by Vaclav Smil
In (largely) gloomy Germany the area needed by PV panels to supply all electricity generation (nearly 560 TWh in 2012) would be considerably larger. With an average PV output of 100 kWh/m2 (the recent annual mean for both roof - and ground-based installations), it would require about 5,600 km2 covered with modules. That would be the equivalent of nearly 1.6% of Germany's total area, 25% of the country's built-up area, or almost 15% of land claimed by settlements and transportation infrastructure; and roughly 2.7 times the total area of all German roofs, based on an estimate of roughly 25 m2 of roof area per person (Waffenschmidt 2008).
What It Would Take
If you are willing to engage in unbounded science and engineering fiction, then, according to Jacobson and Delucchi (2011), this is what it would take to supply the world with 100% renewable energy in 2030 by using electricity (generated by wind, water, and solar PV installations) and electrolytic hydrogen for all purposes: 3.8 million 5-MW wind turbines, 49,000 300-MW central solar plants, 40,000 300-MW solar PV plants, 1.7 billion 3-kW rooftop PV installations, 5,350 100-MW geothermal plants, 270 new 1.3-GW hydro stations, 720,000 0.75-MW wave devices, and 490,000 1-MW tidal turbines. All of that would require only about 0.4% of the world's land for its footprint and 0.6% for spacing, and we are assured that the "barriers to the plan are primarily social and political, not technological or economic" as the energy cost in a new wind-water-solar world "should be similar to that today."
These assurances asides, the simplest reality check shows the fictional nature of these assumptions. In 2013 the worldwide capacity in wind turbines reached about 330 GW, while 13 TW (40 times as much) would be needed by 2030. Total rooftop and large plant PV capacity reached about 100 GW, but 17.1 TW of these installations would be required (170 times as much); moreover, there was not a single 300-MW solar PV plant (five plants rated between 200 and 250 MW), whereas 40,000 would be needed by 2030. In 2013 there was only one central solar power facility rated at more than 300 MW, Ivanpah, at 392 MW, but nearly 50,000 such facilities would be needed by 2030 (an increase of four orders of magnitude). There were fewer than 50 geothermal stations rated at more than 100 MW, but 5,350 would be needed (a 100-fold increase). Pelamis (2014, the world's most advanced wave energy company, produced six 0.75-MW devices by the beginning of 2014, but 720,000 would have to be operating by 2030 (an increase of five orders of magnitude). Finally, by 2013 there were fewer than ten small tidal stations with aggregate installed power of much less than 1 GW, while 490 GW would have to generate by 2030 (two orders of magnitude more).
Such a ramping-up of all kinds of capacities-design, permitting, financing, engineering, construction, all going up between one and five orders of magnitude in less than two decades-is far, far beyond anything that has been witnessed in more than a century of developing modern energy systems. And that still leaves out two other key facts, namely, that such a gargantuan renewable energy system would need an enormous expansion of high-voltage transmission and would require the creation of an entirely new, hydrogen-based society. I still am not sure how we would fly with hydrogen (or electricity) or smelt pig iron. In any case, the chances of a 100% water-wind-solar world to be ready by 2030 are nil, but it is worth while exploring what it would (realistically) take to create an increasingly nonfossil global energy system.
Making Choices
To begin with, even though solar generation has the highest power densities (which may further improve with more efficient PV modules), it would be a very unwise engineering choice to aim at 100% PV-based electricity generation even if moderately good storage options were available. Any sensibly designed all-renewable system would aim at combining different techniques by adding, in all large nations, substantial shares of windgenerated and central solar power, and also (where possible) more geothermal electricity. These requirements would necessitate greatly expanded long-distance high-voltage connections, and this would result in additional large land claims in all countries with extensive territories, but because these claims are difficult to calculate without a great deal of specific assumptions, I will leave them aside.
And although electricity's share of final energy use has been steadily rising (in the United States it is now about 40%), fossil fuels dominate in all major economies, where they have become principal sources for transportation, space heating, and an enormous variety of industrial processes. As a result, their substitution presents a greater transition challenge than displacing significant shares of thermally generated electricity by solar and wind-generated power. Setting aside the early emergence of a hydrogen economy, there are two basic paths toward their replacement. The first one is a complete substitution of fossil fuels by a range of phytomass fuels (wood, liquid biofuels, gasification of phytomass) resembling the current sources and able to fit existing uses.
The second, a more desirable and less land-intensive choice, is to produce some phytomass fuels and substitute a large share of their current use by renewably generated electricity used directly to power electric cars, trains, and many industrial processes, and to provide space heating, and indirectly to produce (with an inevitable reduction of efficiency) storable energies in the form of compressed air, hot water, ice, ammonia, and, of course, hydrocarbons made with captured CO2 and some hydrogen. Consequently, the most realistic approach to delimit the approximate land requirements of new renewable energy sources is to use relatively extreme but still plausible scenarios of future energy systems that might eventually displace all fossil fuels. Small countries would not have the requisite flexibility of choice, but it would seem that large nations should have plenty of space to eventually put in place complex renewable energy systems. How much land such arrangements would claim would depend on the composition of final energy use.
A higher degree of electrification in sunny and windy places would entail smaller land claims than the large-scale substitution of gasoline, kerosene, and diesel by biofuels. I have sketched two options for the eventual total displacement of fossil fuels for the US demand at the 2012 level, which entailed roughly 320 GW of fuel-generated electricity and 1.8 TW of coal, oil, and gas, with 60% of this fuel total going for transportation. I have assumed fairly high power densities, in all cases higher than today's practices, and with wind-powered electricity generation I am counting only the land actually occupied by pads and access roads. The first option-displacing all fossil fuel-based electricity generation by solar and wind electricity, and substituting biofuels for all fossil fuels-would claim about 470 Mha, and the entire system would have a low power density of 0.45 W/m2, above all as a result of the enormous areas required to produce liquid biofuels.
Box 8.1
Land required to displace 2012 US fossil fuel demand by new renewables
Electricity generated from fossil fuels: - 320 GW
Liquid fuels**: --1,100 GW
Solid and gaseous fuels***: - 700 GW
Notes: *Counting only pads and roads. **Nonfuel uses are subtracted. ***Coal and gas used for electricity generation were subtracted.
Massive electrification-with half of all fuels, or about 900 GW, replaced by electricity generated by the same mixture of solar and wind conversionswould reduce the total to about 250 Mha, but as this total does not include losses involved in converting some of the generated electricity to storable fuels, the eventual land claim would be significantly larger. In any case, these approximate calculations delimit plausible extremes: an entirely renewable energy system would occupy roughly 25%-50% of the country's territory (250-470 Mha), compared to about 0.5% (5.5 Mha) of land claimed by today's fossil fuel-hydro-nuclear system.
Of course, there are many plausible adjustments of the grand total, but closer examinations reveal their limits. The land claims for PV-based electricity generation could be cut by a third or a half by placing more panels on roofs rather than amassing them in solar farms in deserts or placing them on abandoned industrial properties or on strips of land along highways. An increasing share of renewable electricity generation can be moved of
fshore: large marine wind farms already exist (EWEA 2013), and various means of ocean power extraction are under development, ranging from classic ocean thermal energy conversion (OTEC) schemes (Faizal and Ahmed 2012) to various wave-power devices and turbines powered by tides or ocean currents and also including utility-scale undersea energy storage (Slocum et al. 2013).
All of these immersed or submersed conversions share two commonalities: they have to operate in a demanding, corrosive environment, and (with the exception of strong offshore winds, which can be harnessed with high capacity factors) the thermal and kinetic energies they are converting have very low power densities. Even OTEC, the oldest idea among ocean energy conversions, has not progressed to reliably operating projects. Wave devices are in the earliest stages of commercial development, and except in Scotland there are no bold plans for their mass deployment. The demand for liquid fuels can be reduced by a third or more as more efficient vehicles get to dominate the market, and the same is true (in the longer term, owing to a slower turnover of housing stock) of household energy use.
But even after cutting the lower estimate of future US renewable energy needs by a third we are left with nearly 170 Mha (about 17% of the country, and that is without additional transmission ROWs that would be needed to link the new renewable generation capacities), that is, with more land than is in annually harvested crops. Large as it is, such a share could be, in extremis and costs aside, accommodated by the world's third largest nation-but that is not the case when analogical land claim calculations are done for smaller countries or for the largest island nations.
Even under the assumption of a very high average power density of 1 W/m2 (which would require a higher rate of the system's electrification), the UK would need about 240,000 km2, or virtually its entire territory. Similarly, McKay (2013, 1) concluded that "in a decarbonized world that is renewable powered, the land area required to maintain today's British energy consumption would have to be similar to the area of Britain." Germany has gone further in installing wind and PV-based electricitygenerating capacities than any other affluent economy, but setting up a completely renewable system based on the best available conversions would require, even with a high power density of 1 W/m2 and even with all roofs covered by PV panels, about 350,000 km2, again essentially the country's entire area.
But it could be a lot worse. Japan could completely decarbonize with nearly 600,000 km2 of land devoted to electricity generation and phytomass fuels, nearly 60% more than the area of the four main islands, and the land requirements of fully renewable national energy systems would surpass entire territories for numerous high-energy countries ranging from such island states as Singapore, Taiwan, and Trinidad to such industrial powers as South Korea or the Netherlands. Again, assorted measures (from rooftop PVs and offshore wind to more efficient cars and lights) could cut these demands by a third or a half, but even then those fully renewable systems would claim impractically large shares of national territories. Given these realities, it is astonishing that Lovins (2011b, 40) would claim that "land footprint seems an odd criterion for choosing energy systems: the amounts of land at issue are not large" and that "for civilian energy production, it's merely an intriguing artifact." Some artifact!
Power densities matter, and this means that the transition from predominantly fossil fuel-based to purely renewable energy systems cannot take place-even in affluent, populous countries with large territories and with excellent conditions for PV-based and wind electricity generation and for phytomass cultivation-by simply following a variant of one of the replacement options just outlined. A large-scale international trade in renewables would help, as it does in the modern fossil fuel system, where trade accounts for almost 20% of all coal use, two-thirds of crude oil demand, and nearly one-third of the natural gas supply (BP 2014; CornotGandolphe 2013). But trading low-energy-density phytomass fuels produced with low power densities in Amazonia would not be obviously the same as trading high-energy-density crude oil produced with exceptionally high power densities in the Middle East.
Moreover, the biospheric realities mean that a truly massive trade in phytomass fuels that could be harvested on a large scale and with high yields could be sourced from only a few tropical countries, predominantly Brazil. Extensive new phytomass production in all other countries with large territories (Russia, Canada, the United States, Australia) is limited either by a cold climate or by recurrent droughts, while serious land scarcity eliminates cultivated phytomass as a major option for the world's four most populous low-income nations, China, India, Indonesia, and Pakistan. Marginal lands, barren hilly slopes, and abandoned, low-productivity farmland are claimed to have a large potential for producing biofuels, but, because of their aridity and poor soils, their low productivity (well below 1 W/m2) would come at a high environmental cost (soil erosion, nutrient loss, biodiversity reduction). A heavy reliance on Brazil would further imperil the always precarious state of Amazonian forests: after several years of decline, the region's deforestation rate rose by nearly 30% between August 2012 and July 2013 (INPE 2013).
Decentralized Energy Supply?
Of course, a purely renewable energy supply would be easier to realize if electricity generation were to be massively decentralized-and decentralization of power and distributed generation have been the leading mantras of renewable energy advocates. The setup entails electricity generation by small units that may or may not be connected to the grid but that are always close to the point of final use, a solution that appeals to green sensibility as it conforms to the small-is-beautiful ideal. However, a reality check is in order: how can this prospect be squared with the growth of megacities whose densely crowded, high-rise blocks may average throughout the year more than 500 W/m2 and reach 1,000 W/m2 during the hours of peak demand?
Box 8.2
Decentralized PV generation in Tokyo
Data sources: Tokyo Metropolitan Government (2006, 2012).
Since 2007 more than half of the world's population has been living in cities. By 2050 that share will be above 70%, and more than half will live in megacities with populations of more than 10 million, areas with the highest power density of final energy uses. Even if the power densities of energy use in many megacities were to decline gradually in the decades ahead, it would be impossible to supply them with decentralized PV-based electricity. The world's largest megacity, today's metropolitan Tokyo, offers a perfect example of these limits.
Even if PV panels operating with 12% conversion efficiency were to cover the entirety of more than 600 km2 of 23 densely populated wards, they could supply only half of Tokyo's energy requirements. And the entire metropolis, whose power density averages about 10 W/m2, could get all of its energy from PV panels only if they were to cover about 70% of its nearly 2,200 km2, an impossibility unless we resort to science fiction visions of cities under plastic bubbles. The shares of annual demand that could be realistically delivered are only small fractions of the total. As previously noted, all potentially available roofs within Tokyo's most densely inhabited twenty-three specials wards are about 64 km2 (Stoll, Smith, and Deinert 2013), that is, just 10% of the total area; but the practical availability would be only a fraction of that, and roofs are a smaller share of a much less densely inhabited area outside the 23 wards.
Extensively used PV panels would thus supply less than 10% of the annual demand of the entire metropolis and less than 1% of the need in its core area, where average power densities are an order of magnitude higher than the metropolitan mean (and could do so only if they were tied to adequate storage capacities). Not surprisingly, the Global Energy Assessment for 2012 concluded that local renewables "can therefore only supply urban energy in niche markets (e.g., low-density residential housing), but can provide less than 1% only of a megacity's energy needs" (Global Energy Assessment 2012, 1347). And, obviously, large-scale electricity storage would be essential in order to supply urban infrastructures that operate 24/7/365 and that need to cover demand that often pea
ks at night owing to household air conditioning in hot climates.
A similar, or even larger, mismatch between the power densities of production and final use will apply to many highly energy-intensive industrial processes, above all to the smelting and casting of metals and to chemical syntheses ranging from ammonia to plastics and composite materials. Obviously, the most efficacious way to supply megacities and energyintensive industries would be by converting energies with an even higher power density in their proximity, either by relying on domestic resources or by importing high-energy-density fuels-while finding large nearby areas capable of supporting large wind or solar capacities might be impractical or impossible. As there is no such renewable option, megacities would have to rely on high-voltage links to distant concentrations of wind and solar generation capacities.
How disruptive that shift will eventually be to traditional centralized utilities remains uncertain. The Edison Electric Institute concluded that despite the risks presented by the rapidly growing penetration of distributed energy, this shift (as long as its degree remains low, as it does in the United States) is
not currently being discussed by the investment community and factored into the valuation calculus reflected in the capital markets. In fact, electric utility valuations and access to capital today are as strong as we have seen in decades, reflecting the relative safety of utilities in this uncertain economic environment. (Kind 2013, 2)
But a study by the Lawrence Berkeley Laboratory showed substantial revenue erosion once the share of rooftop solar generation reaches 10%, although specific effects of higher shares of solar photovoltaics on earning will depend on the type of utility (Satchwell et al. 2014).
Another way to boost renewable generation would be to have a vastly expanded high-voltage link and eventually a global grid. That has been an aspirational goal for decades (Fuller 1981; GENI 2014), but its early emergence (with links crossing the Bering Strait, and connecting North America with Europe via Iceland) is most unlikely, and future large-scale renewablesource electricity transmission will be limited for a long time to regional interconnections (lines from the Algerian Sahara to the EU, or from Arizona to New England, or from Xinjiang to coastal China). Particularly advantageous links will be those that connect distant production and demand areas across several time zones in order to take advantage of different generation and demand peaks (the three-hour difference between CSP plants in the US Southwest and the populous northeastern coast is perhaps the best example).
