4 ms·

Trick is not the total amount of insolation, but how much solar is available in the worst periods of year. Germany has about 100 GW of solar installed; or worst

by anovikov 11mo ago

Trick is not the total amount of insolation, but how much solar is available in the worst periods of year. Germany has about 100 GW of solar installed; or worst weeks it produces about 280 GWh in solar power... we can average over a week because we may assume 2 hours' storage (so for 100 GW nominal, 200 GWh) and that's about 4-5 worst days in a row. Now electricity consumption is about 10TWh a week. Which means, 35x of the current amount of solar is required to cover it entirely if the solar has to supply everything even in the period when there is the lowest amount of sunlight is available (because storage over even longer periods is unrealistic). Which is for the case of "electricity only, in current consumption volumes", means 104,000 km2 or over 1/4 German land surface, with the average solar park density of 1 MW per 7.3 acres. If you try for entire amount of all energy consumed in Germany, not just electricity, then you need to fill entire German territory and that won't be enough. In particular (for your figure of 342.2 GW average), 610,000km2 which is 1.5x+ German surface area. Because 280GWh over a week averages to 1.67 GW average or 1/60 of nominal installed power. Yes, ON AVERAGE, solar panels give about 1/9 to 1/10 of their nominal installed power over the year, but it's 6-7x less during worst weeks. This number can be slightly improved by installing panels over the angle that optimises performance in winter (in summer there will be more than enough anyway), but no more than by 1.5x - still entire Germany will not be enough to supply itself with power from solar that works year round. So your calculation is wrong in this, and also in the fact that solar park is not fully packed with panels because they need pathways to service them and they need to not shade one another. Average area per 1 MW installed power is 7.3 acres in the USA.

ben_w 11mo ago
> 280 GWh > we can average over a week because we may assume 2 hours' storage (so for 100 GW nominal, 200 GWh) and that's about 4-5 worst days in a row […] because storage over even longer periods is unrealistic I don't see any particular logic to either part of this. Better question for "realistic" storage is how much do people actually want to spend, how do the costs change as quantities scale up from hours to days to weeks to seasons. (4kW * 24h * €50-€70/kWh) / 10 year battery life from daily cycles is an unsurprising part of a monthly energy bill (especially as this would basically substitute most of the grid's role in the hypothetical where Germany for whatever reason wasn't still buying nuclear power from France etc.); pumped hydro is estimated around €100/kWh installed, but different maintenance cost and lifetime. > Now electricity consumption is about 10TWh a week. Which means, 35x 9.3, I think you're using old numbers. I find no search results for 280 GWh claim; I will assume the number itself is correct, but given the rate of growth of PV it is of material relevance which year this happened in, e.g. if that 280 GWh was in was in 2020 it's more like 18x, if it was in 2015 it's 13x. > with the average solar park density of 1 MW per 7.3 acres You're double-counting here. 33.9 W/m^2 with 22% cells means the capacity factor is 15% of the name-plate capacity. This is a standard value in much of the world, varies from about 10%-20%, you can't do better than about 32% from anything fixed to the ground even on a completely clear sky on the equator. The map I linked to already accounts for all of that stuff, it's *internal to* how they reach the ~1000kWh/m^2/year for the worst bits of Germany: https://upload.wikimedia.org/wikipedia/commons/1/19/Germany_GHI_Solar-resource-map_GlobalSolarAtlas_World-Bank-Esmap-Solargis.png https://upload.wikimedia.org/wikipedia/commons/1/19/Germany_... Anyway, point is 1 MW/7.3 acres *already* accounts for "winter is bad". > This number can be slightly improved by installing panels over the angle that optimises performance in winter Opposite error! Not when you're talking about covering an entire country it doesn't. Panels would shade each other. > and also in the fact that solar park is not fully packed with panels because they need pathways to service them and they need to not shade one another And again, now you're double-counting. Precisely because 1 MW/7.3 acres number is already accounting for gaps, you've mixed up the optimal output per m^2 of *panel* with the optimal output per m^2 of *land*, and it's land that matters on this scale, the optimal number per unit of land is what you get from laying the panels flat on the ground.
- anovikov 11mo ago
  No, 1MW/7.3 acres is about INSTALLED CAPACITY, that is, peak power. https://solarindustrymag.com/nrel-report-highlights-land-use-requirements-for-solar-plants https://solarindustrymag.com/nrel-report-highlights-land-use... >For direct land use requirements, the capacity-weighted average is 7.3 acres per MW, with 40% of power plants within 6 and 8 acres per MW. And no, 280 GWh was literally a week ago: https://energy-charts.info/charts/energy_pie/chart.htm?l=en&c=DE&interval=week&year=2025&week=48 https://energy-charts.info/charts/energy_pie/chart.htm?l=en&... So we need to consume 10 TWH a week (consumption particularly for week 48, same one), and we produce 280 GWh with 100+ GW of installed capacity. Which means, we need 35x if only power is concerned, and 205x if all energy is concerned, not just electricity.
  - ben_w 11mo ago
    > No, 1MW/7.3 acres is about INSTALLED CAPACITY, One thing I find useful is doing the maths to see if an answer is plausible. Yours wasn't, and lo and behold I found this pretty quickly, directly debunking your citation: Those relying on the earlier benchmarks published nearly a decade ago are, thus, significantly overstating the land requirements of utility-scale PV. - https://www.energy.gov/sites/default/files/2022-01/lbnl_ieee-land-requirements-for-utility-scale-pv.pdf https://www.energy.gov/sites/default/files/2022-01/lbnl_ieee... And even with the changes since then, they waste a lot of land they don't strictly need to waste; see all the gaps in the image of Copper Mountain in Fig. 2. of that PDF. The reason for this difference is historical: PV modules themselves used to be the expensive part of a solar farm, not the land itself, so the early ones wasted a lot of land to put a small number of PV modules in the best possible position. The correct orientation today is "nobody cares" because they're so cheap (the only industrial scale thing I've found cheaper per square meter than PV is a very thin layer of cement, and even then I have to discard labour costs, but that's not important). This is also why quite a few of the Chinese ones are now artistic with their patterns, e.g. this: https://news.cgtn.com/news/334d6a4e7a557a6333566d54/index.html https://news.cgtn.com/news/334d6a4e7a557a6333566d54/index.ht... (One of the startups whose pitches I heard over the summer was a place that realised the cheapest in a lot of cases was just lay the panels flat). > And no, 280 GWh was literally a week ago TY

ben_w 11mo ago

> 280 GWh > we can average over a week because we may assume 2 hours' storage (so for 100 GW nominal, 200 GWh) and that's about 4-5 worst days in a row […] because storage over even longer periods is unrealistic I don't see any particular logic to either part of this. Better question for "realistic" storage is how much do people actually want to spend, how do the costs change as quantities scale up from hours to days to weeks to seasons. (4kW * 24h * €50-€70/kWh) / 10 year battery life from daily cycles is an unsurprising part of a monthly energy bill (especially as this would basically substitute most of the grid's role in the hypothetical where Germany for whatever reason wasn't still buying nuclear power from France etc.); pumped hydro is estimated around €100/kWh installed, but different maintenance cost and lifetime. > Now electricity consumption is about 10TWh a week. Which means, 35x 9.3, I think you're using old numbers. I find no search results for 280 GWh claim; I will assume the number itself is correct, but given the rate of growth of PV it is of material relevance which year this happened in, e.g. if that 280 GWh was in was in 2020 it's more like 18x, if it was in 2015 it's 13x. > with the average solar park density of 1 MW per 7.3 acres You're double-counting here. 33.9 W/m^2 with 22% cells means the capacity factor is 15% of the name-plate capacity. This is a standard value in much of the world, varies from about 10%-20%, you can't do better than about 32% from anything fixed to the ground even on a completely clear sky on the equator. The map I linked to already accounts for all of that stuff, it's *internal to* how they reach the ~1000kWh/m^2/year for the worst bits of Germany: https://upload.wikimedia.org/wikipedia/commons/1/19/Germany_GHI_Solar-resource-map_GlobalSolarAtlas_World-Bank-Esmap-Solargis.png https://upload.wikimedia.org/wikipedia/commons/1/19/Germany_... Anyway, point is 1 MW/7.3 acres *already* accounts for "winter is bad". > This number can be slightly improved by installing panels over the angle that optimises performance in winter Opposite error! Not when you're talking about covering an entire country it doesn't. Panels would shade each other. > and also in the fact that solar park is not fully packed with panels because they need pathways to service them and they need to not shade one another And again, now you're double-counting. Precisely because 1 MW/7.3 acres number is already accounting for gaps, you've mixed up the optimal output per m^2 of *panel* with the optimal output per m^2 of *land*, and it's land that matters on this scale, the optimal number per unit of land is what you get from laying the panels flat on the ground.

anovikov 11mo ago
No, 1MW/7.3 acres is about INSTALLED CAPACITY, that is, peak power. https://solarindustrymag.com/nrel-report-highlights-land-use-requirements-for-solar-plants https://solarindustrymag.com/nrel-report-highlights-land-use... >For direct land use requirements, the capacity-weighted average is 7.3 acres per MW, with 40% of power plants within 6 and 8 acres per MW. And no, 280 GWh was literally a week ago: https://energy-charts.info/charts/energy_pie/chart.htm?l=en&c=DE&interval=week&year=2025&week=48 https://energy-charts.info/charts/energy_pie/chart.htm?l=en&... So we need to consume 10 TWH a week (consumption particularly for week 48, same one), and we produce 280 GWh with 100+ GW of installed capacity. Which means, we need 35x if only power is concerned, and 205x if all energy is concerned, not just electricity.
- ben_w 11mo ago
  > No, 1MW/7.3 acres is about INSTALLED CAPACITY, One thing I find useful is doing the maths to see if an answer is plausible. Yours wasn't, and lo and behold I found this pretty quickly, directly debunking your citation: Those relying on the earlier benchmarks published nearly a decade ago are, thus, significantly overstating the land requirements of utility-scale PV. - https://www.energy.gov/sites/default/files/2022-01/lbnl_ieee-land-requirements-for-utility-scale-pv.pdf https://www.energy.gov/sites/default/files/2022-01/lbnl_ieee... And even with the changes since then, they waste a lot of land they don't strictly need to waste; see all the gaps in the image of Copper Mountain in Fig. 2. of that PDF. The reason for this difference is historical: PV modules themselves used to be the expensive part of a solar farm, not the land itself, so the early ones wasted a lot of land to put a small number of PV modules in the best possible position. The correct orientation today is "nobody cares" because they're so cheap (the only industrial scale thing I've found cheaper per square meter than PV is a very thin layer of cement, and even then I have to discard labour costs, but that's not important). This is also why quite a few of the Chinese ones are now artistic with their patterns, e.g. this: https://news.cgtn.com/news/334d6a4e7a557a6333566d54/index.html https://news.cgtn.com/news/334d6a4e7a557a6333566d54/index.ht... (One of the startups whose pitches I heard over the summer was a place that realised the cheapest in a lot of cases was just lay the panels flat). > And no, 280 GWh was literally a week ago TY