Tuesday, October 29, 2019

Thoughts on a power outage: what worked, what didn't

We lost our electrical power for 45 hours as a result of the wildfires and high winds in Northern California.  This is by far the longest power outage we have ever had to endure, and we learned a lot about how to deal with them.  I thought I'd share some of the lessons.

What Worked

This outage was a lot easier to deal with than it might have been because we had a lot of warning.  PG&E started sounding the alarm about two days before the power was actually cut on Saturday night.  Because of that we were able to stock up on ice, stage flashlights, and so on.  The ice in particular proved to be very useful because that was the difference between saving some perishables and having them, well, perish.  It turns out that our freezer can last 24 hours without electricity (provided you don't open it) but 48 is pushing it.

LED lighting is just awesome.  I grew up on old-school incandescent flashlights that lasted an hour or two on a set of batteries.  LED flashlights will easily last 10-15 hours, and produce a lot more light in the process.  Battery-powered LED candles provide really nice mood lighting late at night. We also had some wall-mounted emergency lights that worked quite well, though we found that the first thing we wanted to do once PG&E pulled the plug was to turn them all off because the bluish light coming up from the wall was really harsh.  We also had no idea how long they would last because their internal batteries are quite small.  They are intended mainly for the use case where the power cuts off at night with no notice so you can find the real flashlights without having to go groping around in the dark, and for that purpose they worked quite well.

What Didn't

We have two phone lines, one of which is an old-school POTS (Plain Old Telephone Service) line that is supposed to keep working even when the power goes out.  It didn't work.  As soon as the power died, so did the phone.  We even kept an old hard-wired phone to use with that line specifically so we would have comms without power.  That experiment was a dismal failure.  This corresponds to a data point collected from an acquaintance who did a similar experiment with similar results.  The days of reliable hard-wired communications in the face of power outages are apparently over.  And unfortunately, our house sits in a cell dead zone so that doesn't work as a backup for us either.

We had uninterruptible power supplies that kept some of our electronics running for a while, but that was mostly a failure too.  Some of the batteries had apparently aged out and quit after only a minute or two.

For those that kept working, we learned the hard way that having a UPS with a power-out alarm that can't be muted is an incredibly bad idea.  I guess the designers thought that the UPS should be as obnoxious as possible in order to make sure that in the event it was being used to power a computer, the user would know to save their work and shut down.  But there are two serious problems with this theory.  First, when the power goes out, it's pretty obvious.  Even during the day, there's almost always something nearby that's powered by electricity that gives you an indication when it's no longer working.  And second, you might be using your UPS to power something other than a computer, something low-power that you want to keep running for a long time, like a DSL modem.  Also, even if you are using the UPS to power a computer, and even if it's the middle of the day so somehow you miss the fact that the power has gone out, there is no excuse for the thing to go on beeping for more than a few minutes, and absolutely no excuse not to provide some way to silence the damn thing short of chucking it out the window.

(BTW, anyone want to buy a slightly used UPS?)

But the worst problem we faced was that our water heater stopped working despite being powered mainly by natural gas.  It turns out that it also has an electrical element (some kind of blower) and when that's not working, the whole heater just shuts down.  We had residual hot water for about twelve hours, which got us through the first night and the following morning.  But after that we decided to check into a hotel.

This turns out to be the hardest problem to solve for future outages.  But for not having hot water, we could probably survive just about any likely outage.  But cold showers are a show-stopper for us.  To fix this, we'd either need to replace the water heater, or get a generator or a Tesla powerwall.  A generator is kind of loud and obnoxious, and a powerwall is some pretty major coin.  Both would need to be wired into the house in order to power the water heater.

Good thing climate change is a hoax or all this could start to get really annoying.

29 comments:

Don Geddis said...

I've just been thinking about the Tesla Powerwall myself. I actually have solar panels already, but unfortunately when the grid goes down, they need to shut off. (The panels seem to need to have a place to dump all their unused excess power; normally, this goes into the grid. But it has to go somewhere. So when the grid goes down, the panels disconnect.)

So I don't even need a particularly large battery. All I need is a place to push excess power, and then I get to use the solar power at least during each day. That would have been a major improvement in my experience last weekend.

Don Geddis said...

Oh, and I've had the UPS-beep problem as well, in the past. It's up there with stupid smoke alarm beeps that go off at 3am, and take a hour to find. Really, really bad UI design.

Ron said...

@Don:

Maybe one of these?

I'm actually thinking seriously about getting one of these. We saw one in operation in our neighborhood during the blackout and they are remarkably quiet.

Also, yes, smoke detector UI is a disaster. All of our smoke detectors are ganged together so when one goes off they all go off. So when we get false positives, which has happened twice, we have to replace them all. The second time we got ones that have a feature to show you which detector initiated the alarm, but the UI for that is an LED that blinks once a minute. So you have to go to one detector and stand there for a minute to see if the LED flashes, then move on to the next one. I'd like to meet the engineer who thought that was a good idea.

Publius said...

Living your values

@Don
>I actually have solar panels already, but unfortunately when the grid goes down, they need to shut off.

This is because the system was designed to power the utility electric grid, not your house. With net metering, this means your solar array is one of the highest cost generators in the system.

> (The panels seem to need to have a place to dump all their unused excess power; normally, this goes into the grid. But it has to go somewhere. So when the grid goes down, the panels disconnect.)

Not exactly. Consider the IV curve of a solar cell. If the solar cell is irradiated but not connected to a load, it's voltage rises to it's open-circuit voltage, Voc. The power doesn't have to go anywhere.

The issue, I believe, is that your AC inverter and the associated controller is too slow to adjust the voltage to compensate for both the 1) varying load in your house, and 2) the varying supply from the solar panels. Adding a battery creates a buffer that allows a controller to properly control voltage and frequency. If your household load is reduced to approximately that of what the solar panels can supply, you shouldn't need a large battery to achieve this stabilization. A larger battery would allow you to maintain power overnight, or use more power than the array can generate (until the battery is drawn down).

@Ron:
>I'm actually thinking seriously about getting one of these. We saw one in operation in our neighborhood during the blackout and they are remarkably quiet.

With the blackouts, Californians are discovering they love fossil fuels!

You're thinking too small. Since you have natural gas service, get a whole-house generator that runs on natural gas. Whole house generators come in sizes from 22 kW to 48 kW. A 19.5 kW generator will run you about $5K.

As for a non-annoying UPS, the CyberPower CP1500PFCLCD PFC Sinewave UPS System, 1500VA/900W, 10 Outlets, AVR, Mini-Tower is a good UPS system. With holiday shopping season coming up, they often have one-day sales where you can grab one for less than $130. Setup a price alert at camelcamelcamel.

Ron said...

@Publius:

> Californians are discovering they love fossil fuels!

Of course we do. We love Nutella too. That doesn't mean that we think that binging on either one is good in the long run.

BTW, all the links in your comment are broken. You need the HTTPS:// prefix otherwise they become local links.

Publius said...

Living your values

@Don
>I actually have solar panels already, but unfortunately when the grid goes down, they need to shut off.

This is because the system was designed to power the utility eletric grid, not your house. With net metering, this means your solar array is one of the highest cost generators in the system.

> (The panels seem to need to have a place to dump all their unused excess power; normally, this goes into the grid. But it has to go somewhere. So when the grid goes down, the panels disconnect.)

Not exactly. Consider the IV curve of a solar cell. If the solar cell is irradiated but not connectd to a load, it's voltage rises to it's open-circuit voltage, Voc. The power doesn't have to go anywhere.

The issue, I believe, is that your AC inverter and the associated controller is too slow to adjust the voltage to compensate for both the 1) varrying load in your house, and 2) the varrying supply from the solar panels. Adding a battery creates a buffer that allows a controller to properly control voltage and frequency. If your household load is reduced to approximately that of what the solar panels can supply, you shouldn't need a large battery to achieve this stabilization. A larger battery would allow you to maintain power overnight, or use more power than the array can generate (until the battery is drawn down).

@Ron:
>I'm actually thinking seriously about getting one of these. We saw one in operation in our neighborhood during the blackout and they are remarkably quiet.

With the blackouts, Californians are discovering they love fossil fuels!

You're thinking too small. Since you have natural gas service, get a whole-house generator that runs on natural gas. Whole house generators come in sizes from 22 kW to 48 kW. A 19.5 kW generator will run you about $5K.

As for a non-annoying UPS, the CyberPower CP1500PFCLCD PFC Sinewave UPS System, 1500VA/900W, 10 Outlets, AVR, Mini-Tower is a good UPS system. With holiday shopping season coming up, they often have one-day sales where you can grab one for less than $130. Setup a price alert at camelcamelcamel.

Don Geddis said...

@Publius: "This is because the system was designed to power the utility eletric grid, not your house. ... If the solar cell is irradiated but not connectd to a load, it's voltage rises to it's open-circuit voltage, Voc. The power doesn't have to go anywhere."

I already said, in my original comment, that when the grid goes down, the solar panels shut off. So I already know that you can just leave the circuit open. That isn't the point.

You seem to be implying that it would be possible to leave the solar panels connected, but only use a portion of the power generated. I believe you are wrong. I've never heard of any solar panel design that can do that.

As far as I know, solar panels are either "on", or "off". If they're "on", you must find a way to absorb all the generated power. In normal use, the grid is the buffer. If your house uses more power than the solar panels are generating, the excess is purchased from the grid. Alternatively, if your house is using less power than the grid is currently generating, then the excess is sent back to the grid.

When the grid goes down, your only choices are to either use all the power (somehow), or none of it. To keep your house powered via the solar panels, when the grid is down, every system I've heard of needs some kind of buffer in the middle. Ron's suggestion of a Tesla Powerwall is a perfect example.

Did I misunderstand you? Were you not trying to suggest that some kind of "properly" designed solar panel system, could be capable of directly powering your variable-need house, without a buffer? Because, as you claim, "the power doesn't have to go anywhere"?

I would love for you to educate me, if I'm mistaken. At the moment, I'm pretty skeptical. Please provide evidence for your claim.

Publius said...

@Don:
As far as I know, solar panels are either "on", or "off".

If the panels are illuminated, they're "on".

@Don:
>I would love for you to educate me, if I'm mistaken. At the moment, I'm pretty skeptical. Please provide evidence for your claim.

Don, did you miss this part:
The issue, I believe, is that your AC inverter and the associated controller is too slow to adjust the voltage to compensate for both the 1) varying load in your house, and 2) the varying supply from the solar panels. Adding a battery creates a buffer that allows a controller to properly control voltage and frequency. If your household load is reduced to approximately that of what the solar panels can supply, you shouldn't need a large battery to achieve this stabilization. A larger battery would allow you to maintain power overnight, or use more power than the array can generate (until the battery is drawn down).

The issue isn't that you need to absorb all of the generated power. If you have 20 kW panels and are only consuming 10 kW, you'd only need to absorb 10 kW from the panels. The issue is that the panels supply DC current to a power controller which has to do a DC-to-AC conversion, and it has to maintain 120 VAC at 60 Hz to your home. However, the controller is too slow to adjust to variations in both the supply from the panels, and the change in load from your house. This is why you need a buffer. A buffer also prevents a brownout if you're generating 20 kW from the panels, but your house load wants 25 kW. The grid can be that buffer, or a battery can be the buffer. Say it's a battery. You can then sort of think of the panels as charging the battery with whatever they're generating, and the controller drawing from the battery to supply you with 120 VAC 60 Hz. The battery doesn't necessary need to be that large to support the buffer / voltage-frequency stabilizing function; however, if you want continuous power to your home, you probably want one big enough to take you through the night, plus some extra. Say 18 hours of demand. Might was well round that up to 24 hours of demand. If you were willing to just live within the generating capacity of your panels, and be without power at night, the battery buffer could be much smaller, say 15 - 60 minutes of demand (I doubt this would lead to a very satisfying experience).


Don Geddis said...

@Publius: Yes, I totally understand how it all works when you add a battery. (Which is why I am currently considering upgrading my system with a battery.)

I was trying to figure out your proposal for a design of a solar system, without a battery (and without the grid). You seemed to be implying that my system was designed poorly, "to power the utility electric grid, not your house". And that my technology was primitive, as my "AC inverter and the associated controller is too slow to adjust the voltage". And that I was wrong about needing a sink for all the generated power: "The power doesn't have to go anywhere."

So that's the design I'm interested in hearing about from you. A solar system with no grid, and no battery, with an inverter that is "fast" enough, and with the solar system generating significantly more power than the house uses. That's what you seemed to be implying, so I'd like to hear how your design works.

Meanwhile, this seems essentially self-contradictory: "The issue isn't that you need to absorb all of the generated power. If you have 20 kW panels and are only consuming 10 kW, you'd only need to absorb 10 kW from the panels." That seems to match, not contradict, my earlier claim that all the power has to go "somewhere". You either need to use it immediately, store it (e.g. in a battery), or send it to the grid. If the solar panels are generating 20kW, then that all has to be sent somewhere (or else they need to completely shut off).

Otherwise, all of your comments -- which had a tone of "correcting" me -- instead seem to just be perfectly compatible with my very original comment on this thread. I already said that my panels shut off when the grid goes down, and I already said that if I add a battery, then I can still use my panels even without a grid. And I already said that I don't even need a large battery. So it seems that, with my original comment, I already said essentially everything you are saying now.

If you don't have a non-grid, non-battery, solution, then what exactly did you add to the conversation?

Publius said...

Solar Panels Don't Detonate

@Don:
>If you don't have a non-grid, non-battery, solution, then what exactly did you add to the conversation?

What I am correcting you on is this:

>You either need to use it immediately, store it (e.g. in a battery), or send it to the grid. If the solar panels are generating 20kW, then that all has to be sent somewhere (or else they need to completely shut off).

You don't need to use all the power generated, and it doesn't all need to be sent somewhere. You can draw from the panel just what you need and no more. The solar modules have an IV curve, and the load attached will determine where on the IV curve the circuit is operating. If the panel is unconnected, the voltage rises to it's open-circuit voltage, Voc, which for the panel in the link, is 39.7 V. On the other end of the IV curve, if the panel is shorted, the voltage will be 0 V, but the panel will supply 9.61 A of current (Isc).

So the reason your panels disconnect when the grid goes down isn't because the power has to be "sent somewhere." It doesn't. Otherwise, a disconnected panel would detonate. The reason is that your controller circuit can't maintain stable 120 V 60 Hz operation to your home without a buffer (either battery or the grid). This isn't to say that there is a controller that could do that -- there isn't.

Don Geddis said...

@Publius: OK, so this reads like a contradiction to me: "You don't need to use all the power generated, and it doesn't all need to be sent somewhere. You can draw from the panel just what you need and no more. ... your controller circuit can't maintain stable 120 V 60 Hz operation to your home without a buffer (either battery or the grid). This isn't to say that there is a controller that could do that -- there isn't."

So, if there is no controller that offer you a little bit of power when the panels are generating a lot ... then how exactly can you "draw from the panel just what you need and no more"?

You keep claiming that something can be done, but you don't seem to have offered any design for how to do it.

Publius said...

Let me clarify:

So the reason your panels disconnect when the grid goes down isn't because the power has to be "sent somewhere." It doesn't. Otherwise, a disconnected panel would detonate. The reason is that your controller circuit can't maintain stable 120 V 60 Hz operation to your home without a buffer (either battery or the grid). This isn't to say that there is a controller that can operate without a buffer -- there isn't.

Consider a standalone solar power installation that isn't tied to the grid. The system has batteries as the buffer. Now consider that the batteries are fully charged, the sun is shining, and no one is at home. The only load is single clock radio, drawing 45 mW of power. The panels are capable of producing 20 kW of power. No power is flowing into the batteries, as they are fully charged. Only 45 mW of power is being drawn from the solar array to power the clock radio. The power drawn from the panel is the intersection of the load with the IV curve of the panel array.

The system you have wasn't designed to power your house, it was designed to power the grid. When the grid goes down, your system has to be isolated from the grid for safety reasons -- if you were still supplying power to the grid, it could injure utility workers. Your system doesn't disconnect because the "power has to go somewhere" (it doesn't), it disconnects for safety. It can't stay on to power your home, as the controller can't adequately control the voltage to 120 VAC at 60 Hz without a battery buffer. It's not a power flow issue, it's a circuit design / performance issue.


Ron said...

To clarify:

> This isn't to say that there is a controller that can operate without a buffer -- there isn't.

It's not that it is *impossible* to design such a controller, it's just that no one has done this and put it on the market. It could be done, but no one has. And the reason no one has is that very few people would want it. People want to be able to power their house even when the sun is not shining, so a battery kills two requirement birds with one engineering stone.

Don Geddis said...

@Publius: "Otherwise, a disconnected panel would detonate."

No, it's pretty obvious that you can just leave a circuit open, and no electricity flows. Obviously the case we're concerned about is when we want some electricity, but the panels are generating far too much.

"The only load is single clock radio ... Only 45 mW of power is being drawn from the solar array to power the clock radio."

I don't think that's true. In the scenario you describe, as far as I understand the system, the solar panels would be disconnected at that point, and the clock radio would be powered from the battery.

I don't think it's possible for the panels to be generating 20kW, and for the system to only extract 45mW and just ignore the rest. This is exactly the situation I was originally hoping for; my understanding is that it doesn't happen.

Instead, the solar panels shut off, until the battery has drained sufficiently that it is capable of absorbing the full 20kW from the panels again, in order to recharge.

"The system you have wasn't designed to power your house, it was designed to power the grid."

You keep using this very odd, almost derogatory phrasing. As far as I can tell, according to your logic, no solar system on earth is "designed to power your house" (at least, without a buffer). If all you mean is that mine didn't come with a buffer, then that was pretty obvious from the very beginning. But besides that, nothing "else" about the solar design "was designed to power the grid" but "wasn't designed to power" my house.

"Your system doesn't disconnect because the "power has to go somewhere" (it doesn't) ... the controller can't adequately control the voltage ... without a battery buffer"

And just why can't the controller control the voltage without a buffer? It's pretty clear that if there's a place for the excess power to go, that it can work just fine. You are rejecting my explanation that the excess power "has to go somewhere". But you have yet to offer any alternative explanation for why, in the absence of a buffer, the controller is unable to provide the desired power.

@Ron: "It's not that it is *impossible* to design such a controller, it's just that no one has done this"

That's an interesting claim, thanks.

"And the reason no one has is that very few people would want it."

I'm not actually convinced by your market explanation. I think most home solar installations don't have battery buffers. If it were trivial to add the feature of still using solar power even when the grid is out, I suspect that would be a standard feature on regular home installations. The fact that it isn't, leads me to suspect that such a design (even if physically possible) must be very complex or expensive, and so perhaps not worth the minor benefit in most cases.

Ron said...

> The fact that it isn't, leads me to suspect that such a design (even if physically possible) must be very complex or expensive, and so perhaps not worth the minor benefit in most cases.

Yes, that's exactly right.

Peter Donis said...

@Don:
If the solar panels are generating 20kW

There is a misconception lurking here which might be contributing to misunderstanding in this discussion. Solar panels don't "generate" power in the sense that a conventional generator does. They work differently, and the difference affects system design.

Suppose you had a DC load that you wanted to power directly from a solar panel. Say it's an LED light that runs on 5 V DC and consumes 1 W of power when it's on and at full brightness. You hook it up as the load in a straight DC circuit in which the solar panel plays the role of the battery. What happens when the sun shines on the panel?

The solar panel works by the photoelectric effect: photons of sufficient energy can knock electrons out of atoms, and those electrons can then flow freely through the material and produce a current. For simplicity, we'll assume that all of the incoming photons are of just the right energy to knock the electrons out of the atoms and give them enough extra energy for an effective voltage of 5 V, so the voltage is exactly matched with that of the LED panel.

The question now is how much current flows in the circuit. 1 W of power at 5 V is 0.2 A (note that I'm not claiming these numbers are realistic, I'm just picking simple ones for illustration). That amount of current translates to some number N of electrons per second. And the light shining on the photoelectric cell will knock out some number L of electrons per second from atoms. So basically we have two possibilities:

(1) L <= N. In this case, an amount of current corresponding to L electrons per second will flow in the circuit, and the brightness of the light will depend on the ratio L / N. In the limit L = N, the light will be at full brightness.

(2) L > N. In this case, the light will be at full brightness and 0.2 A of current will flow in the circuit, regardless of how large L gets.

The obvious question now is, in case #2, what happens to the extra electrons? And the answer is, the same thing that would happen if the cell were just sitting there in the sun not connected to any electrical circuit at all: the electrons end up being recaptured by atoms and their extra energy gets converted to heat, so the cell heats up.

In other words, I think Publius is actually right that "extra power" from a solar cell does not need to go anywhere. It just goes into heating up the cell. As long as the cell design allows for appropriate heat transfer to the environment, so that under bright sun conditions at no load (i.e., cell disconnected so all the energy of electrons knocked out of atoms by the light gets converted to heat) the cell's temperature will stabilize at some acceptable value, there is never an issue with the circuit drawing less current than the equivalent of the number of electrons the light shining on the cell is knocking free. The excess electrons just get recaptured and heat up the cell, but less than the cell would be heated up if it were disconnected under the same light conditions.

More in a follow-up post.

Peter Donis said...

Follow-up to my previous post:

The problem with the nice, simple setup I described in my previous post is that our homes don't run on DC power. They run on AC power. So the DC power that comes from a solar array has to be converted to AC power in order to power your home.

(Note that this raises one obvious option for solar cells, since many of the loads in your home are actually DC loads: have a separate set of DC wiring in your home for those loads, which a solar array can just power directly. You could have a battery pack in the middle, but it might well be easier just to have every such DC device have its own--some, like laptops and smartphones, already do. Even TVs nowadays are mostly DC devices, since they are just big LED panels or something similar. But this is a much more ambitious redesign of a typical home than most people are up for, which is probably why I haven't seen a solution like this advertised.)

Converting DC power to AC power is a chore in itself, because the AC power has to be of a certain voltage, frequency, and phase to match what is currently supplied to your home and what all of the devices plugged into your home's electrical outlets expect. (Note that devices which are actually DC, like LED TVs, laptops, and smartphones, are taking that AC power and converting it back to DC again. If this looks inefficient to you when you already have a solar panel producing DC power on your roof, see my previous parenthetical comment above.) And if your home is connected to the grid, and there is no buffer between your solar panels and everything else, your solar panels are now extra sources on the grid, which greatly complicates the control problem that the grid's operators have to deal with (to the point that they basically have gotten governments to put regulations in place that prevent it from happening).

Putting a buffer between the panels and everything else, and shutting off the panels if for some reason that buffer goes away, removes that extra complexity of grid control, because to the grid, your solar panels now are no longer a source; all they do is reduce the average load your house draws from the grid. And since, as Ron points out, most people want electricity when the sun isn't shining, they need a buffer (a battery pack or something with equivalent function) anyway, so it isn't worth trying to design a system without one (and then have to try to get it accepted by grid operators who don't want the extra complexity in their controls).

Don Geddis said...

@Peter Donis: Thanks for the additional examples, but I still have a few questions. Going in reverse order...

Grid control should not be more complex: I'm happy to disconnect my house from a non-working grid. I just want my solar panels to keep supplying my house, even when the grid is down. From the grid's perspective, it should look completely identical; solar panels only send power to the grid when the grid is up. So "grid complexity" doesn't seem like a reason not to do it.

Wiring DC in the house: great idea! Not quite yet ... but maybe soon. I'm starting to see USB chargers available more regularly, including in power bar extension cords, and some house wall outlets. The idea of whole-house parallel electric networks hasn't quite caught on yet, but it may be coming. I'm pretty sure that Google did something like that for its data centers: All the PCs have power supplies which convert AC to DC. Instead of that, they started wiring up data centers with DC, and then got to remove the power supplies from each of the rack-mounted servers. Clever!

Re: your LED example. Surely solar panels are not producing power at 5V. But maybe that part is easy to change. I did find this post, which indeed seems to suggest offering the house a direct DC feed via a voltage regulator, before going through the battery and DC->AC inverter. Clever! So maybe that would just work, to have straight DC (except that houses aren't wired to distribute that, today).

I still don't fully understand why the inverter can't run without a buffer available. You and Ron both say "people don't want it", but I'm not convinced. Ron does also say that it would be expensive and complex, but it isn't clear to me why a voltage regulator is trivial, but an inverter without a buffer is not.

Ron said...

> it isn't clear to me why a voltage regulator is trivial, but an inverter without a buffer is not

Because a voltage regulator and an inverter are completely different beasts.

A voltage regulator is (or at least can be) nothing more than a big honking transistor and a simple feedback circuit that controls the transistor to keep the output voltage constant. You will also need a big honking heat sink attached to your big honking transistor. But other than that, it's pretty simple. Also, DC loads are almost invariably low-voltage and low-power so there aren't very many bad things that can happen even if you get it wrong. You're not going to injure yourself with 24VDC, and you're not going to burn your house down with a 100W load.

Turing DC into AC is a completely different matter, and turning low-voltage DC into high-voltage AC is a completely different matter yet again. On top of that, AC loads are typically high-power, and there are inductive loads. So now all kinds of bad things can happen if you get it wrong. You can burn your house down. You can destroy expensive appliances. You can kill yourself and others. All those details add up to a very challenging engineering problem, and by the time you've solved them all, adding a battery is a very small additional effort for a very large payoff.

Ron said...

Also...

> You will also need a big honking heat sink attached to your big honking transistor.

Even if you leave off the heat sink, the worst that will happen is that you will fry your transistor. And transistors, even big fat honking ones, are pretty freakin' cheap nowadays.

Don Geddis said...

Sorry to keep harping on this, but I feel like I'm still missing something.

Yes, I totally get that "turning low-voltage DC into high-voltage AC is a completely different matter yet again". You are absolutely right that the engineering (and safety!) problem is much more difficult, than just transforming DC voltage.

But I already have the DC->AC inverters. (My solar system has an Enphase microinverter on each panel, so my system as a whole is already automatically generating AC power.)

My question is more specific: why do inverters require a buffer? What is it about converting DC to AC makes it so difficult, if you don't have a buffer available to send the unused power? That's apparently (according to you, Peter, and Publius) not a problem in a pure DC system. Yet once we convert to AC, it becomes a major problem. Why?

Ron said...

> What is it about converting DC to AC makes it so difficult, if you don't have a buffer available to send the unused power? That's apparently (according to you, Peter, and Publius) not a problem in a pure DC system. Yet once we convert to AC, it becomes a major problem.

No, YOU are the one who originally wrote:

> The panels seem to need to have a place to dump all their unused excess power;

That's wrong, and Publius tried to explain why that was wrong. Unfortunately, some of his phraseology was a bit misleading. (Publius is quite skilled that way.)

You CAN plug a solar panel directly into an inverter and it will work -- until you plug in a refrigerator and the motor fires up and generates a huge inductive load. Then all kind of Bad Things can happen. Or maybe you'll get lucky and they won't. The only way to know for sure is to try it.

But if you want to design an inverter that will serve that use case and sell it on the open market then you have to make sure it complies with building codes, and then you can't just wing it. You have to make sure that you've dealt with all the possible contingencies. That is the hard part, and that is the thing for which there is no market because it would not be cost-competitive with a system that can also run at night.

Peter Donis said...

Grid control should not be more complex

I'm not sure what "should" has to do with it. A grid in which there are only a very small number of sources, whose output is controlled by the grid operators, is much simpler to control than a grid with a very large number of sources, most of which have output not controlled by the grid operators. Basically, in the first type of grid, the grid operator can assume that the load is negative everywhere except a small number of controlled points. In the second type of grid, the grid operator must be prepared for the load to become negative at any of a very large number of points, at unexpected times. The latter is a much more difficult control problem.

From the grid's perspective, it should look completely identical; solar panels only send power to the grid when the grid is up.

Any source only sends power to the grid when the grid is up. That's obvious.

But a source that can send power to the grid at times not expected by the grid operator is much more difficult to manage from the grid operator's point of view. Particularly if there are a very large number of them. That's not a problem for you, but it *is* a problem for the grid operator, and it is why, as I said, solar installations are generally not allowed by regulation unless there is a battery pack or similar buffer in between them and the grid.

Surely solar panels are not producing power at 5V.

I explicitly said I was only choosing the numbers for illustration and was not claiming that they are accurate. The exact voltage does not matter; the fact that "excess power" (i.e., more electrons are being knocked out of atoms by incoming solar photons than are needed to supply the current for the load) does not need to "go anywhere" (it just heats up the panel) is the key point.

I still don't fully understand why the inverter can't run without a buffer available.

It can. The laws of physics don't prevent it. But, as noted, regulations do (because grid operators got governments to put them in place).

Peter Donis said...

But, as noted, regulations do (because grid operators got governments to put them in place).

Note: This is a *different* set of regulations than the ones Ron is talking about, the ones in building codes, etc., that are there to prevent Bad Things From Happening when a large AC load like a refrigerator or an air conditioner starts up.

Don Geddis said...

@Ron: "it would not be cost-competitive with a system that can also run at night."

You seem to be suggesting that the batteries are essentially "free", and the bulk of the cost is in a "safe" inverter.

I'm not sure I believe that. My solar system already has inverters, and already (with the grid) powers my house (or sells back to the grid). That hardware has already been installed.

Now, I understand that with a battery pack, there needs to be another inverter to generate (something like) 110V/20A AC, from the DC batteries. You think this is an enormously complex device, that overwhelms the cost of the batteries? The Tesla Powerwall certainly seems to be charging for kWh of battery capacity. And yet you think that is fake, and the real cost is in the inverter?

@Peter: "Any source only sends power to the grid when the grid is up. That's obvious."

No, that's not true ... at least, not by "physics". The power company can shut the grid down, and home solar could still power the nearby local lines. This is dangerous for the power company repair crews, so building codes require home solar to disconnect from the grid, when the grid itself has gone down.

But that's an active process, requiring a device that is installed and inspected. It doesn't "have" to happen. It's only a good idea. It is required by law, not by physics.

"as I said, solar installations are generally not allowed by regulation unless there is a battery pack or similar buffer in between them and the grid"

No, that's not correct. I -- and almost everyone with home solar that I know -- have solar panels that send power to the grid ("net metering") ... but have no battery (or any other) buffer. This claim is just wrong.

So the grid design (at least in my local area) already needs to account for "the load to become negative at any of a very large number of points, at unexpected times". That is part of the local power grid design already.

That's not what we're talking about. We're only talking about the unusual case, when the grid goes down, and the solar panels disconnect from the grid (as required by regulations). In that case only, I'm curious why my house can't have an inverter that still powers the house from the solar panels -- without a buffer. Yet it seems not a thing that anyone has.

But buffers are not needed to interact with a live grid. That is definitely not part of the story.

Ron said...

> You seem to be suggesting that the batteries are essentially "free", and the bulk of the cost is in a "safe" inverter.

No. I'm suggesting that someone ran the numbers and determined that there isn't a business case for it. I don't know the details, but if someone put a gun to my head and forced me to speculate it would be because the addressable market of people who are willing to pay money for a backup power system that doesn't work after sundown is too small to recoup the NRE expenses. But I don't really know.

It's possible that the bean counters got this wrong, and there's a huge untapped market out there ripe for the picking. If you believe that, the proper response is to start a company that sells this product, use the proceeds to buy a case of Cohibas, bring one of them over to my place, light it with a $100 bill, and blow the smoke in my face while laughing manically. Or something like that.

Don Geddis said...

OK, Ron, I understand your challenge. It doesn't work for me, but let me explain:

We both observe the same thing about the world: solar panels generate power in the afternoon; when the grid is down, the house remains unpowered. It seems like a puzzle: why not just use the solar panel power that's right there?

Neither of us know the real answer to that question.

You look at the physics / engineering, conclude that you believe it should be possible to use that solar power, and therefore infer -- but without any direct evidence -- that there must be economic reasons why such a feature is not widely deployed.

I look at the economic situation, conclude that if it were ("easily") possible to allow the use of solar panel power during a blackout, then surely the free market would have already provided this solution. So then I infer (without direct evidence) that there must be a physics / engineering reason (which I'm not currently aware of) that prevents this from being offered as a solution.

I proposed what I thought might be the physics / engineering problem: that the power all has to "go somewhere". You guys tell me I'm wrong. Could be. Maybe I have it completely backwards. Perhaps the physics / engineering problem is not one of "excess power", but instead one of insufficient power. What if your panels are only generating 2kW (morning, late evening), but your house is trying to draw 5kW? In "usual" circumstances, the electrical system will "fill in" the remainder of the demand, either from the grid, or from the battery backup system.

But if you have no grid, and no battery ... what actually happens to the computers, TVs, clocks, refrigerator, clothes dryer, etc ... when there is 5kW demand but only 2kW power from solar? Maybe that's not an acceptable, stable, electrical scenario for consumer power.

But I'm just speculating here. You guys clearly know the electrical physics better than I do. I'll just say that I don't at all buy your economic explanation about why this feature is not readily available, and remain convinced that there is surely some physics / engineering flaw in the idea. Even if I don't know what it is yet.

Peter Donis said...

@Don:
I -- and almost everyone with home solar that I know -- have solar panels that send power to the grid ("net metering") ... but have no battery (or any other) buffer

Hm, interesting. Obviously my information about regulations, and on the amount of investment grid operators have made in being able to manage intermittent sources, is way out of date.

Ron said...

> Perhaps the physics / engineering problem is not one of "excess power", but instead one of insufficient power. What if your panels are only generating 2kW (morning, late evening), but your house is trying to draw 5kW? In "usual" circumstances, the electrical system will "fill in" the remainder of the demand, either from the grid, or from the battery backup system.

Yes, that is exactly right.

> But if you have no grid, and no battery ... what actually happens to the computers, TVs, clocks, refrigerator, clothes dryer, etc ... when there is 5kW demand but only 2kW power from solar?

You get a brownout. And that can cause damage.