Power Density

by Vaclav Smil

  On the other hand, many small hydroelectric stations built in China's countryside since the late 1950s as a part of the Maoist quest for inexpensive, mass labor-based solutions to the country's energy shortages silted very rapidly and were abandoned or dismantled after only a few years of unreliable electricity generation (Smil 2004a). Similarly, large numbers of surface or shallow underground mines opened by Chinese peasants (legally and illegally) since the 1950s operated for just a few years before less dangerously produced fuel became available. And the latest extraction technique produces short-lived wells: the average life expectancy of wells drilled for hydraulic fracturing of shales will be no more than 15-20 years (see chapter 4).

  Taking these different longevities into account when calculating power densities of specific processes is not easy. The only possible way is to make serial assumptions regarding their ultimate life spans-and such assumptions may have large errors. The redrilling of old reservoirs and the secondary recovery of previously unobtainable oil have extended the lifetimes of many oil fields to three, four, or even five generations, while the initial expectations may have been for just 20-30 years of production. Horizontal drilling and hydraulic fracturing created commercial reserves out of resources that two decades earlier were in the uneconomical category.

  In contrast, the availability of cheaper hydrocarbons led to the abandonment of many coalfields (in Europe, Japan, Taiwan) long before they reached their expected life spans. In the Dutch case, that unexpected shift away from domestically produced and imported coal to natural gas from the newly discovered Groningen reservoir took place in just a few years (Smil 2010b). And because we do not have enough accumulated experience with new conversion techniques, assumptions have to be made about the longevity of different forms of solar energy capture and the durability of wind turbines. The standard assumptions are for 20-30 years of service, but the first designs of large commercial wind turbines, introduced during the late 1980s, and even many better models of the 1990s were retired after less than a decade of operation, and the sites were repowered with larger machines.

  But these uncertainties do not make any difference for land claims made only by fixed structures. For example, a natural gas liquefaction plant that occupies a relatively small area of coastal (or reclaimed) land will process the gas with the same power density for as long as its annual capacity remains constant. And when calculating the power densities of processes that entail both fixed and incremental land claims, it must be made clear to what specific operations the latter category refers. For example, the fixed land claim of a coal-fired power plant with 1 GW of installed capacity remains constant as long as its operation is not expanded or curtailed. And as long as that plant burns only coal from an adjacent surface mine that taps a seam with reserves good for decades, then its annual incremental (variable) land claims (for coal extraction and the disposal of captured ash and desulfurization slurry) would hardly change and its operating power density, calculated on the basis of annual output, would remain nearly identical.

  Land Restoration

  Similarly, it is very difficult to offer useful guidance about the probability with which a specific land cover destroyed or altered by energy development will return (fully or partially) to its predevelopment use. Surface coal mining, one of the most common landscape-altering activities, illustrates the range of possible outcomes. On the one hand, there are some relatively rapid recultivation efforts: land disturbed by surface coal mining may be rather promptly reshaped, re-covered with topsoil, and replanted, and within a decade or two after the seam was extracted the landscape may assume a pleasing aspect composed of a combination of water reservoir (flooding the deepest land cuts) and slopes and flat surfaces replanted with grass, shrubs, or trees, preferably in the more natural, variable, multispecies clumplike fashion rather than (as used to be done) in tree or shrub monocultures set out as regular rows of saplings.

  An increasingly common way of restoring land and waters claimed by energy projects is by dismantling dams, mostly to improve the upstream access of anadromous fish (Whitelaw and MacMullan 2002). In the United States, the first notable removal dates to 1973 (the Lewiston Dam on the South Fork Clearwater River in Idaho), and recently accomplished demolitions include that of the 6.6-m-tall Embrey Dam (on the Rappahannock River in Virginia), in 2004, and the 33-m-tall Elwha Dam (on the Elwha River in Washington state), the largest dam removal project so far, which was completed in 2012 (NPS 2013). In Europe the most important recent decision was to remove the Poutes dam on the Allier (a main tributary of the Loire) to allow passage of Atlantic salmon (RiverNet 2012).

  In contrast to some notable examples of prompt land use restoration there are many old, derelict energy landscapes (some going back for generations) that have yet to be returned to agriculture, forestry, or recreational use. These areas include old coal-mining heaps, land disturbed by surface mining that was never even leveled or shaped into surfaces suitable for replanting; abandoned oil fields with derelict machinery, burst pipes, and oil pools; and crumbling structures of shut-down refineries and coal-fired power plants. In the United States, there are estimates of up to 500,000 abandoned mines (including all ore and nonmetallic minerals), some 4,000 km2 of unreclaimed coal mine land, and a data portal showing many details by state (Abandoned Mine Lands Portal 2013).

  In between the two extremes of prompt landscape restoration and longterm abandonment is the increasingly common creation of brownfields. After the end of their useful life, many energy sites, much like industrial facilities such as defunct iron and steel mills or chemical and manufacturing (mainly textile and automobile) plants, are simply stripped of all structures and, if need be, the soils are appropriately decontaminated. While some of these newly created brownfields may eventually be replanted, most of them are not renaturalized but turned into another kind of commercial development, most often warehouses, commercial spaces, and apartment blocks.

  Concurrent Uses of Space

  Space, rather than land, is a more appropriate term here because the two most important examples of concurrent uses do not have any direct footprints on land. PV panels installed on the rooftops of houses and commercial and industrial buildings make no new land claims, and the electricity they supply to the grid (in excess of the needs of the buildings underneath them), by using existing power lines, also helps to reduce additional need for land devoted to transmission from new central generating plants that would have to be built in the absence of new rooftop PV capacities. And by far the most important example of concurrent space uses by the traditionally most important renewable energy conversion does not refer to land but to reservoirs impounded by large dams.

  In calculating the power densities of hydroelectric generation, reservoirs' entire areas are used as a denominator, but they also supply drinking and industrial water for cities and irrigation for farming, provide flood control (whose economic benefits, namely, preventing or reducing major downstream inundations, may be greater than the value of the electricity generated), improve upstream navigation, and offer new opportunities for recreation as well as for freshwater aquaculture, one of the fastest-expanding modes of modern food production. On the other hand, the negative environmental impacts associated with large reservoirs may actually extend their spatial impact.

  When new reservoirs flood arable land and established settlements and industries, new areas must be reclaimed to maintain food production and to accommodate resettled populations, while irrigation may result in excessive salinization and abandonment of the affected land. These additions and losses are directly or indirectly attributable to the creation of large reservoirs, but we have no satisfactory way to weigh them against the just noted multipurpose uses and to come up with a net adjustment for a specific reservoir area (which in some climates can show significant inter - and intra-annual fluctuations).

  Uses that have no effect on the original design of a reservoir can be ignored: swimming, boating, or a limited diversion of drinking wate
r (amounting to a fraction of natural evaporation from a reservoir) would be in that category. But if the volume of the reservoir was deliberately designed to store water for both electricity generation and seasonal irrigation, or if a reservoir's exclusion area has been made larger to accommodate periodic floods, then it would be logical to apply appropriate correction factors and not to use a reservoir's entire area as a denominator in calculating power densities.

  Applying spatial discounts for other uses would raise the average power densities of electricity generation, but such a correction would be relatively easy to do (although still arguable) only in the case of clear and measurable dual use, for example when data on water withdrawals for irrigation could be compared with the water volume used for electricity generation. In the case of multiple and nonconsumptive uses (recreation, aquaculture), it would be arbitrary. Moreover, even if some acceptable generalized way to correct for non-electricity-generating uses could be found, the power density's order of magnitude would not change; it would undergo merely a fractional adjustment, and hence it might be argued that it is sensible to ignore multiple reservoir uses.

  Multiple uses may also mean that the loss of the land claimed by a reservoir may be more than compensated for by the creation of new productive land. Egypt's High Aswan Dam is a perfect illustration of this reality. The dam created a large lake that flooded mostly desert wasteland in southern Egypt and northern Sudan and caused the loss of only a narrow strip of previously cultivated land along the river that had benefited from silt deposition by annual flooding. At the same time, irrigation water from the reservoir enabled Egypt to expand its cultivated area in the Delta and along the Nile Valley and to convert many previously single-cropped fields to double - and triple-cropping: the overall gain of farmland amounted to about 50%, and more food production was added by the reservoir's commercial fishery (Abu-Zeid and el-Shibini 1997). On the other hand, silt deposition in the reservoir cut off the downstream farmland from its annual addition of nitrogen-rich sediments, a loss that has to be compensated for by applications of synthetic nitrogenous fertilizers.

  Perhaps the most commonly cited example of concurrent land use that involves new renewable energy conversions (and that is often stressed by their proponents) is the case of wind-generated electricity: concrete turbine foundations occupy only small areas (on the order of 15-25 m2) and, except for permanent access roads, the surrounding landscape can be used for grazing or crop cultivation. Moreover, depending on the crops grown and wind-site royalties, a farmer may earn more from the latter than from the harvest of the former. As Lovins (2011b, 2) put it, "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, -99% of its area remains available to drive, park, and walk in."

  That is a correct statement-but it is also a complete misunderstanding of the power density concept. Power density informs us about the energy flux that can be usefully derived from a given area by a particular conversion, regardless of how close or how far apart the individual converting or extracting facilities may be, and regardless of the commercial activity (if any) taking place in most of the area that is not occupied by structures or infrastructures indispensable for extraction or conversion. Consequently, using a wind farm's entire area in the power density's denominator is not conceptually different from using the gross land claims of other energyproducing activities, even though the actual disruptive footprint of their infrastructures adds up to only a very small share of the overall claim. For example, in terms of the ratio of occupied to unaffected land, it is very similar to using the entire area of an oil field, where only a fraction of land is taken up by well pads and access roads and most of it is planted to crops or is grazed by cows: many American landscapes, from California to Texas, offer numerous examples of this common reality.

  Growing crops or grazing cows has nothing to do with the fact that even in not very sunny mid-latitudes, PV cells will always produce electricity with a higher power density than wind turbines, or that even old oil fields dotted with pumps will have a higher power density of fuel extraction than converting soybean harvests to biodiesel. And the power densities of rooftop PV panels obviously do not tell us how much additional land they occupy (none at all) but what the existing or potential electricity-generating capability of such installations is, particularly when those rates are compared to average electricity demand rates of buildings underneath the panels.

  Boundaries and Technical Advances

  Two critical considerations apply to all power densities: the boundary problem in calculating the rates, and technical advances that extend the realm of possible achievements. The first reality creates numerous uncertainties: because there are no binding rules for the inclusion or exclusion of analyzed components, different choices of system boundaries will result in substantially different outcomes. Again, this is not a challenge unique to power densities for it is repeatedly encountered in analyses of energy costs (embedded energies). They can be narrowly limited to a specific process (smelting steel) or they can include the energy costs of as many preceding steps as is practicable (mining and transporting iron ore, producing coke and natural gas, mining limestone, producing temperature-resistant furnace linings, etc.).

  The resulting differences in final energy costs are commonly two - or threefold. For example, the energy costs based on an input-output analysis of the steel industry were twice as high as those calculated by basic process analysis (Lenzen and Dey 2000), and similar comparisons showed the energy cost of structural iron and plywood used in Swedish apartment buildings to be three times as high when input-output data were used (Lenzen and Treloar 2003). Some calculations presented in chapters 3-5 will show that differently set analytical boundaries can result in even greater disparities for power densities, some easily as large as an order of magnitude.

  This is particularly true when areas of requisite distribution (transportation and transmission) infrastructures of the fossil fuel industries and of all forms of electricity generation are added to spaces claimed by extraction and power plant facilities. Surfaces that are actually transformed by highvoltage transmission lines, the most extensive of these distribution infrastructures, are minimal: the foundations of transmission towers occupy only a small fraction of the ROWs set aside for the lines, and the land underneath the lines can be used as before if it was grazed or cropped, or it can be converted to plant nurseries or Christmas tree plantations. Similarly, land above buried pipelines can be used for grazing or can be planted to seasonal crops; in Canada, although not in the United States, even trees are allowed in a pipeline's ROW as long as they are less than 1.8 m tall and are at least 1 m away from the line, and the only land whose use has been grossly altered is for pumping and compressor stations placed along the line.

  Of course, as already noted, similar situation arises with wind turbines, and there is no accepted norm to deal with this large disparity between transformative (footprint) and spacing (ROW) land requirements. Counting only the former clearly underestimates the overall impact because land use within ROW corridors (or within spaces occupied by large wind farms) is obviously restricted, and using the overall spacing or ROW power density conveys the fundamental limits on the capacity of machines, lines, or pipes that can be accommodated within a unit of land. At the same time, an ROW claim is not obviously equivalent to such (if only temporarily) destructive transformations as surface coal mining or the impervious surfaces of many energy facilities.

  Perhaps the best way to deal with the inescapable boundary problem is to follow these three basic rules: to make clear what goes into a particular power density calculation; where appropriate, to present alternative values of power densities within narrower and wider (and always properly explained) boundaries; and to make sure that comparisons of different modes of energy extraction, conversion, and use are done for accounts that have been prepared in identical, or at least very similar, ways. I will try to follo
w these precepts in this book. Finally, a few comments about the evolving value of the power densities of energy production and use are in order.

  The power densities of renewable energy flows and energy densities of fossil fuel resources are fixed. The first category is circumscribed by the amount of solar radiation reaching the Earth and its subsequent transformation into wind (via differential heating of the planet's surfaces), flowing water (via evaporation), and phytomass (via photosynthesis). Stores of fossil fuels present a fixed outcome of long underground transformations involving elevated temperatures and pressures (in theory, these resources are not finite, but their rate of formation is negligible compared to our current rates of their extraction). In contrast, the power densities with which these resources are converted to useful energies are-within the natural and thermodynamic constraints-anthropogenic, and as such they evolve, driven by a combination of technical and managerial advances.

  The conversion of phytomass to heat and light began hundreds of thousands year ago (dates for the earliest controlled use of fire remain uncertain), the conversion of running water and wind (by waterwheels and windmills) dates, respectively, to more than two millennia and about 1,200 years back (Smil 1984). All of these very low power densities and rates began to rise only with the advent of modern high-yield cropping and forestry and with the invention and commercialization of water turbines (starting in 1830s) and modern wind turbines (in the 1980s). The direct conversion of solar radiation was impossible until the first deployment of PV cells on Earth's satellites in the 1960s, and the power densities of solar electricity generation have been rising with improving cell efficiencies.