Hacker News Reader

Comment by shadowgovt | Hacker News Reader

One question about the grid I've never wrapped my head around:

So the entire grid is in phase, by which I mean every generator, load, &c is operating on the same AC pulse (handwave-handwave ignoring smaller loads, transformations, etc.). But that phase isn't instantaneous; it's near-light, and the grid is long enough for that to matter in places.

So every point in the grid can't be perfectly in phase with every other point, right? Because we have both lightspeed delays and loops, so even if point A is receiving power from two substations in-phase, point B (with different lengths of wire to those two substations) should be receiving it out-of-phase, right? How do we balance that in the grid?

View on HN

Others answered your question. A related, but different problem is “how do you make sure a new power source you add to the grid is in sync with the grid the moment you connect it?”.

That’s described on https://en.wikipedia.org/wiki/Synchronization_(alternating_c....

Firstly, you design your generator to have the same wave form and phase sequence as the grid.

Then, you power up your generator and make sure the voltage, frequency and phase angle of the electricity it generates matches that of the grid.

The moment that’s (more or less) the case, you connect the grids, preferably on a zero crossing. That Wikipedia page describes a setup with incandescent bulbs that can be used to manually detect that moment, but nowadays, it’s done using electronics.

You just synchronize to wherever you are on the wave. Your offset will not be the same as your neighbors offset, but that's fine.

Maybe imagine a line of buoys in the ocean as a wave passes.

But if I have power from two sources coming into my facility, they wouldn't be in phase with each other, right? How do I use both?

(Perhaps the obvious answer is correct here: "Grids don't work that way; you'd never do that.")

The sum of two sine waves of differing phase is a sine wave with some phase between the two.

I wonder if the solution is similar to routing traces on PCBs where there's high-frequency stuff going on - you add and subtract length from the line pairs until the phases match up. Sometimes the dumb solutions are the best ones?

Weird how all the replies so far misunderstood your question.

There's a lot of high-frequency stuff going on, but the line length is so much that most of it irradiates away. You don't go adding more line to compensate anything, but you do add filters at some places.

An airport or hospital might be fed from two different nearby substations. The substations are probably fed from the same transmission lines with parallel circuits. The substations will have transformers with the same winding configuration for a consistent phase shift from transmission to distribution level voltages.

The airport or hospital will procure transformers that have the same winding configuration and impedance so the phase shift and voltage drop across them will also be the same. The low voltage windings of the transformers at the hospital can be paralleled so they both feed the load. Reverse power protection would be implemented so the hospital can’t back feed the distribution line.

What phase difference do you expect in this case? What effect would it have?

In most cases where you do have more than one source (say datacenter for redundancy) you don't just tie them together directly, in simplest config you'd just switchovered from one to another if one failed.

Imagine you've got two cables coming into your factory. One is 500 meters longer than the other. That would mean that the sine waves are 0.01% out of sync.

I'm pretty sure for home electronics the phase difference doesn't even matter that much. Everything sensitive seems to be converted to DC anyway.

Maybe it matters in large industrial applications, I have no idea. But I also imagine in their situation it's probably pretty straightforward to clean up the power supply.

Would be fun to see this turn up as a case study for two identical assembly lines driven by motors. One on the “fast” frequency and the other on the slow.

In my imagination, the specialists are brought in only to learn that the RPMs in one motor were juiced because someone didn’t practice cable management

The electrical phases would be the exact same frequency still, 60 hz.

In reality, the phase difference is always small, and it ends up not mattering.

But... If you had a worldwide electricity grid running at 60Hz, it would start to matter, and you would make sure to use local capacitors/inductors to make phase shifts to make sure that you didn't have big circulating currents in loops (they just are wasting energy)

Napkin math: Speed of light is 300,000 kilometers/second. At 60 hz, the wavelength would be 5,000 km. 1% of that would be 50 km.

Imagine a triangle with 2 generators 100km apart, and a factory almost due south of one generator:

ga ......... gb

__f

I think the offset would be something like sin(gb.f) - sin(ga.f)

But really this comment is bait for an EE to school me.

Here's a video of old school mercury arc rectifiers as payment: https://www.youtube.com/watch?v=YhaQqgXrMMU

you would have to be in a location to straddle synchronous grid. There may be some places that allow that. Otherwise, the two sources are already in approximately in phase anyways. To put it simply though, you put some devices on one feed and some devices on another feed. So it is a non-issue.

If you have local generation you want to combine with grid power you usually just sync your local generation to the grid. If you don't want to do that you can just use a DC intertie that is local. Basically you'd have two AC -> DC converters and a single DC -> AC converter.

So what happens if you've got three power plants in different states, connected in a triangle?

50Hz wave length is at ~6000 km so it frankly isn't a problem.

You'd have to have 2 different routes from same power source that differed in hundreds of kilometers in length and that just doesn't really happen. And small phase difference would just cause uneven loading

Also that power would be "used up" closer to the power source and you'd be sucking off power from closer sources.

> So every point in the grid can't be perfectly in phase with every other point, right? Because we have both lightspeed delays and loops, so even if point A is receiving power from two substations in-phase, point B (with different lengths of wire to those two substations) should be receiving it out-of-phase, right? How do we balance that in the grid?

As far as I know, in your example, one of the two circuits will be transmitting more power from the substations to point B than the other. There's a device called a phase-shifting transformer (https://en.wikipedia.org/wiki/Phase-shifting_transformer) which can be used to adjust the phase angle of the circuits, and that way, adjust how much of the power is carried by each circuit.

(The following sentence from that article probably goes to the core of the answer for your question: "For an alternating current transmission line, power flow through the line is proportional to the sine of the difference in the phase angle of the voltage between the transmitting end and the receiving end of the line.[1]")

I looked this up once, and the answer is: They basically just ignore it.

If everything is in a line, you just sync to your local phase, but if you have wires in a triangle that's impossible, and they basically just ignore it, in practice it doesn't happen often enough to cause any problems.

You can think of electrical phases like any other kind of wave: maybe like shaking a slinky. The power plant sends the phases downstream to consumers, and once it leaves the factory it doesn't know what consumers do with it (beyond detecting voltage drops, etc.).

So the grid doesn't need to balance it, no.

And from the receiving end's perspective, power will be exactly one phase out of sync if the cables differ in length by 5000 km. I'm assuming that isn't really a problem people have to worry about, but my background is in physics and not engineering (i.e. maybe there's some industrial applications that I'm not thinking of).

I think it's more like doing the wave at a large sports stadium.

Everyone isn't always in phase with every other station's phase at that exact moment in time. They're just in phase with their local part of the grid's phase. Though amount that is different is milliseconds, not seconds as in the stadium wave example.

This mostly came from reading this: https://electronics.stackexchange.com/a/291328

> So every point in the grid can't be perfectly in phase with every other point, right?

not in phase as such, but in sync.

because you only have to match your 60hz(or 50hz) to what you receive locally, so long as the waveform is reasonably stable its not that much of an issue.

Because everything is effectively a bunch of elastically linked pendulums, you tend to reach equilibrium, so long as the load doesn't act too much as a dampener

You could introduce a back to back HVDC link [1] on one of the two incoming legs, allowing you to alter frequency and phase freely.

[1] https://en.wikipedia.org/wiki/High-voltage_direct_current#Ba...

random guess: the entire grid is huge but the difference in lengths between those two substations is tiny and doesn't matter?

Reply on news.ycombinator.com