3 ms·

I actually did out the math, and per tonne of silica (quartz) vs tonne of uranium ore mined, silicon has about 5x as much energy as uranium in a breeder reactor

I actually did out the math, and per tonne of silica (quartz) vs tonne of uranium ore mined, silicon has about 5x as much energy as uranium in a breeder reactor. Compared to current reactors, the silicon metal in a solar panel will produce ~500x more energy per kg. Note that this doesn't count the silica used in the glass covering etc. Solar should be considered right up there next to nuclear in terms of energy density, at least as far as mining is concerned.

kaybe 9y ago
I think you could publish this as a short paper if you did it rigorously enough. And feed it to the media to boot.
philipkglass 9y ago
I've done this math too, and my results show that silicon is well short of uranium (though far above fossil fuels). Assumptions: - 20% capacity factor for solar - 25 year lifetime at nominal capacity factor (In practice, most panels work for more than 25 years, but also lose a bit of power output over time due to e.g. fogging of glass covers and cracks in cells from thermomechanical fatigue. I'm fudging the two factors together.) As of 2016 it takes 4.8 grams of silicon, on average, to manufacture one watt of silicon PV capacity: https://www.pv-tech.org/news/lower-silicon-consumption-due-to-wafer-innovations-to-hit-polysilicon-deman https://www.pv-tech.org/news/lower-silicon-consumption-due-t... That means that one kilogram of silicon made into PV can produce (1 / 4.8) * 24 * 365 * 25 * 0.2 = 9125 kWh of electricity over its lifetime. For comparison, one kilogram of natural gas will produce ~5.6 kWh of electricity when burned in a combined cycle gas turbine plant. One kilogram of uranium produces about 38300 kWh of electricity over its lifetime, using these numbers: http://www.world-nuclear.org/information-library/facts-and-figures/world-nuclear-power-reactors-and-uranium-requireme.aspx http://www.world-nuclear.org/information-library/facts-and-f... The lifetime electricity yield from a kilogram of uranium is about 4.2 times as much as that from a kilogram of silicon. EDIT: oh, my mistake! You were comparing ore volumes. Medium grade uranium ore contains less than 1% uranium, while quartzite is super-abundant and contains more than 45% silicon. That accounts for your ratios.
- hwillis 9y ago
  It also has to be enriched from the .72% natural concentration of U-235 and U-234. The SWU (energy) that goes into that is negligible, but the concentration process isn't perfect and you still need several times more natural uranium to produce something that can be used in a reactor (3-4%). I used a 33-40 year lifetime weighted by measured degradation rate[1]. Interestingly, NREL blames a lot of the capacity loss on the degradation of solder interconnections. I also used real insolation values in four regions of the US and 15%-23% (median to high efficiency) efficient cells, but assumed perfect MPPT tracking. I did guesstimate the amount of silicon though, by assuming 60-200 micron panels, which according to your link is off by a factor of 4-10x. So in the end real world solar loses to fusion (obviously) but beats light water reactors handily. It's initially a tossup when compared to breeder reactors, but only if you neglect silicon recycling (should that happen). At that point fission obviously loses. [1]: https://www.nrel.gov/pv/lifetime.html https://www.nrel.gov/pv/lifetime.html
  - philipkglass 9y ago
    The World Nuclear numbers use mined natural uranium on the input side, so enrichment losses are implicitly included for nuclear. There are a lot of different PV degradation mechanisms. Most of them have known engineering solutions. I personally think that some modules installed today will have an operational lifetime in excess of 50 years, but I'm conservative when it comes to picking defensible numbers for extrapolation. Did you account for cell-to-module energy losses and inverter loading ratios > 1 when you made your estimates? I chose 20% capacity factor, despite published national (US) EIA numbers in excess of 25% for utility scale projects, because EIA reports AC (post-inverter) numbers. With typical utility scale inverter loading ratios now at 1.3 or so, the module-level capacity factor is more like 20%. None of these issues change your major conclusions. I'm just going into details because I rarely find anyone else interested in the subject matter.

kaybe 9y ago

I think you could publish this as a short paper if you did it rigorously enough. And feed it to the media to boot.

philipkglass 9y ago

I've done this math too, and my results show that silicon is well short of uranium (though far above fossil fuels). Assumptions: - 20% capacity factor for solar - 25 year lifetime at nominal capacity factor (In practice, most panels work for more than 25 years, but also lose a bit of power output over time due to e.g. fogging of glass covers and cracks in cells from thermomechanical fatigue. I'm fudging the two factors together.) As of 2016 it takes 4.8 grams of silicon, on average, to manufacture one watt of silicon PV capacity: https://www.pv-tech.org/news/lower-silicon-consumption-due-to-wafer-innovations-to-hit-polysilicon-deman https://www.pv-tech.org/news/lower-silicon-consumption-due-t... That means that one kilogram of silicon made into PV can produce (1 / 4.8) * 24 * 365 * 25 * 0.2 = 9125 kWh of electricity over its lifetime. For comparison, one kilogram of natural gas will produce ~5.6 kWh of electricity when burned in a combined cycle gas turbine plant. One kilogram of uranium produces about 38300 kWh of electricity over its lifetime, using these numbers: http://www.world-nuclear.org/information-library/facts-and-figures/world-nuclear-power-reactors-and-uranium-requireme.aspx http://www.world-nuclear.org/information-library/facts-and-f... The lifetime electricity yield from a kilogram of uranium is about 4.2 times as much as that from a kilogram of silicon. EDIT: oh, my mistake! You were comparing ore volumes. Medium grade uranium ore contains less than 1% uranium, while quartzite is super-abundant and contains more than 45% silicon. That accounts for your ratios.

hwillis 9y ago
It also has to be enriched from the .72% natural concentration of U-235 and U-234. The SWU (energy) that goes into that is negligible, but the concentration process isn't perfect and you still need several times more natural uranium to produce something that can be used in a reactor (3-4%). I used a 33-40 year lifetime weighted by measured degradation rate[1]. Interestingly, NREL blames a lot of the capacity loss on the degradation of solder interconnections. I also used real insolation values in four regions of the US and 15%-23% (median to high efficiency) efficient cells, but assumed perfect MPPT tracking. I did guesstimate the amount of silicon though, by assuming 60-200 micron panels, which according to your link is off by a factor of 4-10x. So in the end real world solar loses to fusion (obviously) but beats light water reactors handily. It's initially a tossup when compared to breeder reactors, but only if you neglect silicon recycling (should that happen). At that point fission obviously loses. [1]: https://www.nrel.gov/pv/lifetime.html https://www.nrel.gov/pv/lifetime.html
- philipkglass 9y ago
  The World Nuclear numbers use mined natural uranium on the input side, so enrichment losses are implicitly included for nuclear. There are a lot of different PV degradation mechanisms. Most of them have known engineering solutions. I personally think that some modules installed today will have an operational lifetime in excess of 50 years, but I'm conservative when it comes to picking defensible numbers for extrapolation. Did you account for cell-to-module energy losses and inverter loading ratios > 1 when you made your estimates? I chose 20% capacity factor, despite published national (US) EIA numbers in excess of 25% for utility scale projects, because EIA reports AC (post-inverter) numbers. With typical utility scale inverter loading ratios now at 1.3 or so, the module-level capacity factor is more like 20%. None of these issues change your major conclusions. I'm just going into details because I rarely find anyone else interested in the subject matter.