Next Generation Energy is managed and written entirely by Seed editors and expert guest bloggers. Read more about them Solar energy is poised to become the dominant form of electricity this century.
Two forms of solar energy are ready to deliver large quantities of cost-effective carbon-free power: solar photovoltaics, or solar PV, and concentrated solar thermal power, CSP, or as I call it, solar baseload. If the world becomes serious about avoiding catastrophic climate change, we could be adding 50 GW a year or more of baseload solar within a decade, at which time it is likely to be the lowest cost and most widely available form of baseload (or load following) power.
Solar PV could make a major contribution post-2020, more than 1,500 GW (peak), over half a wedge, by 2050. The price could well be below $.12 a kilowatt hour unsubsidized within the decade.
None of this requires a massive government program aimed at solar technology breakthroughs.
BASELOAD SOLAR
After more than a decade of neglect, concentrated solar power has begun rapid growth with more than a dozen providers building projects in two dozen countries. In 2006, the Arizona Public Service Company dedicated the first new CSP plant in the United States in two decades--a 1 MW-concentrated solar trough system with an engine used for decades by the geothermal industry. In June 2007, Nevada Solar One, the state's first CSP plant, went online. On 275 acres near Boulder City, it provides 64 MW of electricity from 98 percent solar power and 2 percent natural gas. And in California, PG&E has created deals with three major CSP companies to generate electricity for the Golden State. Another 10 plants are in the advanced planning stages in the Southwest, along with nine plants in countries that include Israel, Mexico, and China.
Utilities in the Southwest are already contracting for power at 14 to 15 cents per kWh. The modeling for the California Public Utilities Commission puts California solar thermal at 12.7 to 13.6 cents per kWh (including six hours of storage capacity), and at similar or lower costs in the rest of the West. A number of players are adding low-cost storage that will delivers peak power when demand actually peaks, rather than just delivering a constant amount of power around the clock. Thermal storage is far less expensive with a much higher round-trip efficiency than electric storage. That is why I call this baseload solar.
Equally important, CSP has barely begun dropping down the experience curve as costs drop steadily from economies of scale and the manufacturing learning curve. The California Public Utilities Commission analysis foresees the possibility that CSP could drop 20 percent in cost by 2020. A 2006 report by the Western Governors Association, "projects that, with a deployment of 4 GW, total nominal cost of CSP electricity would fall below 10¢/kWh."
It also asserts that deployment will likely occur before 2015. Indeed, the report noted that the industry could, "produce over 13 GW by 2015 if the market could absorb that much." The report also notes that 300 GW of CSP capacity can be located near existing transmission lines.
As an aside, wind power is a very good match with CSP in terms of their ability to share the same transmission lines, since a great deal of wind is at night, and since CSP, with storage, can be dispatched in a controllable manner.
A new report from Environment America, "Solar Thermal Power and the Fight Against Global Warming," explains how the United States could achieve 80 GW of CSP by 2030. A number of industry and academic experts recently discussed the possibility of 10,000 solar GW globally by 2050 at an energy forum in Hanover, Germany.
CSP plants can also operate with a very small annual water requirement because they can be air-cooled -- although that does add about 10% to the cost of power. CSP has some unique climate-friendly features. It can be used effectively for desalinating brackish water or seawater. That is useful for many developing countries today, and it's a must-have for tens, if not hundreds of millions, of people if we don't act in time to stop catastrophic global warming and, as a result, dry out much of the planet. Such desertification would, ironically, mean even more land ideal for baseload solar.
The technology has no obvious bottlenecks and uses mostly commodity materials-- steel, concrete, and glass. The central component, a standard power system routinely used by the natural gas industry today, would create steam to turn a standard electric generator. Plants can be built in a few years--much faster than nuclear plants. It would be straightforward to build baseload solar systems at whatever rate industry and governments needed, ultimately 50 to 100 GW a year growth or more.
SOLAR PV
The best-known form of solar is PV, direct conversion of sunlight to electricity. PV has historically been quite expensive, but its costs have been coming down for decades, and sales have been growing at some 50 percent per year recently. Last year, global PV installations surpassed 2,800 MW of new capacity, which represents growth of more than 60 percent from 2006 levels.
It is difficult to compare PV costs with traditional baseload sources like nuclear because, on the one hand, PV delivers power only about 20 percent of the time. On the other hand, PV can be installed directly on the roofs of buildings. PV therefore avoids transmission and distribution costs and associated losses, while providing power directly to retail customers when it is typically most expensive-- during the sunny days of the summer.
Because it is a modular, low-maintenance consumer product, PV can make use of innovative financing strategies whereby the customer does not own the equipment, but merely purchases the power. SunEdison company is a leader in providing such solar energy services with no upfront costs. In a recent interview, Jigar Shah, the company's chief strategy officer, explained to me that his company could deliver Florida more kilowatt-hours of power with PV--including energy storage so the power was not intermittent-- for less money than Progress Energy has said its planned new nuclear plants would cost. And PV would have no risk of price escalation in the face of construction delays or rising prices for uranium.
Shah projects that by 2015, solar PV will be able to provide electricity directly to the customer for $.12 per kWh unsubsidized. PV could provide 100,000 MW of U.S. capacity in 2020, and 350,000 MW by 2030.
Between energy efficiency, wind power, solar PV, and solar baseload (along with its existing nuclear and hydro plants), the United States can generate all of the low -carbon electricity it will need for decades.
Comments
"Solar energy is poised to become the dominant form of electricity this century."
Yes - absolutely - and the quicker the US understands this fact, the sooner we'll start to lessen CO2 levels.
Posted by: Fair Trade | August 14, 2008 3:42 PM
Standard PV may only provide direct powre 20% of the time but if you look at this new discovery http://www.truthout.org/article/major-discovery-from-mit-primed-unleash-solar-revolution you have an exciting new possibility. Use PV to run your home during the day and to split water into O and H then use the gas at night to power a fuel cell to produce power for the home at night as well as refueling your H powered car, or recharging your electric vehicle. Now combine that with this new fuel cell design http://www.sciencedaily.com/releases/2008/07/080731143916.htm which will dramatically lower the cost of fuel cells as the amount of Pt needed as a catalyst is negligible. If this can be made to work then who needs energy companies anymore, at least for the home.
Posted by: Doug Alder | August 14, 2008 9:04 PM
so why do I still hear the 'new nukes are still needed' call all the time? How can the good news of CSP baseload be communicated to the people and our government? 0.12$ per KwH seems pretty competitive to me. I'm paying 0.11 right now in VA, which I assume is mostly coal based, with some from the Lake Anna nuclear facility, nat gas and some wind.
Posted by: darth | August 15, 2008 6:12 PM
It's unclear exactly which customer price they're talking about. $0.12/kWh at the production end could translate to upwards of $0.20/kWh at the residential end.
Posted by: Anthony | August 15, 2008 6:43 PM
I usually defer the analysis of solar energy costs to the esteemed Dr. Buzzo of Depleted Cranium:, but in June I came across some a solar array data set on The Gristmill, to which commenters had applied useful and illuminating analyzed on this usually very dark subject. The data was on Navada Solar 1, a 64MW CSP facility completed last year. The Nevada Solar 1 site covers 400 acres, and was completed in 2007 at a cost of over 260 million dollars. Nevada Solar 1 was upon completion the third largest solar generating facility in the world. Gary Lipow discussed Nevada Solar 1, in a May 21 post on Gristmill, in which he reported that Nevada Solar 1 produced 134,000,000 kWh in a year. Such statement are typical of renewable supporters, big numbers that hide inefficiencies and costs. Lets brake the 134,000,000 kWh figure down. Assuming 8 hours a day of useful sunshine, Nevada Solar 1 averages producing 45 million KWh. This gives us a 23% capacity factor. A link which Lipow supplies tells us that the Nevada Solar 1 facility cost more than 260 million to build.
We now have some data upon which to build an analysis. In a comment on Lipow's post “Skeptico”noted that the Navada 1 plant cost over $17,000 per kW.
Lipow responded "If you look at costs over 20 years and include interest and O&M that is a cost somewhere between 10 and 12 cents per kWh - not an unreasonable cost for peaking power."
Skeptico shot back, "$260,000,000 / 134 Million KWhr/year /20 years = 10c per kWHr. But that is before any maintenance and before interest. The entire $250 million is paid up front. At only 5% that increases costs by a factor of 2.65, and that's before maintenance."
Sean Casten asked, "How do you get 10 - 12 cents from these numbers, and then added, that assuming a reasonable 10% rate of return on the $260 million, "[p]aying off that money requires a $30.5 million/year annuity. On 134 million kWh, that equates to 23 cents/kWh, and is before including any maintenance or operating costs, or returns above capital recovery. Drop the rate to 5% (e.g., absurdly low) and you're still looking at 16 cents/kWh."
Sean Casten repeated his question, "How do you get 10 - 12 cents from these numbers?"
Lipow responded with one word, "Subsidies."
Ah ha, we have some light!
In a June 14 comment on Gristmill, Gar Lipow accused Skeptico of using a dishonest argument, because "Skeptico took the kWh per year, treated them as baseload, and calculated capacity based on treating a peaking plant as a baseload plant." Lipow's accusation is strange because solar advocates seldom refer to solar plants as limited to peak load, even though that is largely the case, Lipow had not mentioned the words "peak load" in his May 21 post.
Lipow offered a further response to Sean Casten:
I looked at this particular plant because it happens to be the most recent plant. But it is only 64 MW - tiny and not able to grab the full economies of scale needed for CSP. . . . But CSP of 300 MW or greater is normally considered to pen out to a levelized cost (with interest) of 9-12 cents per kWh."
Lipow offered as proof of this a link to a pamphlet on CSP from Sandia Lab, which claims:
"Current technologies cost $2–$3 per watt. This results in a cost of solar power of 9¢–12¢ per kilowatt-hour. New innovative hybrid systems that combine large concentrating solar power plants with conventional natural gas combined cycle or coal plants can reduce costs to $1.5 per watt and drive the cost of solar power to below 8¢ per kilowatt hour."
Thus we have a total disconnect between the theoretical world inhabited by Lipow and Sandia scientists, and the empirical world of Nevada Solar 1. Lipow's sole excuse for the disconnect is found in the words "economies of scale." No case for cost lowering through economies of scale is laid out.
Another issue emerged from the same discussion on June 14 discussion. “Nucbuddy” noted a growing controversy on water use by desert solar facilities. Nevada Solar One near Boulder City uses 400 acre feet of water for 64 megawatts of power. Nucbuddy also notes that the $260 million cost of Nevada Solar1, does not include the cost of electrical transmission lines.
We now have enough data to do an initial comparison between Concentrated Solar Power and nuclear power. If Nevada Solar 1 were scaled up to produce the equivalent annual electrical production of a AP-1000 nuclear plant. Our solar power plant would occupy about 42 square miles of desert, and would cost $17 Billion. Overnight storage of heat, electrical transmission lines, and interest would carry additional costs. Our 1 GW solar facility would annually consume nearly 27000 acre feet of rare desert water.
In contrast, the two 1.7 GW Mitsubishi's Advanced Pressurized Water Reactor (APWR) Luminant Energy is planing to build at Comanche Peak are currently estimated to cost $5-6 billion each.
I have more solar (CSP) construction cost and performance data. Solana CPS facility under construction at Gila Bend, Ariz., will have a name plate capacity of 284 megawatts. The facility will cover about 1,900 acres. And its cost is piously estimated at 1 Billion. No power output estimate is reported but the power will reportedly sell for 4 billion dollars over a 30 year period of time. Assuming a 23% capacity factor that we calculated from Nevada Solar 1, the daily power out put would be 24 X 280 MWs X .23 = 1.5456 GWh Per day, The $1 billion dollar figure would appear to be proportional the Nevada Solar 1 construction costs which ran little more that $4 million per name plate MW. $4 Million X 280 MWs = $1120 Million. 1 Billion would be about 10% less, but who wants to bet on the $1 Billion figure considering inflation? We had to go with a dummied up capacity factor from Nevada Solar 1, but as we will see the data is not out of line. We get an average of 192 MWh electrical production. Now lets try out our .23 per KWh cost figure. 192 MWh is worth 192,000 X .23 X 8 = $253,280 x 365 = $128,947,200 which is remarkably close to $4 Billion divided by 30 years = 133,333,333. This is assuming something that is not in fact clear, that is that stated construction costs include interest.
We also have a similar land use pattern. 280 MW of name plate power requires 3 square miles of land. We don't have any information on water use yet. The Solana data set is far from complete and our cost figure is far from final but the data I do have increases my confidence in the Nevada Solar 1 data set and conclusions I have drawn from it.
My conclusion is simple. Solar investments exists because of government subsidies: The government now pays 30 percent of the capitol investment costs of businesses that invest in solar power to meet our energy need.
Renewable energy production tax credit: This program gives wind, solar, geothermal and other renewable power sources a leg up with a 1.9-cent per kilowatt-hour tax credit, which makes them more competitive with natural gas or coal-fired power plants. Every advocate of solar power believes that solar power in an "infant industry" that needs to be supported by such lavish subsidies. Failure to do so, the solar advocates tell me, will doom the human species to disappear from the face of our planet.
In addition to investments in solar arrays rate payers or tax payers are going to pick up the tab on a $1.5 million per mile cost for new transmission lines.
Depending on various factors, building one MW of solar energy can involve an investment of up to $7 million. That is before interest, and does not include overnight energy storage. Solar theorist claim that solar investment costs are going to drop to a $3.5-5 million soon. It is not clear if that figure includes inflation, because the word inflation never appears in discussions of solar power. According to solar experts in the next few years the cost of solar facilities may drop as low as $2.5 million per MW. That is expected to happen shortly after the Starship Enterprise gets its warp drive coil.
"Sunflower" in a comment during June on Grustmill provided the following prices quotes for CSP: Ausra (line focus) claims $100/m2, BrightSource (power tower heliostats) $150/m2, Matrix Solar Dish (me) $100/m2.
Let's do a little analysis. A square kilometer at $100 per square meter would cost $100X1000X1000 = 100,000,000 per square kilometer or $247,000,000 per square mile. This represents an improvement over the $260,000,000 for 400 acres figure we got for Nevada Solar 1, or the $1 billion for 1900 acres figure we got for Solana, but again the word inflation did not appear in the discussion.
The 150/m2 estimate gives us $370500000 per square mile, still a little better than Nevada Solar 1 in price.
Ausra and PG%E have announced a 1 square mile line focus generating facility with a name plate power rating of 177 MWs. The facility is to be located at San Luis Obispo. No price tag has been placed on it yet, so it is impossible to tell if the $100/m2 figure will hold.
BrightSource has a contract with PG%E that calls for the construction of three solar facilities producing a total of 500 MWs. The cost appears to be estimated in the two to three billion dollar range. The size of the facility does not appear in press releases. However, press releases contained a a statement by John Woolard, CEO of BrightSource that the United States production tax credits are "absolutely critical" for his development. "Otherwise these plants will get built all over the world and not in the U.S." It would appear then that the BrightSources facility is not so much of a bargain, since it is too expensive to build without a massive federal subsidy.
According to the Tree Hugger, the first BrightSource facility has a name plate rating of 100 MWs, and will produce 246,000 MWs of electricity a year. That means that the BrightSolar facility will produce peak power with at an .84 capacity factor for 8 or so hours a day. The cost for the first BrightSource Unit would run between four and six hundred million, thus would fall in the range the Nevada Solar 1 range.
While we have not direct report of the size of BrightSource facilities, their mirror test facility in Israel occupies 12,000 Square Meters. The facility produces 1.5 MWs of electricity. This would scale up to about 800,000 Square Meters or 0.8 square kilometers. Subflower gave us the estimate of $150 Per M(2) for BrighSource,and for a 100 MW facility that should cost about $120,000,000 which is way lower than my $400,000,000 to $600,000,000 guestimate. And remember that the guestimate was based on BrrightSource's own statement.
A further consideration would be that BrightSources own estimated cost falls within the cost range of current cost estimates for nuclear power plants costs. For the basically the same price as a 1 GW BrightSource generating facility PG&E could buy a 1 GW reactor that would generate power day and night, rain or shine with 3 times the daily electrical output of the BrightSource facility.
Posted by: Charles Barton | August 23, 2008 12:02 PM