Amory B. Lovins, "Renewable Energy's 'Footprint' Myth," Electricity Journal, vol. 24, July 2011, pp. 40-47. Reproduced with permission from Elsevier.
"An average square meter of land receives each year as much solar energy as a barrel of oil contains, and that solar energy is evenly distributed across the world within about twofold."
Environmental scientist Amory B. Lovins is chairman and chief scientist of the Rocky Mountain Institute, a research organization dedicated to sustainable living. He is the author of such books as Small Is Profitable and Natural Capitalism. In the following viewpoint, Lovins rebuts arguments that the land footprint of renewable energy production facilities—such as windmills and solar arrays—is larger than traditional fossil-fuel and nuclear energy production facilities. Lovins contends that wind farms are often placed in regions in which the intervening land between the windmills can be used for other purposes; he states that this is not the case for nuclear facilities that may have a smaller building footprint but normally prohibit the use of land around the buildings. Furthermore Lovins claims that photovoltaic solar panels are typically placed on building roofs, thus requiring no additional real estate for energy production. Finally Lovins points out that nuclear power plants require even more land to mill and refine the fissionable materials and to bury the hazardous waste products.
As you read, consider the following questions:
- According to Lovins and the Brookhaven National Laboratory, how much land does a 1 GW plant use over a forty-year life cycle?
- What percent of the land on a wind farm site can be used for purposes other than producing energy, in Lovins's opinion?
- As Lovins states, what percent of photovoltaic accumulators are placed on building roofs, thus using no extra land?
Land footprint seems an odd criterion for choosing energy systems: the amounts of land at issue are not large, because global renewable energy flows are so vast that only a tiny fraction of them need be captured. For example, economically exploitable wind resources, after excluding land with competing uses, are over nine times total national electricity use in the U.S. and over twice in China; before land-use restrictions, the economic resource is over 6 times total national electricity use in Britain and 35 times worldwide—all at 80-meter hub height, where there's less energy than at the modern ≥100 m [meter]. Just the 300 GW [gigawatt] of windpower now stuck in the U.S. interconnection queue could displace two-fifths of U.S. coal power. Photovoltaics [PV], counting just one-fifth of their extractable power over land to allow for poor or unavailable sites, could deliver over 150 times the world's total 2005 electricity consumption. The sunlight falling on the Earth about every 70 minutes equals humankind's entire annual energy use. An average square meter of land receives each year as much solar energy as a barrel of oil contains, and that solar energy is evenly distributed across the world within about twofold. The U.S., "an intense user of energy, has about 4,000 times more solar energy than its annual electricity use. This same number is about 10,000 worldwide [so] ... if only 1 percent of land area were used for PV, more than 10 times the global energy could be produced," [states a US Department of Energy and Electric Power Research Institute 1997 report].
Nonetheless, many nuclear advocates argue that renewable electricity has far too big a land "footprint" to be environmentally acceptable, while nuclear power is preferable because it uses orders of magnitude less land. If we assume that land-use is an important metric, a closer look reveals the opposite is true.
For example, Stewart Brand's 2010 book Whole Earth Discipline cites novelist and author Gwyneth Cravens's claim that "A nuclear plant producing 1,000 megawatts [peak, or about 900 megawatts average] takes up a third of a square mile." But this direct plant footprint omits the owner-controlled exclusion zone (∼1.9-3.1 mi2 [square mile]). Including all site areas barred to other uses (except sometimes a public road or railway track), the U.S. Department of Energy's nuclear cost guide says the nominal site needs 7 mi2, or 21 times Cravens' figure. She also omits the entire nuclear fuel cycle, whose first steps—mining, milling, and tailings disposal—disturb nearly 4 mi2 to produce that 1 GW plant's uranium for 40 years using typical U.S. ores. Coal-mining to power the enrichment plant commits about another 22 mi2-y of land disturbance for coal mining, transport, and combustion, or an average (assuming full restoration afterwards) of 0.55 mi2 throughout the reactor's 40-year operating life. Finally, the plant's share of the Yucca Mountain spent-fuel repository (abandoned by DOE [US Department of Energy] but favored by Brand) plus its exclusion zone adds another 3 mi2. Though this sum is incomplete, clearly Brand's nuclear land-use figures are too low by more than 40-fold—or, according to an older calculation done by a leading nuclear advocate, by more than 120-fold.
Exaggerating Renewables' Land Use
This is strongly confirmed by a new, thorough, and authoritative assessment I found after completing the foregoing bottom-up analysis. Scientists at the nuclear-centric Brookhaven National Laboratory and at Columbia University, using Argonne National Laboratory data and a standard lifecycle assessment tool, found that U.S. nuclear-system land use totals 119 m2 [square meters] /GWh [gigawatt hour], or for our nominal 1 GW plant over 40 years, 14.5 mi2—virtually identical to my estimate of at least 14.3 mi2....
Of this 119 m2/GWh of land-use, Brand counts only 2.7 m2/GWh—1/16th of the power-plant site—or 2.3 percent. Not that he's unaware of the concept of a fuel cycle, which he bemoans for coal. His land-use errors for renewables, however, are in the opposite direction. "A wind farm," he says, "would have to cover over 200 square miles to obtain the same result [as the 1 GW nuclear plant], and a solar array over 50 square miles." On page 86 he quotes Jesse Ausubel's claim of 298 and 58 square miles respectively. Yet these windpower figures are about 100-1,000 times too high, because they include the undisturbed land between the turbines—about 98-99+ percent of the site—which is typically used for cultivation, grazing, wildlife, or other uses (even solar collection) and is in no way occupied, transformed, or consumed by windpower. For example, the turbines that make 15 percent of Iowa's electricity rise amidst farmland, often cropped right up to the base of each tower, though wind royalties are often more profitable than crops. Saying that wind turbines "use" the land between them is like saying that the lampposts in a parking lot have the same area as the parking lot: in fact, about 99 percent of its area remains available to drive, park, and walk in.
The area actually used by 900 average MW [megawatt] of windpower output—unavailable for other uses—is only about 0.2-2 mi2, not "over 200" or "298." Further, as noted by Stanford's top renewables expert, Prof. Mark Jacobson, the key variable is whether there are permanent roads. Most of the infrastructure area, he notes, is temporary dirt roads that soon revegetate. Except in rugged or heavily vegetated terrain that needs maintained roads, the long-term footprint for the tower and foundation of a modern 5 MW tubular-tower turbine is only about 13-20 m2. That's just about 05 mi2 of actual windpower footprint to produce 900 average MW: not about 50-100 times but 22,000-34,000 times smaller than the unused land that such turbines spread across. Depending on site and road details, therefore, Brand overstates windpower's land use by two to four orders of magnitude.
Solar Panels Require the Least Additional Land Use
His photovoltaic land-use figures are also at least 3.3-3.9 times too high (or ≥4.3 times versus an optimized system), apparently due to analytic errors. Moreover, some 90 percent of today's photovoltaics are mounted not on the ground but on rooftops and over parking lots, using no extra land—yet 90 percent are also tied to the grid. PVs on the world's urban roofs alone could produce many times the world's electricity consumption. [In 2006] the National Renewable Energy Laboratory found that:
In the United States, cities and residences cover about 140 million acres of land. We could supply every kilowatt-hour of our nation's current electricity requirements simply by applying PV to 7% of this area—on roofs, on parking lots, along highway walls, on the sides of buildings, and in other dual-use scenarios. We wouldn't have to appropriate a single acre of new land to make PV our primary energy source! ... [I]nstead of our sun's energy falling on shingles, concrete, and under-used land, it would fall on PV—providing us with clean energy while leaving our landscape largely untouched.and concludes: "Contrary to popular opinion, a world relying on PV would offer a landscape almost indistinguishable from the landscape we know today." This would also bypass the fragile grid, greatly improving reliability and resilience.
Table 1 summarizes, then, the square miles of land area used to site and fuel a 1 GW nuclear plant at 90 percent capacity factor, versus PV and wind systems with the same annual output.
Thus windpower is far less land-intensive than nuclear power; photovoltaics spread across land are comparable to nuclear if mounted on the ground in average U.S. sites, but much or most of that land (shown in the table) can be shared with livestock or wildlife, and PVs use no land if mounted on structures, as about 90 percent now are. Brand's "footprint" is thus the opposite of what he claims.
Material Production Costs for Nuclear and Renewables
These comparisons don't yet count the land needed to produce the materials to build these electricity supply systems—because doing so wouldn't significantly change the results. Modern wind and PV systems are probably no more, and may be less, cement-, steel-, and other basic-materials-intensive than nuclear systems—consistent both with their economic competitiveness and with how quickly their output repays the energy invested to make them. For example, a modern wind turbine, including transmission, has a lifecycle embodied-energy payback of under seven months; PVs' energy payback ranges from months to a few years (chiefly for their aluminum and glass housings); and adding indirect (via materials) to direct land-use increases PV systems' land-use by only a few percent, just as it would for nuclear power according to the industry's assessments. Indeed, a gram of silicon in amorphous solar cells, because they're so thin and durable, produces more lifetime electricity than a gram of uranium does in a light-water reactor—so it's not only nuclear materials, as Brand supposes, that yield abundant energy from a small mass. Their risks and side effects, however, are different. A nuclear bomb can be made from a lemon-sized piece of fissile uranium or plutonium, but not from any amount of silicon. Only for that purpose is energy or power density a meaningful metric. For civilian energy production, it's merely an intriguing artifact. What matters is economics and practicality.