going to intentionally short the secondary circuit if they were doing dangerous work, and if anyone’s inverter was trying to energize it, that was the inverter’s problem.

I’m genuinely unsure what purpose anti-islanding actually serves.

(My inverter is moderately intelligent and formed a one-house microgrid all by itself. That being said, this capability may be at odds with helping the grid survive a major disturbance. When my inverter decides to disconnect from the grid, it cannot support the grid regardless of what its software and the regulators think.)

[0] By which I mean line workers at a company that PG&E contracts with to maintain their distribution network. Apparently PG&E outsources real work. Go figure.

[KennyBlanken]

> I’m genuinely unsure what purpose anti-islanding actually serves.

Scroll up to read about the HNer required to carrying $1M in insurance for having more than 10kW in solar on their roof, which is just...insane. Solar is mired in bullshit to make it seem dangerous, make it look ugly, make it as complex and expensive to install as possible. That's how you end up with regulations requiring what looks like an electrical substation on the side of your home, covered in neon-colored labels. Gotta make sure someone knows the system is "RAPID SHUTDOWN EQUIPPED" from 100 feet away!

Electrical grid operators want you to buy electricity, not make it for them. Their nightmare is becoming "just" a grid.

Their biggest nightmare, however, is you realizing that you no longer need them at all. Lots and lots of people, especially those out in suburban or rural areas, could easily go off-grid these days. Homes are better insulated, heatpumps are quite common, solar costs a fraction of what it used to, lithium ion battery prices are crashing and LiFePo batteries are getting commonplace, etc. So what's a utility to do? Push microinverters that require a grid connection and so on.

[Syonyk]

> I’m genuinely unsure what purpose anti-islanding actually serves.

I think it mostly prevents inverters from trying to destroy themselves feeding into a dead short, or creating weird instabilities - the grid segment is either up or down from the inverter's point of view. Any sane inverter will detect it's trying to feed a dead short and shut down. And if you're grid tied and your grid segment is down, it looks like a dead short.

It's a useful enough filter for people who understand power systems and power system work, because, as your linemen pointed out, if safety is a question, the workers simply create a bolted short across all the phases and neutral, and you're not simply not going to make that ring.

And, yes, migratory line workers are a thing. I don't get it, but relatively few people I've talked to out here (Idaho Power territory) actually work for Idaho Power - it's mostly migratory contract workers doing the power pole inspections and such.

[Veliladon]

> I’m genuinely unsure what purpose anti-islanding actually serves.

Because a typical solar inverter can't handle transient loads. It just spits out whatever power is coming in from the panels. It can't shed if too much is coming in and it can't magically create power to fix a shortfall. The power has gotta go or come from somewhere.

If you have a battery you can soak the transient load and disconnect from the grid and operate as an island but without the grid to act as a buffer for the load the inverter is useless on its own.

[amluto]

Shedding excess power is very, very easy for solar, especially as compared to any other power source. The MPPT can move away from the maximum power point. (Compare to, say, wind or hydro where moving a turbine away from its optimal speed and torque can be quite destructive without extreme care. I visited a small hydro installation with a monstrous space heater to dump power in the interval between when a load disappears and when valves can adjust to reduce the flow of water.)

I am curious how SolarEdge’s inverters reduce output, though. It’s not fundamentally hard, but the inverter does not have appear to have a particularly high speed data connection to the MPPTs, and I haven’t found the underlying mechanics documented anywhere. I’m guessing that the inverter pulls the incoming voltage down such that the MPPTs hit their preprogrammed output current limits and curtail production.

(SolarEdge’s system can’t just dump excess power into a battery — their common configurations have the battery behind a DC-DC converter with considerably lower capacity than that of the inverter’s output. The DC-DC can’t fully absorb the solar string’s output if the sun is shining and an island’s load goes away. The battery itself likely could, at least for a little while, but there isn’t any way for the power to get there.)

[Syonyk]

Indeed, came here to write basically this. Solar tripping off is solar obeying the rules set on them by the utility that is "You will trip if anything so much as looks funny."

And then a few decades later, they figured out this was a terrible idea and the newer 1741-SA/CA Rule 21/etc stuff with far greater ridethrough for reasonable disturbances showed up, but I don't think it's the majority of solar installs.

I should probably get a Grid Guard code for my SMA inverters and use the revised ridethrough rules - my utility handwaved at it when I installed stuff, and it's using the CA Rule 21 stuff, but they now have far more explicit guidelines on what they want programmed in the inverters.

[heartbreak]

It’s such an own goal to take massively distributed and resilient power generation with no inherent single point of failure and just give it one anyway.

[makomk]

The generation isn't naturally massively distributed and resilient, that's the thing. Assuming that the grid will always be there and stable simplifies the problem quite a bit: all the inverters have to do is sync up with it and then dump all the power they've got into the grid. If you want the solar generation to be the grid, rather than just boosting it, suddenly you have to deal with all the hairy control theory problems involved in grid operation and stability, except that you have to do it using millions of generators rather than thousands - and the behaviour of those generators is closely correlated, so you still have to deal with problems like a large proportion of generation going offline at once as well as the extra complexity of a much bigger system.

Old-fashioned spinning generators also naturally act to stablilize the grid by pulling energy out and dumping energy back in through their inertia as necessary to counteract changes due to varying demand and supply. Solar and wind don't inherently do that: if you want similar behaviour from them, that requires additional software and sensing, and obviously for home solar that would have to be done in a distributed fashion which complicates matters further. (The grid is not in fact at a single frequency and phase everywhere, and can develop all kinds of interesting and undesirable oscillations.)

[dpierce9]

With modern inverters the ability to provide reactive power for voltage support could be signal driven. It doesn’t inherently require overbuilding or sacrificing production all the time, just when needed. Further, it could be compensated thus justifying the lost production.

[aaaaaaaaaaab]

So noone was bothered to run a simulation to see the second-order effects of their regulations on the grid?

[londons_explore]

The conversation went something like this: "well as long as all the solar disconnects when we have an emergency, then we are left with a grid without solar. And we already know we can handle that fine".

The rulemakers were expecting solar to be a failure and there to never be anything more than a handful of geeks with it installed. The problems only become major when 1+% of statewide generation is solar. 1% is (very approximately) the amount of generation an electricity grid can suddenly lose without failure.