Scientific American published a grand plan for nationwide terrestrial solar power generation system in the January issue. Its highlights:

• A massive switch ... to solar power plants could supply 69 percent of the U.S.’s electricity ... by 2050.


• A vast area of photovoltaic cells [and solar concentrator power plants] would have to be erected in the Southwest.


• Excess daytime energy would be stored as compressed air in underground caverns to be tapped during nighttime hours....


• A new direct-current power transmission backbone would deliver solar electricity across the country.


• But $420 billion in subsidies from 2011 to 2050 would be required to fund the infrastructure and make it cost-competitive.

It's an ambitious plan that could sharply decelerate CO₂ emissions and increase the US output of "green" power. Heroic plans require heroic proof. A critical analysis follows.

Some high level critiques are the following:

• Shifting peak load from day-time to night-time would not occur until solar displaced all natural gas plants and other swing units--i.e., all of excess air-conditioner demand over night-time demand, and all of the additional day-time usage that would occur as the price between day-time and night-time power usage equilibrated. This obviates the need for any wind-storage of solar power until well later.


• Compressed-air energy storage will become less useful as the price gap between day and night power diminishes. This undermines the case for near-term night to day storatge and will only be economical under this plan for day-to-night storage after day-time power is sufficiently cheaper to support the capital outlay. (Ironically since the solar installation is of the hockey-stick variety, compressed air storage may become viable for night-to-day energy storage well before solar becomes a relevant portion of energy supply.)


• Current photovoltaic production is about 2 GW of which US installation is about 8%. The plan calls for 84 GW of US installation by 2020 which would require 45% increases in solar installation every year for 13 years. Capping the installation at 10 GW/year installed, the ramp up becomes 70% per year 2006-2014.


• These growth rates are implausible without a $2.80/watt subsidy taking the installed price of $4/w to $1.20/w which is equivalent to $0.05/kwh. That would mean $234 billion in subsidies just to get to 3% of needed installed capacity by 2050.


• Polysilicon shortages are holding back photovoltaic growth so in 2007 and 2008 a growth rate of 20% is more plausible. That would require doubling production every year from 2009-2014 to hit the installed base of 84 GW by 2020.


• 84 GW by 2020 would be just 16% of average load and with a peak watt of electricity generating only 6 or 7 hours per day in the Southwest, it would be about 5% of total electric power generated.


• For this 5% of energy generated, we would be subsidizing it over 200% of the value of the energy generated--that is for $0.06 of electricity, it would require $0.14 in subsidies.


• At the end of the period, there is no guarantee that prices will be low enough to compete with coal, natural gas, nuclear or wind.


• If solar becomes viable and can compete with other energy types and begins to displace other types of power, prices for those types of power will drop. The total cost of solar will have to beat the marginal cost of coal or nuclear to dismantle an existing plant.

Consider investing in terrestrial solar power for security reasons or as a contingency, but it's a lot of faith to get the case to work for half of daily electricity demand.

It has never been clear to me whether the cost per watt refers to peak power production or to average power production. It appears that the cost quoted is per peak watt. It is cost per average watt that should be of interest, since, to produce a given amount of energy will require four to eight times the quantity of solar cells that a naive person would calculate based on peak power ratings. It is perplexing that articles in the popular press never make the distinction.

It has never been clear to me whether the cost per watt refers to peak power production or to average power production.

For solar, the usual measure is per peak watt, under some standard clear-sky condition (AM 1 or AM 1.5, I don't call which).

Napoleon Numbers or "horseback approximations": peak at AM1 is roughly 600 watts per square yard, 700 per square meter.

You can't keep a big enough array aimed at the Sun; if nothing else, adjacent segments shade one another. Also, the cells are less efficient when the incident radiation isn't normal to the surface. So actual power collected is proportional to the cosine squared, which integrated over a day is half the peak. So divide by two: 350 W/m^2.

Multiply by the cosine of the latitude. For the American Southwest, call it 32 degrees, cosine 0.84+. Result: whatever the figure they give you for power collected is, take 40% as a realistic estimate. Then multiply that by the efficiency, which for poly cells is generally around ten percent -- some go to twelve.

Thirty watts per square meter, net, over the day, average over a year. 250 WH/m^2 per day in the winter, when you need it most.

Yes, there are more efficient cells. They're crystalline (=5X to 10X cost) and require concentrators, which require aiming, which doubles the land area required and requires energy to build and run the aiming devices. Result is more or less push and pull.

And if you think PETA makes a big deal out of windmill bird strikes, think of what the environmentalists are going to say about an impenetrable roof over, say, Maricopa County, AZ.

Terrestrial solar collectors are wonderful for small-scale use; I have one on my barn. It charges a battery which runs a pair of RV fluoros (12V input) so I can do feeding and the like at night. It saved me a couple of grand because I didn't have to run power wires out there. Solar makes no sense whatever for large-scale use.

Regards,
Ric

It's viable, as long as you can't see the vast cell farms from Sen. Kennedy's house.

Scientific American is okay as long as you avoid the tacky leftist politics masquerading as "science".

Cost per watt is none of the above.

Current cost of solar panels is about $3.80 per watt when you buy a LOT of panels. This is the cost to buy a panel, usually a large panel of about 200 watts. The "watt" is defined under standard test conditions of 1000 watts/m2 (sea level) and 25 degrees C.

Panels output more power when they are colder and vice versa when hotter.

The SCI-Am article is stupid in that the storage method that they talk about is completely unworkable on a large scale.

Aren't their adiabatic losses whenever you compress a gas?

Large scale solar makes perfect sense...in space.

I also assume there'd be some losses in the direct current transmission or the conversion of DC to AC It would be ironic if this topic re-ignited some the AC/DC discussions of Edison and Tesla.

Typically solid critique from Sam and enlightening comments, especially Ric's (not to disparage anyone else). On the plus side, I would point out that peak power loads, caused by widespread use of air conditioning on hot summer afternoons, could be nicely addressed by photovoltaics. Presuming the installation costs to drop (and I expect we've all seen the under-$1/watt announcements/rumors by now) and modularity, such that starting with a few hundred watts and adding a bit at a time is easy -- I think the remaining constraint is regulatory: will it be 1) legal to do this and especially 2) possible to sell power into the grid? I can imagine a decentralized approach of adoption by individual homeowners having a significant effect at the margin. But Federally-mandated conversion/construction of huge arrays will turn into another ethanol subsidy scam pretty quickly; better to build more nuclear power plants.

The SCI-Am article is stupid in that the storage method that they talk about is completely unworkable on a large scale.

Which storage method was that, and why? They mention at least two.

I agree with Sam's comment that using storage to transfer baseload (read: nuclear) power from night to day seems more likely to occur before PV becomes cheap enough for the opposite to occur.

On the plus side, I would point out that peak power loads, caused by widespread use of air conditioning on hot summer afternoons, could be nicely addressed by photovoltaics.

They would be more cheaply addressed by thermal storage systems and time-sensitive electricity pricing. It would be cheaper, especially in large buildings, to make ice at night and use that ice to provide cooling during the day. This is already done where the pricing system allows it to be taken advantage of.

Ah. Thank you, Dennis. I suppose it makes perfect sense that advertising hype would be based on absolute best case conditions. What that means is that you can take two-thirds of the numbers I gave as your baseline -- unless you're prepared to site the arrays in some of the most remote, inaccessible areas of the country, where construction costs are maximized, proximity of the border is a significant worry, and environmentalists' shrieks about "destroying delicate ecosystems" reach 100 dBA and begin edging into the supersonic range. (After all, you're talking about leveling all the hills and covering the result with an impenetrable roof. I think that qualifies as "damage".)

Put compressed air out of your mind. The blasting to excavate the required caverns would require all the energy the panels would ever collect, leaving a zero net without ever figuring in the other manufacturing and construction costs. If anything this insane is ever done, the energy storage medium will be old-fashioned lead-acid batteries; it's the only thing the resources exist for in the quantities required. I once figured that it would only take a year's world output of lead to do it. If you think polysilicon production is a bottleneck, look up battery materials sometime. Which, of course, is why the insane compressed air notion. Lead is eeeeeeevil, and they have no concept of scale.

In fact, the whole thing smacks of a high school science fair project blown up by several orders of magnitude without any consideration whatever of scaling effects. I have to clean my panel about once a week. Who's gonna do that on an array of a hundred square kilometers? And how much bigger does it have to be if you figure that that's impossible?

Regards,
Ric

Well, if we put it next to the southwestern border, we can just hand out squeegees to the people fleeing Mexico, and put them to work right there. ;^)

Re: Paul's earlier comment, I would expect time-sensitive electricity pricing to be a major driver of all types of alternatives to peak-load power generation, including rooftop photovoltaics. They're certainly the only thing that (if cheap enough) would make any sense for me personally; I spend less than $1/day on electricity except during the summer months. The only significant electric bills I get are due entirely to air conditioning.

The blasting to excavate the required caverns would require all the energy the panels would ever collect, leaving a zero net without ever figuring in the other manufacturing and construction costs.

This argument appears to be independent of where the electricity is coming from, so you appear to be saying that CAES is energy negative regardless of the energy source. If this is the case, why are there successful CAES systems operating today?

Anyway, CAES can be made to work with aquifers, which require no excavation.

"Storage" of off-peak generating capacity has been going on since the early '70s at the Ludington Pumped Storage plant in Michigan.

re: energy storage

The compressed air storage probably wouldn't work economically except for existing caverns that can be made airtight (e.g., salt domes, existing natural gas wells, existing coal mines). Take an existing underground (or above ground) structure and coat its interior with something airtight and puncture resistant, then add air that costs $0.03/kwh to pump in, then drive your turbines with it when power costs $0.09/kwh. It's not economical now on a widespread basis because of losses and the installation cost per watt of the pump and turbine.

Re: peak watt

I think this is laboratory peak wattage directly facing the sun at azimuth on a clear day, etc. Even if the sun is up 12 hours or so on average, they only get 6 or 7 kwh pwer peak watt at the best sites in Arizona, New Mexico and California. That is, the combination of the clouds, cosine squareds (one for the latitude deviation, one for the time of day deviation) and everything else result in 25% of rated peak capacity (which would be higher in the brighter light of space without the atmosphere and Earth in the way more than 4% of the time). Note that the land area required is maybe twice as much as the light-collecting area or else we run into shading difficulties. The good news for solar advocates is that peak sunlight is during the day and the annual system peaks tend to come on sunny days during sunny seasons so that solar is worth the $0.10/kwh natural gas price instead of the blended $0.06 of coal or nuclear including $0.03 night time power. This implies the 5% of power generated by 84 GW of solar would be worth about 8% of the dollars people spend on electricity. If solar was 70% of peak capacity today (with corresponding reductions in other types of power), it would be only worth $0.03/kwh because the peak price would shift to non-sunny times.

re: reverse metering and distributed uptake

Many states and countries including California (and Germany which represented 55% of 2006 world wide installations) are heavily subsidizing solar by buying it back at retail rates, giving renewable energy certificates (or similar) subsidizing installation ($4/w in Austin--the only remaining cost is the ladders, nails and labor) and so on. Still, uptake is not substantial because the expense is high enough to attract only off-grid uses (like call boxes or the barn) or true believers.

The compressed air storage probably wouldn't work economically except for existing caverns that can be made airtight (e.g., salt domes, existing natural gas wells, existing coal mines).

I believe the Huntdorf and McIntosh CAES facilities use salt dome storage cav-ities that were specifically solution-mined for themselves, not repurposed from some other preexisting uses. Solution mining also does not require any blasting or excavation.

"I expect we've all seen the under-$1/watt announcements/rumors by now"

Nanosolar.com has started shipping $0.99/Watt solar panels:

http://www.nanosolar.com/blog3/2007/12/18/nanosolar-ships-first-panels/

Ed:
Read that press release carefully. it says:"- the world’s lowest-cost solar panel – which we believe will make us the first solar manufacturer capable of profitably selling solar panels at as little as $.99/Watt"

It doesn't say they're selling at that price, it just says that they could make a profit at that price. Panels in the kilowatt size range currently retail for about $4.50/watt. http://store.solar-electric.com/ I have bought from this vendor and their prices are in line with the industry. That price is finally dropping a bit (http://www.solarbuzz.com/) after several years of slow upward growth. Installed systems, where the homeowner doesn't provide free labor and mounting hardware but includes the grid-tie inverter, wire, switchbox, etc sell at about $9. per Watt. Grid-tie inverters alone sell at about $1/Watt in this size. At this price, and somewhat lower for bigger systems, homeowners can recoup the investment in about 8 years assuming both State and Federal rebates and no cost of money. That's offsetting consumer costs at the retail rate, about 4 times higher than the wholesale rate utilities deal with. This is why utilities buy from solar thermal system owners rather than PV; they're cheap enough to turn a profit at wholesale price + state mandated renewable incentives. http://news.zdnet.com/2100-9595_22-6166113.html

As for storage, (Snellenr mentioned above a specific site), there is currently about 18,000 MW of pumped storage in the US. Base power generated at night pumps water uphill, and runs generators in the daytime to meet peak needs. I don't know the measured cost competitiveness of water compared to compressed air storage, but the day/night wholesale price differential is sufficient to support these now.

Is it ... possible to sell power into the grid?

IIRC, not just possible, but manadatory that the utility accept the power, and pay you based on their highest marginal cost production plant.

On the other hand, solar is well-suited for distributed generation; as the US power grid continues to age and power demands continue to rise, avoidance of transmission costs also helps makes the case for solar vs. construction of new powerplants (requiring ever-more-expensive grid upgrades).

Polysilicon shortages are holding back photovoltaic growth so in 2007 and 2008 a growth rate of 20% is more plausible.

If poly-Si shortages are holding back PV manufacturing growth, they haven't been doing so very effectively. PV manufacturing has increased 48% per year since 2002, and went up 50% last year.

The issue with poly-Si is not that silicon is inherently rare, or that there's some inherent limit on the number of facilities we can make for purifying it, but rather than Si PV cells may be economically disadvantaged if they have to pay for the full cost of purifying the element, rather than piggybacking on the IC industry. But as the following link notes, PV became the dominant consumer of poly-Si in 2006, and this didn't slow the growth.

http://thefraserdomain.typepad.com/energy/2007/12/fyi-solar-cell.html

If poly-Si shortages are holding back PV manufacturing growth, they haven't been doing so very effectively.

Good point. US solar buyers can certainly outbid other countries and other competing uses for polysilicon (e.g., microchips) and grow quickly in the near term. But that would entail the price staying near $5/watt entailing another $1/watt in subsidies to encourage US uptake. Dedicated facilities for low-grade polysilicon utilized by solar cells will be coming on line in the next year or two. Thin film is another wild card. I think it's a viable second source and its cost would track more LCD TV costs and price inversely with demand for LCD TVs.