4 ms·

". I have no idea how much a multi-gigawatt HVDC line costs per kilometre," Well there's the crux, ain't it? They're bloody expensive. Also, to your 3.5% loss

". I have no idea how much a multi-gigawatt HVDC line costs per kilometre," Well there's the crux, ain't it? They're bloody expensive. Also, to your 3.5% loss / 1000km line loss add in the transformation losses. Then consider that a 20k line must be a DC line (to avoid reflection losses). Also, why did you use 20k km? Such a curious number, half the world's circumference, that I suspect you mean to propose to wrap the world in power lines to get power at midnight from noon elsewhere Well 20km for your loss calculations is not nearly enough since you cannot connect two sites in a straight line. For starters the earth' topography won't allow it. But then there's the issue that you won't connect every city to every generating site, but have a hub-spoke model (like ethernet). To give you an idea of the added distance compare the straight line distance of Atlanta to Pittsburgh to the interstate distance (520 mi to 680 mi). Finally, the proposal of connecting two sites on opposite ends of the earth is fundamentally hegemonist. What are you going to do with the fiercely independent people that tend to live in deserts when they oppose you wallpapering their desert with panels? When the Bedouins of Algeria are blowing up the electricity needed in Japan, will the EU send storm troopers to punish the locals?

ben_w 5y ago
> Also, to your 3.5% loss / 1000km line loss add in the transformation losses. Which is going to be a fixed amount regardless of distance? While I don’t have the numbers, I would expect HVDC to be a dead end already if it made a difference to my conclusion. > 20k km? Such a curious number, half the world's circumference, that I suspect you mean to propose to wrap the world in power lines to get power at midnight from noon elsewhere Indeed, though this is likely excessive: summer/winter is significantly harder for storage to deal with — and hence the mostly likely to be solved with transmission — than day/night, given most people sleep at night but nobody hibernates though winter. > But then there's the issue that you won't connect every city to every generating site, but have a hub-spoke model (like ethernet) Surely a web is the best topology? Existing grids don’t all run on hub-and-spoke models: https://openinframap.org/#2/19.82/13.02/L,P,S https://openinframap.org/#2/19.82/13.02/L,P,S > When the Bedouins of Algeria are blowing up the electricity needed in Japan, will the EU send storm troopers to punish the locals? While true, I’m not even going to try to play at global geopolitics; all I can say is that while deserts are a good place for PV, and while PV is the best thing I (being neither a biologist nor a geologist) know what to do with a dessert, the land area required for PV is low enough that even the UK — entirely north of the US-Canadian border — would be able to power itself in winter with no transmission and just a day/night storage system using 1% of its land area [0], so tiling the entire Sahara is unlikely: The same equations say that even if we electrify all world energy consumption, Algeria alone could supply everyone 6 times over, but you’d only do a global grid like this precisely if you didn’t want the huge storage system I’m implicitly adding to worsen the performance of my opinion and give your criticism the best possible chance. [0] illumination = (cos(51 north + plus axial tilt) / π) * 1kW/m^2 ~= 87 W/m^2; efficiency = 20% ~> 17.5 W/m^2; 35 GW / (17.5 W/m^2) ~= 2000 km^2; U.K. land area ~= 242,495 km^2. I think this is a bad idea, but economics beats aesthetics and someone would need to run the numbers that I don’t have to say which of storage and transmission is actually better. Perhaps the minima for the UK is PV in Algeria and HVDC, perhaps it is local x*PV + y*wind + z* batteries. It probably varies from any one country to the next.
- sasfasaf 5y ago
  "Which is going to be a fixed amount regardless of distance" Sure, but knock off another 10% "I would expect HVDC to be a dead end already if it made a difference to my conclusion." I didn't get that. Could you re-phrase? "Surely a web is the best topology? " My point you're underestimating the line length since you cannot connect generation to consumption with a straight line. Also, the hub-spoke or a web network would need the connections to be much larger than needed for that city (since the city would also need to transmit to it's neighbors) "While true, I’m not even going to try to play at global geopolitics; [..] worsen the performance of my opinion and give your criticism the best possible chance." You cannot dismiss cultural concerns. Look what happened in Kabul last week!
  - ben_w 5y ago
    > Sure, but knock off another 10% > "I would expect HVDC to be a dead end already if it made a difference to my conclusion." > I didn't get that. Could you re-phrase? If the conversion losses were as high as 10%, why did anyone bother using HVDC for e.g. a mere 85 km stretch in the Trans Bay Cable? But even if they were 10%, that’s a multiplier not an adder so it makes a difference of 10% to final energy prices, it wouldn’t break anything (though it might shift the minima to batteries, as those are also cheap and getting cheaper). > My point you're underestimating the line length since you cannot connect generation to consumption with a straight line. Also, the hub-spoke or a web network would need the connections to be much larger than needed for that city (since the city would also need to transmit to it's neighbors) I think you’re overestimating how much the losses matter. You only lose half your power from 20,000 km, adding an extra 10,000 km from wiggling around a bit takes you down to 34% of original generation, and even then webs usually have high-capacity connections between important locations and lower-capacity links from the high capacity nodes to the low use nodes. Is anywhere in the US more than about 2400 km from the coast? Because 2400 km is a multiplicative loss of 0.918 compared to whatever losses you get from a trans-oceanic link to wherever (0.77 from Algeria to DC, according to distance from Wolfram Alpha, likewise 0.66 from Queensland to San Francisco). And the USA already has (several) grids, so you’d probably just want to link any global HVDC connection like this into each grid, rather than to each city. > You cannot dismiss cultural concerns. Look what happened in Kabul last week! “I am not qualified to discuss them” absolutely is not the same thing as dismissing them. But my main point is more of we don’t need to tile the Sahara (or any single particular place), so it shouldn’t even come up in the first place.

ben_w 5y ago

> Also, to your 3.5% loss / 1000km line loss add in the transformation losses. Which is going to be a fixed amount regardless of distance? While I don’t have the numbers, I would expect HVDC to be a dead end already if it made a difference to my conclusion. > 20k km? Such a curious number, half the world's circumference, that I suspect you mean to propose to wrap the world in power lines to get power at midnight from noon elsewhere Indeed, though this is likely excessive: summer/winter is significantly harder for storage to deal with — and hence the mostly likely to be solved with transmission — than day/night, given most people sleep at night but nobody hibernates though winter. > But then there's the issue that you won't connect every city to every generating site, but have a hub-spoke model (like ethernet) Surely a web is the best topology? Existing grids don’t all run on hub-and-spoke models: https://openinframap.org/#2/19.82/13.02/L,P,S https://openinframap.org/#2/19.82/13.02/L,P,S > When the Bedouins of Algeria are blowing up the electricity needed in Japan, will the EU send storm troopers to punish the locals? While true, I’m not even going to try to play at global geopolitics; all I can say is that while deserts are a good place for PV, and while PV is the best thing I (being neither a biologist nor a geologist) know what to do with a dessert, the land area required for PV is low enough that even the UK — entirely north of the US-Canadian border — would be able to power itself in winter with no transmission and just a day/night storage system using 1% of its land area [0], so tiling the entire Sahara is unlikely: The same equations say that even if we electrify all world energy consumption, Algeria alone could supply everyone 6 times over, but you’d only do a global grid like this precisely if you didn’t want the huge storage system I’m implicitly adding to worsen the performance of my opinion and give your criticism the best possible chance. [0] illumination = (cos(51 north + plus axial tilt) / π) * 1kW/m^2 ~= 87 W/m^2; efficiency = 20% ~> 17.5 W/m^2; 35 GW / (17.5 W/m^2) ~= 2000 km^2; U.K. land area ~= 242,495 km^2. I think this is a bad idea, but economics beats aesthetics and someone would need to run the numbers that I don’t have to say which of storage and transmission is actually better. Perhaps the minima for the UK is PV in Algeria and HVDC, perhaps it is local x*PV + y*wind + z* batteries. It probably varies from any one country to the next.

sasfasaf 5y ago
"Which is going to be a fixed amount regardless of distance" Sure, but knock off another 10% "I would expect HVDC to be a dead end already if it made a difference to my conclusion." I didn't get that. Could you re-phrase? "Surely a web is the best topology? " My point you're underestimating the line length since you cannot connect generation to consumption with a straight line. Also, the hub-spoke or a web network would need the connections to be much larger than needed for that city (since the city would also need to transmit to it's neighbors) "While true, I’m not even going to try to play at global geopolitics; [..] worsen the performance of my opinion and give your criticism the best possible chance." You cannot dismiss cultural concerns. Look what happened in Kabul last week!
- ben_w 5y ago
  > Sure, but knock off another 10% > "I would expect HVDC to be a dead end already if it made a difference to my conclusion." > I didn't get that. Could you re-phrase? If the conversion losses were as high as 10%, why did anyone bother using HVDC for e.g. a mere 85 km stretch in the Trans Bay Cable? But even if they were 10%, that’s a multiplier not an adder so it makes a difference of 10% to final energy prices, it wouldn’t break anything (though it might shift the minima to batteries, as those are also cheap and getting cheaper). > My point you're underestimating the line length since you cannot connect generation to consumption with a straight line. Also, the hub-spoke or a web network would need the connections to be much larger than needed for that city (since the city would also need to transmit to it's neighbors) I think you’re overestimating how much the losses matter. You only lose half your power from 20,000 km, adding an extra 10,000 km from wiggling around a bit takes you down to 34% of original generation, and even then webs usually have high-capacity connections between important locations and lower-capacity links from the high capacity nodes to the low use nodes. Is anywhere in the US more than about 2400 km from the coast? Because 2400 km is a multiplicative loss of 0.918 compared to whatever losses you get from a trans-oceanic link to wherever (0.77 from Algeria to DC, according to distance from Wolfram Alpha, likewise 0.66 from Queensland to San Francisco). And the USA already has (several) grids, so you’d probably just want to link any global HVDC connection like this into each grid, rather than to each city. > You cannot dismiss cultural concerns. Look what happened in Kabul last week! “I am not qualified to discuss them” absolutely is not the same thing as dismissing them. But my main point is more of we don’t need to tile the Sahara (or any single particular place), so it shouldn’t even come up in the first place.