When the APS Forum on Physics and Society publication "Physics and Society" published an article by one Wallace M. Manheimer suggesting with little evidence that renewable energy was useless, I felt obligated to respond with some more complete information. They published my letter in the latest issue; I've reprinted it below.
Wallace M. Manheimer's article  on energy choices in the April 2012 issue makes a number of important points, but also goes wrong on many fronts, and I hope Physics & Society will allow at least some correction of these misstatements.
To start at the end, Manheimer asserts that "one cannot talk about climate and ignore energy supply. Yet, these organizations [AIP and APS] have done just that." One need only read the same issue of Physics and Society to know that claim is false - the book review by Paul P. Craig  mentions "the first APS energy study [...] in 1973", which has been followed by many others. Manheimer himself cites the recent APS "Energy Efficiency Report" - and then appears to dismiss it as parochial. This is ironic since he earlier claims that cutting US energy use would be "worse because distances are much greater in the United States, it is colder here, and we have responsibilities as a major world power" Manheimer's argument pertains to Italy, but in general technology developments allowing efficiency gains in the US apply equally well or better elsewhere.
An examination of the numbers in Manheimer's first concluding paragraph shows one deep inconsistency that permeates his piece. He asserts we need 30 TW of "energy" supply by 2050. He suggests meeting most of this with 20 TWth of nuclear power - but notice the little ‘th' appended, meaning "thermal", not "electric" energy. Energy quality matters, and lumping different types together in this way is quite misleading. The actual electric supply from 20 TWth nuclear reactor capacity would be about 7 or at most 8 TW of actual useful energy. He then refers to the "3-4 TW" from renewables as a "small amount," but that would be entirely in the useful electric form, i.e. even in his budget by 2050, renewables supply half as much electric energy as from nuclear power.
The 30 TW by 2050 is a "thermal," not "electric" number. About 40% of thermal power used now [worldwide – see ref. 3] goes to steam turbine generators that run at about 30% efficiency in converting to electricity; another 20% is in transportation where conversion of thermal to mechanical power is similarly low relative to electric-powered transportation. A good fraction of the rest goes to space heating which has very low energy quality requirements: a given quantity of room-temperature heat can be obtained through a heat pump running against outside or underground temperatures on 1/3, 1/5 or even less electric energy (ref. 4). Only a small fraction (perhaps 20%) of the world's "thermal" energy consumption actually goes to high-temperature industrial process heating that makes efficient use of close to the full energy content of the consumed fuels. So the 30 TW of thermal energy Manheimer worries about translates to perhaps 6 TW of high-quality thermal energy and 24 TW of low-quality thermal energy, which can be met equally well with about 8 TW of electric energy. The real requirement for energy supply by 2050 is 14 or 15 TW of high quality (say electric) power, not 30.
Manheimer's analysis of solar and wind power is not in substantive numerical error, although modern PV panels are usually close to 20% efficient (not 10%), and solar will most likely be deployed where incoming sunlight is well above the global average. However, he leaves a lot out. He makes no mention of the dramatic fall in costs for those technologies, particularly solar photovoltaic systems, nor their dramatic growth rates of 40% or 50% or more per year in recent years . Solar photovoltaic production is almost 3 orders of magnitude higher than it was 20 years ago. Manheimer notes that 1 TW of solar would take roughly 25,000 km2, but says little about how realistic or unrealistic such a level of installation would be. 2011 installations amounted to about 25 GW (peak) or about 5 GW average power. So, 1 TW is 200 years production with no further growth and does still seem distant. Another 20 years of photovoltaics growth like the last, however, and by 2032 we would have not just 1 TW average power from photovoltaics, but close to the full 15 TW the world needs. The area used, 15 times 25,000 km2 or 375,000 km2 (to use Manheimer's estimate), amounts to just 0.25% of Earth's land area. Also, there is no fundamental reason these have to be placed entirely over land. Neither historical production growth rates nor Earth's surface area are limitations on powering the world's 2050 energy needs entirely from the Sun, if that's what we choose to do.
The other key question then is one of economics, but Manheimer's analysis of that is purely based on subsidy levels, and even there is mostly speculative. Capital-intensive energy sources like solar power have fundamental economic characteristics very different from those of fossil-fuel systems. Nuclear power is somewhere in between. Other than interest costs (which can be highly variable but at present for the US government are close to zero), the annualized cost of a solar facility costing $10,000 per average kW (roughly where present-day solar stands) is almost entirely that up-front capital divided by the likely plant lifetime. If the solar plant can be expected to run for 25 years, that comes to $400/kW-yr or about 5 cents/kW-hr. If a 60-year lifetime could be realized, amortized solar plant costs would be less than 2 cents/kW-hr. Interest costs, maintenance and transmission and distribution (and utility profits) would add another few cents/kW-hr, but even now the base cost is not at a point that it would break anybody's electric bill. There is one complication of large-scale solar or wind adoption, also not mentioned in Manheimer's article. Due to the variability of renewable sources, some mechanism for large-scale energy storage needs to be simultaneously integrated to the grid, and grid capacity itself needs to be enhanced to allow power to flow under these more variable conditions. The experience in Germany, which has recently seen over 50% of electric power coming from solar  shows this is not an impossible barrier, but it does mean a small additional cost associated with renewable sources.
The more important economic point is that solar's $10,000/kW-avg is continuing to drop quickly. The "learning curve" for solar photovoltaics has been consistently over 20% (about 22%) per doubling over more than 10 doublings [7, 8]. That is, costs per kW for an annual production level of 50 GW-peak or 10 GW-avg (double the present year's) should be $8,000/kW-avg or less. At 1 TW-avg, with the same historical learning curve, costs would be below $2,000/kW-avg, less than one fifth what they are today. The importance of subsidies is in allowing the industry to scale more quickly to the higher production levels and lower cost that make it truly competitive without long-run subsidy. The Solyndra case is one worth examining in detail: the reason the company received a subsidy is that its costs looked good several years ago, but solar PV prices dropped so fast it became uncompetitive. This price drop is a good thing, but nowhere to be found in Manheimer's article.
The fact is that fossil fuel plants themselves have capital costs in the range of one to several thousand dollars per average kW. Recent experience with constructing nuclear plants has seen costs several times as high - as much or more than the cost of solar photovoltaics. Both fossil and nuclear facilities also have much higher annual continuing costs than solar, from fuel and operations.
Manheimer hardly discusses economics in his article, other than to claim that a switch to renewables "[will] almost certainly condemn the vast majority of the human family to abject poverty". That sounds rather "alarmist" - while he repeatedly accuses the APS and climate scientists of "alarmism." Manheimer states that "standard ‘renewable' energies, solar and sequestration, are nowhere near ready to provide for societal energy needs, and likely never will be" but the reality of the situation is quite otherwise if we give any weight to historical patterns of development in the solar industry (wind has been similar though slower to improve).
Manheimer is absolutely right that global society will need substantially more energy than it uses today. Solar and other renewable technologies are perfectly capable of delivering on that need, with prices lower than fossil fuels and nuclear power well before mid-century. Market forces will take over from that point, but the timing and populations that benefit from the transition will be dependent on the action of entities able to provide the hundreds of billions of dollars needed. Climate change is only one of the compelling reasons we should be pushing the United States and other governments into earlier and faster action on the energy technologies of the 21st century.
1. Wallace M. Manheimer, "American Physics, Climate Change, and Energy," Physics & Society, vol. 41(2), p. 14, (April 2012).
2. Paul P. Craig, review of "Physics of Sustainable Energy II: Using Energy Efficiently and Producing it Renewably," Physics & Society vol. 41(2), p. 24 (April 2012).
3. US Department of Energy International Energy Outlook 2011
4. US Department of Energy, "Geothermal Heat Pumps"
5. See for example the following recent news reports on solar industry growth: cleantechnica.com/2012/03/19/worldwide-solar-pv-market-grew-in-2011/ or www.sciencedaily.com/releases/2011/09/110905085957.htm
6. "Solar power generation world record set in Germany", The Guardian, 28 May 2012
7. Alvin Compaan, "Photovoltaics: Clean Electricity for the 21st Century", APS News (April 2005)
8. Kees van der Leun, "Solar PV rapidly becoming the cheapest option to generate electricity," Grist magazine (11 Oct 2011)