Hacker News Reader

Comment by vegetablepotpie | Hacker News Reader

This is unlikely. While transmission lines may go DC, all of the distribution, the lines that goes from a substation to peoples houses, in the US is AC. Although it’s possible to wire a house for DC, and people have done that, many of the appliances we use, use AC power.

Although AC phase matching is a delicate technical problem, it’s one we’ve solved for over a hundred years. DC presents other engineering challenges that are non-trivial. For example, circuit breakers for AC power are designed to “break” when the AC curve hits zero volts. This eliminates the chance of arcing and makes breakers smaller and cheaper to manufacture. A DC breaker has a chance of arcing and it may be necessary to make them larger, or use exotic gasses with high dielectric values to prevent this from occurring. Either of these increase costs for homeowners.

View on HN

Transmission has a bunch of problems with DC as well. Transformers are fundamentally AC devices; if you want to use them in a DC circuit, you need an inverter anyway to convert the DC to AC and back again. There are ways to step up DC voltage, but none are as cheap or reliable at utility scale as a transformer is. If you don't step up the voltage, you'll lose basically all your power to transmission losses, since delivering high power at low voltage requires high current, and power loss increases with the square of the current.

High-voltage DC is also extremely dangerous, as it's prone to arcing and electrocution.

> you need an inverter anyway to convert the DC to AC and back again

But you have complete freedom to choose their frequency, so you can use much cheaper transformers.

> A DC breaker has a chance of arcing and it may be necessary to make them larger, or use exotic gasses with high dielectric values to prevent this from occurring.

That's a problem for mechanical switches (were conductors move to make contact or disconnect).

If you use semiconductors to do the switching, it becomes a problem of how fast they switch, how much energy is dissipated during the switch, and how much energy those semiconductors can absorb momentarily (thermal mass).

For small equipment, this is a solved problem. Fast switching FETs are cheap & robust.

For utility-scale, semiconductors are an entirely different ballgame. Big advances have been made over the last decades.

So a HVDC grid might in theory be possible. But in practice, it'll be an engineering tradeoff between HVDC+semiconductors almost everywhere vs. HVAC+more traditional gear like transformers.

And even if a HVDC grid were practical with modern tech, in most places there's existing AC-based grid & power plants. I suspect the "sync AC phases" is an easier problem to solve than "re-do the grid to use HVDC".

But for 'simple' point-to-point connections like an offshore windpark or long international lines, HVDC is sometimes practical (and used, if so).

I think a DC grid is likely. But it's still 100 years away.

Over time, more and more components will be built DC (DC long distance cables are already popular, due to being slightly cheaper. DC for electronics is popular due to AC being poorly suited to microprocessors/logic. DC sees wide use in cars. USB-C brings computer peripherals into the DC world).

Eventually, whenever two DC bits of power infrastructure are touching oneanother, someone will notice that removing the DC->AC->DC conversion steps saves money and increases efficiency.

Eventually enough bits of the grid will be DC that AC 'islanding' can occur - whenever every link from A to B is DC, there is nothing to keep the phase locked between place A and place B. Initially that will be solved with software locking means.

But finally maintaining that anti-islanding tech will be too costly, and all remaining bits of the AC grid will be removed.

But it's gonna take 100 years because grid tech changes slowly, and infrastructure like buried cables can be 70+ years years old.

The problem is that the DC from the grid is going to be way higher voltage than the voltage to the house. There are lots of AC-AC transformers that would need to be replaced to use DC for distribution and hard to switch incrementally without a dual converter in every house. Also, AC is better for medium distance transmission, DC can boosted to super high voltage for long distance but that isn't practical inside a city.

Then, the house DC voltage is going to be higher than the electronics DC. So you'll need to have a box at every single outlet to convert DC-DC. The appliances are going to need the higher voltage house DC. And the house DC voltage is going to be more dangerous than AC. Also, there are no standards or even proposals for DC electrical system: no voltage and no outlets.

The problem is that replacing DC-AC-DC with DC-DC-DC and there isn't much savings from all those conversions. Would you replace all of your appliances for 1% savings in electrical cost?

And those reasons are why it'll take 100+ years...

However, the AC-AC transformers currently use a lot of Steel+copper. That's expensive. New developments will be pushed towards solid state alternatives which are theoretically cheaper (and exist today, but aren't widely used).

Outlets in your house I suspect will get replaced with super-USB-C. Ie. something which is 5 volts and then negotiates a higher voltage as needed. A future version I bet will support 3 kilowatts for hair dryers, etc. That will be safer. It'll also be pushed by device makers who currently hate the headache of making different versions of electrical devices for every country with different plugs. Fancy houses already have USB outlets in every socket. Cheapo devices like flashlights already use USB power input for worldwide universality.

I could imagine rules might push people to super-USB-C too. Laying AC lines requires highly qualified labour, but plugging in super-USB-C cables into a super-USB-hub can be done by anyone - the safety is in the design, rather than requiring careful installation.

When every outlet in your house is super-USB-C, it won't take much for newly built houses to instead use DC everywhere (maybe even negotiated voltages too - ie. your house only receives 5 volts until any device needs more power, and then it'll ramp up).

> However, the AC-AC transformers currently use a lot of Steel+copper. That's expensive. New developments will be pushed towards solid state alternatives which are theoretically cheaper (and exist today, but aren't widely used).

In what world is steel and copper more expensive than semiconductors?

A small CPU in 1974 cost $360. A functionally similar CPU today costs $0.03.

Whereas a pound of copper went from $0.89 to $3.91 in the same time.

Transformers of equal wattage cost more today than they did in 1974. Power electronics cost substantially less.

I don't see this trend reversing.

It is impossible for super-USB-C to carry that much power. The new 240V is at the limit of what USB-C can do and you are talking about order of magnitude difference.

First, the wire thickness needs to be like regular wires to carry enough current. Changing the wiring in the walls is the hardest thing. Second, the voltage needs to be like regular service to carry enough power over regular sized wires. Higher voltages, like 500V, are better since DC loses more energy over distance than AC. Third, the plug needs to be similar size to power plugs to not arc, and DC arcs worse than AC.

Is replacing everything worth the effort to increase efficiency by a little bit? You are optimizing for low-power DC devices at the expense of high-power AC appliances. The only way I can see DC power happening is in isolated community like the Moon or Mars.

> It is impossible for super-USB-C to carry that much power.

5 amps, 2000 volts. 10 kilowatts. Made safe by milliamp precision leakage detection (you won't manage to kill anything with a milliamp of current, and more than a milliamp of current will trigger the protection).

2000 volt insulation made of PVC needs to only be 0.1mm thick (say 0.5mm for a 5x safety factor). That can easily fit in plugs and sockets of the existing size. The leakage detection would also detect current leaked between the power conductors, so dielectric breakdown can be protected against.

The cable would be designed to deliberately cause dielectric breakdown when heated before catching fire, so that in cases of a fatigued conductor or wire draped across a stovetop, a fire is impossible.

This design is only safe if implemented correctly, so each end of the connection will frequently test the other end of the connection to ensure it is adhering to the spec. It would do this by sending an encrypted message saying "In 37 seconds I am about to cause what looks to you like a baby chewing through the insulation". Then, in 37 seconds, that happens, the power gets turned off, and a few microseconds later the encryption key is send over proving that it was a test and the power is turned back on again. Both sides would have capacitors so the user wouldn't notice the power going off for a few microseconds per minute.

There may be a potential for more localized grids that are interconnected by HVDC transmission lines. I'm not sure what the impetus for that would be, however. I agree that it would be unlikely.

Interestingly, HVDC actually becomes a more efficient method of transmission over longer distances. Perhaps it's feasible to generate electricity half a continent away. Maybe tile the Sahara with solar panels and power all of Africa with it.

Naw, we’ll just take the long path and send the electricity to a richer continent:

https://en.m.wikipedia.org/wiki/Xlinks_Morocco%E2%80%93UK_Po...

You're assuming that, in the future, we'll still have huge, centralized plants. With solar, it's very possible that we'll have small stations in each town or area, and only need to balance between them infrequently, greatly reducing the amount of energy we need to carry across large distances.

> You're assuming that, in the future, we'll still have huge, centralized plants.

You're assuming that we won't.

> With solar, it's very possible that ...

How nice the world would be we could all lean on "well because it is possible it must eventually happen."

- Things will be the same, except for X.

- Maybe other things will be different too.

- Oh, so everything will be different?!

> You're assuming that we won't.

That's really not how I read that comment.

Inertia will keep the status quo for a very very very long time. Combine that with heterogeneity (i.e. not everyone will be using local solar plants) and that the distribution concerns noted still apply to local plants outside your house and it doesn’t seem like it’ll go away without some kind of central planning forcing the issue.

I think municipality-sized microgrids are a big part of the future, but they still don't remove the requirement for a grid, simply because of weather. Most renewables are very dependent on local weather conditions: you don't get solar when it's cloudy, and you don't get wind when it's calm. The grid needs to equalize power generation and consumption, and it's probably more economical to have a few transmission lines running between cities and to remote power generation facilities than it is add the utility-scale batteries needed to power through a week of cloudy weather.

I could however see a future where cities refuse to subsidize rural homeowners and communities, disconnect from the country-level grids that exist today, let them de-energize and fall into disrepair, and then maintain only a few transmission links over major transportation corridors to connect with other major cities.

Sure, but when you only need to transmit the difference, and even that only to charge batteries, the transmission lines can be much smaller.

That's not true, and stems from mistakenly thinking of electricity as a fluid instead of working through the math of the laws of electrodynamics themselves.

With AC, there is no net current anyway - nothing physical is being transmitted any substantive difference. The actual electrons travel on the order of 0.2 microns per 60Hz cycle, and then move back the other way. [1]

In reality, in the absence of fundamental electric components like resistors/capacitors/transformers, all points of a circuit have the same voltage and same current. The transmission lines are just connecting different cities into the same circuit, there's nothing flowing between them. This allows your solar array in the Mojave Desert to power your data center on the Columbia River, but there aren't fewer electrons traveling between them just because you also have a hydro power plant on the Dalles. The load from all devices on the grid is shared across all generation sources.

[1] https://en.wikipedia.org/wiki/Drift_velocity

You're oversimplifying. US residential power is currently supplied by a pair of 120 V wires, but most of the stuff in your house only uses one leg of that pair. That's alright because it's AC, and the ground wire doesn't actually need to carry any significant current. Power just returns time-lagged through the same wire when the voltage changes.

If you switch to DC, that doesn't work any more. Every amp in requires an amp out. The wiring in your house just got a lot more complicated, not to mention the wiring in the local grid.

Also, running your neighborhood on 85 V (the DC equivalent to 120 V) isn't exactly efficient. Even large ground-mounted transformers only provide power to 10-15 houses at most, and pole-mounted transformers may only service one house. The main power to your neighborhood is 7.2 kV because it's more efficient to to send power at high voltage and low current.

It's not impossible for a residential solar setup to output thousands of volts, but it's not easy and it's pretty constraining for designs. There's also a world of difference between that voltage at the street and that voltage in the house. Things go wrong in the house.

> You're oversimplifying. US residential power is currently supplied by a pair of 120 V wires, but most of the stuff in your house only uses one leg of that pair. That's alright because it's AC, and the ground wire doesn't actually need to carry any significant current. Power just returns time-lagged through the same wire when the voltage changes.

This isn't how it works.

Maybe my model of how AC works is wrong, but my understanding is that neutral wires exist and are necessary in addition to safety earth, even though neutral and ground are eventually tied.

> and only need to balance between them infrequently

I think it's exactly the opposite. There will frequently be a need to balance them.

With wind and solar in the mix, generation will fluctuate with the weather. In a given area, it could be cloudy one day and sunny the next. Or windy one day but not the next. And consumption won't be correlated with that, so that creates an extra source of mismatches between demand and consumption within each area.

Transmission is one way to solve that. You could also solve it with storage (within every area), but that's probably less efficient and/or more expensive.

It's easier for me to bring you some food when you're running out than to bring you food every single meal.

Sure, but you need the same road to get there regardless of how many hours a week it gets used.

It's a trade off. More transmission lines means less overbuilding and less storage. Really it's an optimization problem that will be solved on a case by case basis given geography and so on.

For infrastructure it’s the opposite. You have to buy a food truck big enough to feed everyone on the worst day of the year. And at that point you may as well use it every day because you already paid for it.

The arcing is no joke. In my building they have a new assembly of three coupled water pumps and an electrical box that is waiting to be installed that has a scary warning about how the electrical box could create a dangerous arc.

Reply on news.ycombinator.com