2 ms·

Yeah, at scale this kind of coordination is really, really difficult. With only 10 houses it's pretty easy.

manicdee 12y ago
At scale it becomes easier, since with 10 houses each house represents 0.1 of maximum load, and with 100 houses each house represents 0.01 of maximum load, so there are more opportunities for any particular device to draw power from the grid since there will be more other devices ready to turn off at that time. Upgrading to city-scale means that the individual consumer will be less likely to have to wait for their electricity ration. The hard part is defining the system for such a "smart grid" to safely coordinate activities while being resistant to tampering, spoofing, or other cyber warfare. As an example, a smart grid will be able to deal with "time of use" charging, meaning that solar electricity produced during the early morning or late afternoon will be far more expensive than what is produced at noon/zenith. A malicious party might attempt to fake the production of electricity in order to get third parties to "buy" electricity that was never produced. This will lead to brown-outs as there is more demand than generation, while the fraudster "selling" the fake electricity will be quite hard to catch. Even worse, a fraudulent network company might decide to under-meter your production so they can make a profit from the extra electricity that you produced that they aren't paying you for. Starting at the "four neighbours build a DC network" level, how do you determine which of your three neighbours is not producing electricity when they claim they are?
- msandford 12y ago
  I guess what I meant by "difficult" is from the algorithmic side of things. Let's suppose that scheduling demand is NP hard. For N of 6 this is trivial. For N of 100,000 this is a nightmare, so you end up not with ideal demand scheduling but a bunch of heuristics and approximations. I can think of a bunch of ways to handle the honesty problem. If it was just say 6 houses I'd probably wire the whole thing as a star and put a current measurement device on each of the 6 hot wires that all connect at the central location. It's then very easy to tell where power is coming from and going to. Each person could also install their own current measurement device where the power attaches to their in-house system and monitor internally and compare so that way if the central unit gets compromised any individual knows it and can stop participating until it's fixed.
BrandonMarc 12y ago
Rather than lots of difficult coordination, why not go a simpler route and make it based on the last digit or two of the street address? It's not a perfect plan, and some smoothing out may be needed to account for the distribution of addresses in a city / state / region, but the plus side is it can be done immediately and without need to figure out what others are up to. Or maybe not the street address ... use the lat-lon, or phone #, etc.

At scale it becomes easier, since with 10 houses each house represents 0.1 of maximum load, and with 100 houses each house represents 0.01 of maximum load, so there are more opportunities for any particular device to draw power from the grid since there will be more other devices ready to turn off at that time. Upgrading to city-scale means that the individual consumer will be less likely to have to wait for their electricity ration. The hard part is defining the system for such a "smart grid" to safely coordinate activities while being resistant to tampering, spoofing, or other cyber warfare. As an example, a smart grid will be able to deal with "time of use" charging, meaning that solar electricity produced during the early morning or late afternoon will be far more expensive than what is produced at noon/zenith. A malicious party might attempt to fake the production of electricity in order to get third parties to "buy" electricity that was never produced. This will lead to brown-outs as there is more demand than generation, while the fraudster "selling" the fake electricity will be quite hard to catch. Even worse, a fraudulent network company might decide to under-meter your production so they can make a profit from the extra electricity that you produced that they aren't paying you for. Starting at the "four neighbours build a DC network" level, how do you determine which of your three neighbours is not producing electricity when they claim they are?

msandford 12y ago
I guess what I meant by "difficult" is from the algorithmic side of things. Let's suppose that scheduling demand is NP hard. For N of 6 this is trivial. For N of 100,000 this is a nightmare, so you end up not with ideal demand scheduling but a bunch of heuristics and approximations. I can think of a bunch of ways to handle the honesty problem. If it was just say 6 houses I'd probably wire the whole thing as a star and put a current measurement device on each of the 6 hot wires that all connect at the central location. It's then very easy to tell where power is coming from and going to. Each person could also install their own current measurement device where the power attaches to their in-house system and monitor internally and compare so that way if the central unit gets compromised any individual knows it and can stop participating until it's fixed.

BrandonMarc 12y ago

Rather than lots of difficult coordination, why not go a simpler route and make it based on the last digit or two of the street address? It's not a perfect plan, and some smoothing out may be needed to account for the distribution of addresses in a city / state / region, but the plus side is it can be done immediately and without need to figure out what others are up to. Or maybe not the street address ... use the lat-lon, or phone #, etc.