by Vaclav Smil
The measure is deceptively simple-power (energy flow per unit of time) per unit of area, in standard form expressed in W/m2-but both the numerator and the denominator hide important differences and complexities. Neither of them can convey the quality of the two measured variables: watts and square meters are measures that will accurately quantify power and area without telling us directly anything either about the quality of a specific energy flow (its environmental externalities, its ease of use, its reliability, its cost, its durability) or about the initial quality of land that was claimed by energy production and use and about the eventual extent of its destruction or alteration. A closer look at some of the most important qualitative attributes of the two variables is important in order to appreciate the inherent deficiencies of the measure.
This information loss is a common problem when using single measures for variables whose quality matters at least as much as, or more than, their quantity. Food consumption is an excellent example of this information deficiency. Every foodstuff can be quantified in terms of energy density (kcal/g, MJ/kg), and average per capita food intake can be expressed in kcal/ day or J/day, but these commonly used quantities hide fundamental nutritional qualities. They tell us nothing about the shares of macronutrients (carbohydrates, lipids, proteins) in foodstuffs, about their specific qualities (all animal proteins are nutritionally superior to plant proteins), or about the presence of micronutrients. As a result, people can consume energetically adequate quantities of food but be malnourished; unfortunately, that is not an uncommon situation for some population groups even in many affluent countries, particularly as far as some key micronutrients (above all iron and zinc) are concerned (Smil 2013c).
Energy Qualities
Quantification of the numerator is subject to frequent but minor inconsistencies arising from the conversion of fuels to common energy equivalents. The most common cause is the slightly different assumptions made regarding average energy densities Q/kg or J/m3) of fuels and the choice (often undefined) of using either higher or lower heating values. But by far the greatest challenge arises from converting various forms of electricity to a primary energy equivalent. With electricity generated by fuel combustion, it is simply a matter of adjusting for the overall efficiency of thermal generation: an aging coal-fired station may have an efficiency of just 33%, the latest model may convert 42% of coal into electricity, and for combined-gas cycle generation the rate may be as high as 60%. Consequently, the primary energy equivalents of the three processes will be respectively 10.9, 8.6, and 6.0 MJ/kWh.
But there is no universally valid (or accepted) way of converting primary electricity that includes nuclear, solar, wind, geothermal, and hydro generation. The two commonly used procedures differ by roughly a factor of three: the simple thermal equivalent is 3.6 MJ/kWh; conversions using the prevailing efficiency in large central fossil-fueled thermal stations range from 8.6 to 11.0 MJ/kWh. This disparity explains most of the differences in primary energy totals offered by data aggregators and statistical services. For example, British Petroleum's widely quoted Statistical Review of World Energy uses the straight thermal equivalent (BP 2014), while the International Energy Agency uses that equivalent (3.6 MJ/kWh) for all forms of renewablesource electricity generation but calculates the primary energy equivalent of nuclear electricity by assuming that 1 kWh equals 10.9 MJ (IEA 2014). These realities also mean that no matter which conversion rates are chosen, the power densities of fuel combustion are always qualitatively different from the power densities of primary electricity use.
The obvious finite versus renewable energy source dichotomy aside, the key difference, and one with enormous environmental and hence socioeconomic implications, is that fossil fuel combustion is always associated with CO2 emissions, the principal source of anthropogenic carbon releases and the leading cause of human interference in the global biogeochemical carbon cycle. The combustion of coals and hydrocarbons also results in the emission of oxides of nitrogen (precursors of photochemical smog), and the burning of coal and of many liquid fuels releases sulfur oxides (key contributors to acid deposition). In contrast, primary electricity (from nuclear fission and renewable energy flows) is not a direct source of CO2 emissions, but carbon emissions do, of course, result from the construction and maintenance of the requisite facilities and converters. Other qualitative differences pertain to the convenience and flexibility of energy use.
Electricity is always easier to use (with a flip or push of a switch) than fuels, although in household settings a modern high-efficiency natural gas furnace with electronic ignition connected to a programmable thermostat comes close in reliable convenience: except for an annual checkup, its operation is entirely automatic. And the only two important commercial applications in which electricity cannot compete with fuel are (as already explained) the production of iron from its ores (where coke or charcoal remains indispensable) and commercial flying, where kerosene remains the only practical choice for jetliners. The capability of experimental unmanned solar-powered airplanes is orders of magnitude behind (Paur 2013). These planes can stay aloft for many hours, and Solar Impulse crossed the United States-but it needed 12,000 photovoltaic (PV) cells to do so, it carried only its pilot, its speed averaged less than 50 km/h, and its wingspan was almost the same as that of a wide-body Boeing 747-400, which can carry more than 500 people at the speed of Mach 0.92.
Space Qualities
The denominator of power densities-a unit of horizontal surface-is seemingly a much simpler measure, but a brief reflection makes it obvious that this ordinary quantity subsumes some very different spatial and functional qualities. For all photosynthetic production in fields, grasslands, and forests the denominator is simply a unit of land covered by cultivated or wild plant species, with a yield expressed in W/m2 rather than (as in agriculture) in t/ ha or (in forestry) in m3/ha. For fossil fuels it is the unit of surface disturbed by extraction (coal mining activities, oil and gas drilling) or otherwise claimed by requisite infrastructures (roads, rights-of-way for pipelines). For thermal electricity-generating plants it is a combination of permanent structures (boiler and generator halls, other buildings, cooling towers) and necessary infrastructures, be they on-site (in coal-fired plants mainly coal storage, switchyard, fly ash depository) or off-site (ROWs for high-voltage transmission lines). For nuclear stations it also includes also the exclusion (safety) zone surrounding a plant.
For renewable energy conversions it gets somewhat more complicated. For the direct combustion of biofuels it is obviously just the area of harvested phytomass: for liquid biofuels it is the cultivated land used to grow plant feedstocks that are converted to ethanol or biodiesel. But the attribution becomes uncertain when crop or logging residues are used to produce cellulosic ethanol: they are by-products of crop cultivation or wood cutting and would be produced even if they were not destined for enzymatic fermentation. For hydroelectricity it is the entire surface of a reservoir created by a dam, but there are several possibilities to consider, including the maximum reservoir design level (which may not be attained for years after dam completion), the average level, determined as the mean of seasonal fluctuations, or the modal level that prevails for most of the year. In any case, for all large projects, land claimed by the dam and associated infrastructure amounts to only a small fraction of the area inundated and periodically partially reexposed by the extreme reservoir levels.
To be comparable with other conversions, the power densities of windgenerated electricity require a transposition from vertical to horizontal planes. The working surfaces of wind turbines are vertical, but the power density of wind generation is calculated as a ratio of electricity produced per unit of horizontal land surface. The power densities for solar PV electricity generation (and also for solar water heating) should be expressed, for conformity's sake, per unit of horizontal surface area, but they are usually given per unit area of actual working surface (on the ground or on a roof), which is fixed at an angle to optimize radiation cap
ture or (a much more expensive solution) which uses automated tracking to maximize the exposure.
As a result, power densities are calculated by using several obviously distinct types of land covers or land uses, and the similarities (often identities) and differences embodied in these rates cut across production modes, infrastructural arrangements, and diverse energy uses. My attempt at a typology of space that becomes the denominator in calculating power densities offers a few fairly simply classifications based on obvious structural and functional attributes: I consider the degree of transformation, project longevity, the likelihood that the land claimed by energy facilities will regain its former function (or at least some of it), and the possibility of concurrent uses of land (or water) that is devoted to the production or conversion of energies.
Space claims by modern energy conversions are hierarchical. On the most intensive end of the spectrum is the obliteration extreme, whereby not only the original plant cover but even the physical appearance of natural surfaces has been entirely erased and replaced by structures that are completely and exclusively devoted to a site's new function, be it extraction, transportation, processing, or conversion. Then comes an extended continuum of impacts of diminishing intensity that eventually joins the other extreme, land that is claimed, owned, or managed by energy industries but whose surfaces have retained their previous (and often entirely undisturbed) soil and plant cover.
The most diverse group in the first category comprises the structures required by modern energy extraction, transportation, and processing and by electricity generation. Every industry has many structures and assemblies of this kind, some highly standardized, others of specific design: buildings that house hoisting machinery, sorting and washing facilities, and storage silos for underground collieries; drill pads, wellheads, gas processing, liquefaction, and regasification plants, compressor stations, refineries, and loading docks for the oil and gas industry; boiler and turbogenerator halls, electrostatic precipitators, desulfurization units, and cooling towers for thermal power plants. In light of the high power densities of fossil fuel combustion and nuclear fission, it is not surprising that most of the structures housing these activities account for only a small fraction of total land claims by thermal power plants.
In addition to buildings that house machinery and processing facilities there are also office buildings, and many energy extraction and conversion sites must have infrastructural components whose construction and operation result in destruction of original soils and plants: these include often extensive areas of impervious surfaces (roads and walkways, large parking lots, storage sheds or open lots with stacked components and parts, railroad yards, switchyards) and frequently even more extensive fuel and waste depositories (large coal yards, oil and gas tanks, ponds storing captured fly ash and sulfates produced by flue-gas desulfurization).
Relatively large shares of areas claimed by energy projects have retained their soils and vegetation but have been fragmented to such an extent that they have lost the capability to provide many of their former ecosystemic services. Satellite images illustrate a wide range of these energy-related fragmentations. The development of oil and gas fields on grasslands or in forests requires the construction of access roads and the drilling of many, often fairly closely spaced, wells; drilling pads and wellheads create pockmarked landscapes (often in a regular, gridlike manner) that are dissected by roads, and further disturbances are created by installing pumps, compressors, and (aboveground or buried) gathering pipelines. Measurements will show that all aboveground structures claim only a small share of the affected area, but while most of the land is undisturbed, its fragmentation excludes any commercial use and degrades its value as habitat for plants and animals. Similar pockmark fragmentation can be also seen (though usually on a smaller scale) with in situ leaching of uranium.
Large-scale PV electricity generation with massed rows of panels fastened to elevated steel supports entails much greater interventions. Support columns, arrayed in regular formations, disturb soil, panels shade the ground, and space must left between the rows for access needed for maintenance and regular cleaning. In contrast, large wind farms can be seen as perhaps the least disturbing example of this fragmentation. As I explain in some detail in the next chapter, turbine siting requires some minimum spacing between adjacent towers, and this distance increases with machine capacity.
Consequently, large modern wind turbines stand hundreds of meters apart, and even when the area of all access roads and transformer stations is added to the space occupied by their concrete foundations, more than 95% or even more than 99% of a wind farm's area is left undisturbed (if somewhat fragmented). In most settings wind farms should not have adverse effects on terrestrial fauna (of course, they are deadly to birds) or on such previous commercial land uses as grazing or cropping. At the same time, assuming that turbines occupy only 1% of land required for the spacing of large turbines would underestimate their overall spatial impact because the noise generated by these large machines requires considerable buffer zones surrounding wind farms. This necessity is no problem in mountains or in regions far from any settlements, but the exclusion of permanent habitation within the buffer zones restricts turbine location in more densely populated European and Asian landscapes.
Many energy projects also include new, deliberately created green belts or buffers designed to provide at least partial visual, noise, and air pollution screens separating them from their surroundings. Such projects are usually site-specific. Some are well planned, others are added as afterthoughts, and yet others are a part of standard construction requirements. For example, India's Ministry of Environment and Forests stipulates that the total green area of the country's numerous new coal-fired power plants is to equal onethird of the total plant land claim, or about 60 ha landscaped and planted for a 1-GW station and almost 120 ha for a large 4-GW plant (CEA 2007).
Finally, many energy production facilities contain land that is entirely undisturbed by their construction; this includes land that was acquired before the project's initiation and that is held in reserve for a possible expansion, land that was deliberately acquired in order to put some distance between a project and the nearest inhabited areas, and, in the case of nuclear power plants, land whose previous uses (unexploited, forestry, cropping, seashore) can continue as long as the land does not include any permanent habitable structures. Some inland water bodies (lakes or reservoirs) associated with thermal power plants also belong to this category: because they are used for water cooling, their area should be counted as a part of the project spatial claim, while most of their former uses are largely unaffected.
As far as the overall impacts of energy infrastructures on land use and land cover are concerned, it should be obvious that they cannot be captured simply by adding up the affected space. Detailed, realistic appraisals require qualitative assessments because energy developments affect spaces ranging from unproductive, barren, hilly surfaces to extensive areas of highly fertile alluvial soils. Similarly, many energy infrastructures have negligible impacts on flora and fauna but others destroy parts of highly biodiverse environments. More often, the ROWs and access roads needed to bring fuels and electricity to distant markets are a major reason for habitat fragmentation, a change that can contribute to loss of biodiversity.
Project Longevities
As for the longevity of specific energy-related land use, it is obvious that only a few structures of modern fossil fuel industries and electricity generation can be called permanent, even if that adjective refers to the span of just a single century. Mining facilities, oil wellheads, refineries, tanker terminals, fuel storage facilities, and thermal electricity-generating plants are designed for service spans of 20-40 years, but many of these enterprises have been around, after upgrading and partial reconstruction, for 50-80 years. What is much more common is that the structures are completely replaced and modernized but the extraction and generation sites remain.
Many European coa
lfields were in production (though, for generations, on a small scale) for more than a century, and some English ones for more than three centuries, before their operation became uneconomical during the closing decades of the twentieth century, not because they ran out of coal (Smil 2010b). California's late nineteenth-century San Joaquin Valley oil fields are still in production more than a century later: Midway-Sunset was discovered in 1894, Kern River in 1899 (SJVG 2012). Even more impressive, Baku's oil fields, where modern oil production began in 1846 (a decade before the United States began drilling in Pennsylvania), are still in relatively vigorous production (Mir-Babayev 2002). And several of the world's largest oil-producing sites are now more than 60 years old: the Saudi al-Ghawar field, by far the world's largest oil field, began oil extraction in 1951, and the Kuwaiti al-Burqan site has been producing oil and gas since 1946 (for more details, see chapter 4).
Many coal-fired electricity-generating stations have occupied the same sites for generations. In 2008 the United States had 10 coal-fired-powered units that were built during the 1920s, including the Sixth Street Generating Station in Cedar Rapids, Iowa, which began to generate electricity in 1921 and was shut down only in 2010 (SourceWatch 2014). There were an additional 110 units that began to work before 1950, and plants of similar age, or even older locations with refurbished generating equipment, can be found in Europe. And most reservoirs created by dams built during the closing decades of the nineteenth century are still in operation, and most of those built since World War II will be around for more than 100 years.
Rheinfelden, on the German and Swiss border, was the first large hydroelectric plant in Europe when it was completed in 1898, and after rebuilding between 2006 and 2011 its four new turbines began producing up to 600 GWh/year (Voith 2011). Other well-known large hydroelectric projects older than 70 years include America's Grand Coulee Dam on the Columbia River (which entered use in 1942), the Hoover Dam on the Colorado River (1936), and Ukraine's Dnieper Hydroelectric Station (1932), and the eventual life span of some of the largest reservoirs will be measured in centuries. That is true even for such impoundments affected by a high rate of silting as Egypt's High Aswan Dam. Negm and co-workers (2010) modeled the reservoir's silting and scouring processes and concluded that the steady value of life span of the dead zone (below the lowest water intake level for the turbines) was 254 years and that of the live zone was 985 years.
