[paulddraper]

> the whole of the US would only require the whole Texas Panhandle to be covered in panels 25% efficient

Are we building Dyson spheres next?

I believe largish orders of mass-market solar panels (currently ~15% efficient, as you mentioned) are about $10/sq ft, but let's make that $5 for sake of argument. So $218k for one acre.

The Texas Panhandle is 16.6 million acres. So assuming solar becomes nearly twice as efficient for half the cost, the solar panels alone for this venture would be $3.6 trillion. I hesitate to guess what construction, installation, batteries, or infrastructure for 16 million acres would add.

---

So I concede that given best-case economics, impractical funding, and technology that doesn't exist, it could perhaps be done.

More of a XKCD "What If?" scenario than a serious solution.

[sinxoveretothex]

3.6 trillions is a lot of money if you intend to bulk buy all of it at once today, but comparatively less over, say, 10 years. US GDP is around 18 trillion/year, 2% of the GDP is a lot, but it's certainly realizable.

Besides, it is somewhat meaningless to look at the price of solar panels without a reference. As I stated above, I have no idea how to estimate the cost of nuclear. I also don't know if the new kind of reactor (like traveling wave reactors) are anything more than a concept (it seems like breeder reactors are pretty much at the level of prototypes today).

I think the important point is what the price difference is rather than whether a one-time purchase of solar panels would cost an amount to large to fit in one's wallet.

[paulddraper]

$3.6 trillion is only a portion of just the materials cost. It's like the price of lumber, as compared to the total cost of building a house.

Elon chose that particular spot because the economic of solar is geography dependent. The infrastructure costs of distributing 100% solar will be astronomical.

I think you could pat yourself on the back if you figured out how to make it work for only 20x the cost of just the panels.

---

> As I stated above, I have no idea how to estimate the cost of nuclear

There is extensive data about nuclear costs, both estimated and empirical.

In 2010, the U.S. Energy Information Administration estimated total captial costs for nuclear power at $5.4k per kW. http://www.eia.gov/oiaf/beck_plantcosts/index.html (This is a rather safe estimate -- the EIA reported inflation-adjusted costs for actual, real-life plants build in the 1960s was $1.5k per kW. But I'll use the higher estimate.)

The U.S. used 3.9 trillion kWh last year, twenty percent of which is already nuclear. So that's $1.9 trillion of capital for 100% of U.S. electricity to come from nuclear.

---

At this point, coal makes the most economic sense of course. And the U.S. has enough for the next couple centuries.

[sinxoveretothex]

> In 2010, the U.S. Energy Information Administration estimated total captial costs for nuclear power at $5.4k per kW.

Matches what I found here: http://www.eia.gov/forecasts/capitalcost/xls/table1.xls (5.530$/kW) (page with full report here: http://www.eia.gov/forecasts/capitalcost/, release date 2013)

Thing is that it is literally more expensive than photovoltaic at around 4k$/kW (same source). Even the Operation and Maintenance is 5 times as expensive (~93$ vs ~26$).

I must say that I did not expect the evidence to be in my favour here (as evidenced by my initial comment).

> Elon chose that particular spot because the economic of solar is geography dependent.

Yes, the Sun hits more at lower latitude: http://energy.gov/maps/solar-energy-potential

> The infrastructure costs of distributing 100% solar will be astronomical.

> I think you could pat yourself on the back if you figured out how to make it work for only 20x the cost of just the panels.

What's nice about solar is that it can be very localized (with the efficiency loss described above).

I agree that the costs of distributing all the power required for the whole of the US from the Panhandle would require… interesting, distribution architecture.

Going by the solar potential map, I guesstimate the Panhandle average to be about 525 Wh/ft^2/day and the worst parts of Washington and Oregon to be at about 350 Wh/ft^2/day. So let's assume the national average to be 440 and so assume that we need about 120% the panels needed before. Actually, let's bump that up to 140% just because my number is rather rough and the population of the US is mostly on the coasts.

Now, what do you expect the other costs to be? Probably batteries. Going by somewhat sketchy graph on the Internet, I'll approximate the energy use as constant throughout the day, with a 2X peak offset just after the sun sets. So let's assume that one third of the energy needed through the day doesn't need to be stored at all, it's used as it's generated.

Going by: http://www.eia.gov/beta/MER/index.cfm?tbl=T02.01#/?f=A&start... the US used about 100000 trillion BTU => 100E31E12 BTU 0.293071 Wh/BTU = 29 quadrillion Wh AKA 30 trillion kWh (10 times your figure for some reason).

So 30E12 kWh * 2/3 * 1/365 = 53E9 kWh of storage needed

Now, Tesla sells the Powerpack for about 2.6 M$/4MWh. 53E9 kWh * 2.6 M$/4E3 kWh = 34.5E12 $ AKA 34 trillion $.

I don't know what the economies of scale would look like on producing 530 million powerpacks, but I'm pretty sure it would make the cost substantially lower (besides, those numbers are not for bulk orders).

As for panels, going by your numbers, I get 10$/ft^2 = 10$ * (3.3 ft/m)^2 = 109 $/m^2. Assuming 15% efficient panels and assuming 420 Wh/ft^2/day * 0.15 = 686 Wh/m^2/day. So 109 $/m^2 / .686 kWh/m^2 = 159 $/kWh

30E12 kWh * 159 $/kWh = 4.77E15 AKA 4.77 quadrillion $.

30E12 kWh / 36524 h 4000$/kW = 13 trillion $, so something is off.

Indeed, your 10$/ft^2 is way off, it is for small residential installations: http://solar-power-now.com/solar-panel-cost-per-square-foot/

Now, I don't know what to think of the ~350-fold price reduction for really large installation, but it seems sensible going by electronics bulk pricing.

Assuming a more conservative 10-fold reduction for the price of batteries, we'd get 13+3.4 trillion $ = 16.4 trillion $.

Going by the EIA report, nuclear would be about 5/4 the cost/kW of solar (presumably without storage), which is about on par with the price of storage + solar: (13+3.4)/13 ≈ 5/4

> The U.S. used 3.9 trillion kWh last year, twenty percent of which is already nuclear. So that's $1.9 trillion of capital for 100% of U.S. electricity to come from nuclear.

Yes, but they may be at the end of their life: http://www.eia.gov/tools/faqs/faq.cfm?id=228&t=21

---

At any rate, I don't have much of an issue with nuclear, at least the newer breeder designs since breeders produce very little waste and the new designs are presumably very safe. But obviously, all costs being equal (assuming no error in my calculations) I'm 100% in favour of renewables over nuclear. Renewables don't have the international concerns linked with exporting nuclear reactor technology for one and that's a serious drawback of nuclear honestly.

[paulddraper]

Thanks for the additional info.

[Dylan16807]

That price is not unreasonable. The US consumes roughly 5 trillion kWh every year. At a low-end estimate of 10 cents per kWh we have a 15 trillion dollar budget to build a facility that lasts 30 years. If we prioritize the environment over cheap power, we could double or triple that budget (and if people use less power in response to higher prices, that makes our job even easier). Obviously it would take a while to ramp up production, but that's not hard, just slow.