Power Density

by Vaclav Smil

  Hydroelectric power projects with large dams and low capacity factors can be as land-intensive as phytomass production, while some of the world's largest plants and alpine-type stations can generate electricity with power densities up to two orders of magnitude higher. The high land requirements of hydroelectric power generation are best illustrated by the fact that this conversion supplies less than 3% of the world's primary energy but accounts for more than half of all land occupied by the world's energy infrastructure. Wind turbine spacing results in power densities hardly better than the best phytomass production, but when only actual (physical) land claims are counted the rate goes up by an order of magnitude and, depending on location, rivals or surpasses the power densities of PV-based solar generation. Central solar power is not significantly less land-intensive but has much higher capacity factors. No other mode among today's commercial conversions of renewable energy can do better than flat-plate solar collectors, which can supply hot water with power densities between 50 and 100 W,/m2.

  Extreme values for the power densities of coal extraction span three orders of magnitude, from less than 100 W/m2 to more than 1,000 W/m2, but typical rates in large modern surface mines are very similar to those in large oil fields, being at least close to 1,000 W/m2 and commonly two to four times higher. Coal and natural gas extraction make up the largest land claims of fossil fuel-based thermal electricity generation, and while the plants themselves are fairly compact, a great deal of space may be claimed by cooling and captured waste disposal infrastructure. This may reduce the power densities of some coal-fired stations to the same order of magnitude as those of large solar PV-based electricity generation, while the fastestgrowing choice, gas-powered turbines, is highly compact, with power densities of 103 We/m2. The power densities of nuclear generation are broadly comparable to those of coal-fired power plants (102-103 We/m2).

  The power densities of energy production range over five orders of magnitude, from 10-1 W/m2 for liquid biofuels to 104 W/m2 for the world's richest hydrocarbon deposits, but the final energy uses of modern highenergy societies fall mostly between 101 and 102 W/m2 for homes, commercial buildings, industrial enterprises, and densely populated urban areas. This means that modern civilization extracts fuels and generates thermal electricity with power densities that are commonly at least one, usually two, and sometimes three orders of magnitude higher than the power densities of final energy uses in urban areas (where most people now live) and in individual buildings and commercial and industrial establishments (fig. 7.4).

  Fossil fuels to supply urban areas are extracted and delivered with power densities that are higher than the power densities of large cities (1030 W/m2). Thermal electricity is typically generated with power densities that are one and often two orders of magnitude higher (300-3,000 We/m2) than the power densities of electricity use in family houses (10-50 We/m2). Liquid fuels for transportation are produced with power densities that are one to two orders of magnitude higher than the power densities of urban traffic. And even the very high power densities (300-1,000 W/m2 for supermarkets, high-rises, factories, and downtowns) either overlap or are slightly surpassed by the power densities with which electricity and fuels are actually produced and delivered.

  The modern energy system produces concentrated energy flows and then diffuses them through pipelines, railways, and high-voltage transmission lines to final users. As a result, the space claimed by the extraction and conversion of fossil fuels is a small fraction of the ROWs needed to distribute fuels and electricity: American extraction, processing, and conversion of coals and hydrocarbons take up less than 20% of the land that is required for pipeline, railway, and transmission ROWs and occupy less than 0.1% of the country's territory. In contrast, future societies powered solely or largely by renewable energies would rely on an opposite approach by concentrating diffuse energy flows captured with low power densities ranging mostly between 0.2 W/m2 for liquid biofuels to 20 W/m2 for solar PV-based energy. Renewable energy systems would have to bridge gaps of several orders of magnitude between the power densities of energy production and use (fig. 7.5).

  As a result, tomorrow's societies, which will inherit today's housing, commercial, industrial, and transportation infrastructures, will need at least two or three orders of magnitude more space to secure the same flux of useful energy if they are to rely on a mixture of biofuels and water, wind, and solar electricity than they would need with the existing arrangements. This is primarily due to the fact that conversions of renewable energies harness recurrent natural energy flows with low power densities, while the production of fossil fuels, which depletes finite resources whose genesis goes back 106_108 years, proceeds with relatively high power densities This power density gap between fossil and renewable energies leaves nuclear electricity generation as the only commercially proven nonfossil high-power-density alternative. That is why further advances in photovoltaic electricity generation, the renewable conversion with the highest power density, would be particularly welcome.

  Several bold proposals would sever the link between renewable electricity generation and extensive land requirements. They include a variety of ocean energy conversions-exploiting the kinetic energy of waves and currents and the difference in thermal energy between surface and deep waters (Charlier and Finkl 2009; Cruz 2008)-and wind generation by turbines placed within the jet stream (Roberts et al. 2007). None of these proposed alternatives is likely to evolve fast enough to supply a significant share of global energy demand (10%-15% of 2013 use would mean 1.7-2.6 TW). Nor are there any realistic prospects for early, large-scale commercialization of landless PV conversions using giant buoyant PV panels in the stratosphere (StratoSolar 2014) or the Moon-based PV beamed to Earth by microwaves (Girish and Aranya 2012).

  Aggregate Land Claims

  Systematic quantifications of many specific kinds of land use (in addition to such standard data as arable or forested land) are fairly recent; even such important categories as urban land or impervious surface area of the United States have been reliably assessed only during the past 10-15 years. Two obvious disclaimers: aggregates hide many qualitative differences, and they have no claims to high accuracy, though the aim is to do better than just getting the right order of magnitude. I am confident that even for the global totals, and definitely for the US aggregates, even the most uncertain numbers have error margins no greater than ±50%, and some are off by less than ±25%.

  Global Energy System

  Global data for the extraction of fossil fuels, their processing and transportation, and for electricity generation and transmission are fairly reliable (BP 2014; IEA 2014; UN 2014), and I use them, converted to annual powers for the year 2010, together with fairly liberal averages of specific power densities, that is, assuming relatively low rates, in order to err on the high side of aggregate land claims. Moreover, to avoid any appearance of unwarranted accuracy, all itemized results are rounded upward to the nearest 100 km2, and all category totals are rounded to the nearest 1,000 km2. The only category that has been deliberately excluded from this global aggregate are the ROWs of railroads that the coal-carrying trains share with other cargo.

  The extraction of nearly 14 TW of fossil fuel (including the on-site processing of coal and natural gas) claimed about 12,000 km2 (and most likely not less than 10,000 and not more than 20,000 km2), and crude oil refining added about 1,000 km2. Tanker terminals and natural gas liquefaction facilities occupied only about 300 km2. With assumptions of either an average ROW width of 15 m or an average throughput power density of 300 W/m2, the world's refined oil product and natural gas pipelines (whose total length reached about 2 Gm in 2010) preempted other land use on almost 30,000 km2. Aggregates for global electricity are more error-prone. Fossilfueled stations required about 1,500 km2, nuclear power plants added at least 600 km2. Any errors in these values are negligible compared to the estimates for hydroelectric power generation; my best estimate of land claimed by water reservoirs used for electricity generation
is on the order of 100,000-150,000 km2.

  The International Commission on Large Dams' register of dams contains data on nearly 38,000 structures taller than 15 m (ICOLD 2014). About 72% of them are single-purpose dams (50% built for irrigation, 18% for electricity generation). Among the multipurpose dams, irrigation (24%) and flood control (20%) are more important uses than power generation (16%). Consequently, only about 20% of all dams are used solely or primarily to generate electricity, and hence even if we use one of the higher estimates of global reservoir area (about 600,000 km2), hydroelectric power generation would claim at least 120,000 km2. ROWs for high-voltage transmission lines can be estimated in two ways: by assuming that the aggregate line length of about 1 Gm claims on average about 5 ha/km, or by assuming that the transmission of 2.3 TW in 2010 proceeded at an average rate of 40 W/m2. The two totals are respectively 50,000 km2 and about 58,000 km2, a close agreement for estimates of this kind of global aggregates.

  This summation (keeping in mind it is just a fair approximation) means that in 2010 the world's fossil fuel-based energy supply-produced by the extraction of coals and hydrocarbons, their processing and transportation, thermal electricity generation, and the proportional claims of highvoltage transmission ROWs (nearly 70% of the total)-took at least 80,000 and no more than 90,000 km2 of the grand total, the latter total being an area smaller than Portugal or Hungary, and with 13.6 TW of primary fossil energy, its average power density (counting the land devoted to fossil fuelfired electricity generation) was roughly 150-170 W/m2.

  The energy system that dominated the global supply during the past 50 years-fossil fuels, thermal, and hydroelectricity generation, delivering 14.34 TW in 2010-claimed roughly 230,000 km2 (200,000-250,000 km2) of land that is either directly occupied or whose uses are restricted by the ROWs imposed by pipelines or high-voltage transmission lines, which means that it has been operating with an overall power density of about 60 W/m2 (fig. 7.6). The mean of 230,000 km2 is slightly smaller than Romania; the higher value equals about half of Spain's land surface and is less than 0.2% of the Earth's ice-free land. Leaving pipeline and transmission line ROWs aside, the grand total comes to nearly 150,000 km2 (less than half of Poland), of which almost 90% is land flooded by reservoirs.

  In 2010, new renewable energy sources (solar PV, wind, liquid biofuels) contributed just 130 GW, or merely 0.9% of all primary commercial energy. PV-based electricity claimed less than 1,000 km2, and wind turbines were spread over about 40,000 km2 (when assuming an actual generation density of 1 W/m2) and less than 1,000 km2 when only the land occupied by turbine pads and associated infrastructures is counted. In 2010 biofuels (dominated by sugar cane and corn grown for conversion to ethanol) supplied only about twice as much power as wind-79 GW versus 40 GW, that is, an equivalent of less than 1.5% of the global crude oil output-but their land claim was more than 260,000 km2, an area slightly larger than the UK and equal to almost 2% of arable land planted to annual crops (FAO 2014). Even when calculations were performed using minimal wind turbine claims and excluding all transmission ROWs, modern renewables (excluding hydroelectricity) required almost 270,000 km2 to deliver 130 GW, with a mean power density of just 0.5 W/m2. They claimed more land than the global fossil fuel-nuclear-hydro system, which delivered 14.3 TW (110 times as much power) with an average power density of about 60 W/m2.

  Box 7.1

  Aggregate land claim of the global energy system in 2010

  Figure 7.6

  Equivalents of land claims of the global fossil fuel-nuclear-hydro energy system in 2010. Carl De Torres Graphic Design.

  Final comparisons are between the power densities of energy production and the final uses of fuels and electricity. The hierarchy of final uses proceeds from annually averaged power densities of around 3 W/m2 for small, densely populated modern economies to between 10 and 30 W/m2 for many urban areas to more than 100 W/m2 for city downtowns. This leads to a number of revealing conclusions. Perhaps most important, the fossil fuel-nuclear-hydro systems that dominate the energy supply in modern affluent economies have been operating with an overall power density that, depending on climate, population density, and level of industrialization, is mostly two to four times that of the power required by large urban areas.

  Box 7.2

  Terrestrial areas modified by human action, circa 2010*

  Notes: *Data from FAO (2014), Hooke, Martin-Duque, and Pedraza (2012), from previously cited studies of urban areas and impervious surface areas, and from box 7.1.

  Fossil fuels (when transportation and transmission ROW needs are included) generally supply energy with power densities higher than those prevailing in city downtowns, and the only instances in which the power densities of energy use surpass those of common ways of energy production are the energy-intensive industrial processes (often well above 1,000 W/m2) and city blocks consisting of densely packed high-rise buildings (on an annual basis they can go well above 500 W/m2) and during short periods of peak demand (driven by winter heating or summer air conditioning) in downtown cores, where they can go to as much 1,000 W/m2 or even more.

  A table and a graph (fig. 7.7) make it easy to appreciate how small is the absolute claim of the modern energy infrastructure in comparison with other human activities that have modified large parts of roughly 130 Tm2 of the Earth's ice-free surface:

  Figure 7.7

  Comparisons of global land use in 2010 (land use in 1,000 km2). Carl De Torres Graphic Design.

  US Energy System

  The land requirements of America's energy system can be estimated more accurately than the total claimed by the global energy production. I use supply specifics from official statistics (primarily the USEIA) and, once again, assume rather conservative means of power densities, and round the totals to the nearest 50 km2. In 2010, fossil fuel extraction claimed roughly 6,000 km2, fuel processing (mainly refining) needed less than 500 km2, and the ROWs of long-distance hydrocarbon pipelines preempted all but a few concurrent land uses on about 15,000 km2. In 2010 about 40% of all mass transported by railroads was coal (AAR 2013), but as only a small minority of lines are dedicated solely to coal shipments the only imperfect way to proceed is to attribute roughly 40% of ROWs of high-density "A" track (approximately 3,100 km2: 30-m ROW for 104,000 km) to coal; that would add about 1,200 km2. Thermal electricity generation (including the domestic part of the uranium fuel cycle) occupied less than 600 km2, reservoirs for hydro generation covered at least 17,000 km2, and transmission ROWs claimed nearly 16,000 km2.

  The entire fossil fuel-nuclear-hydro energy system thus required almost 53,000 km2 (roughly 0.5% of the US territory, an area roughly half the size of Virginia or Tennessee), of which some 30,000 km2 (5 5%) were ROWs and nearly a third water reservoirs, while the fossil fuel-based supply (including pipeline and railway ROWs) claimed only about 19,000 km2. For the sake of completeness, the nationwide fossil fuel-based total should be enlarged by the area of abandoned wasteland created by the surface extraction of bituminous coal that has yet to be reclaimed. During the first decade of the twenty-first century the difference between new permits for surface coal mining and bond releases (issued after the completion of planned reclamation) was on the order of 250 km2/year, and that would imply the growth of unreclaimed land debt by more than 2,000 km2/decade.

  But that is both too much and too little: too little because in forested areas (particularly in Appalachian mountaintop removal) even the best reclamation effort cannot recreate the original plant composition, and an ecosystem that would closely resemble it may get reestablished only after decades, even centuries; too much because in some locations (particularly in grassy regions), unreclaimed areas may naturally revegetate in a matter of years without any deliberate reclamation effort. In any case, it should be remembered that the overall land claim attributable to the US coal extraction is definitely much higher than my approximation, and, choosing to err on the high side, I add 2,000 km2.

  Box 7.3

  Aggregate land claim of
the US energy system in 2010

  Note: *Total power includes the burning of wood waste.

  Hydroelectricity is the traditional source of renewable energy, with by far the largest land claim among mature energy conversion, while the relatively small contribution of geothermal electricity (less than 2 GW in 2010) requires less than 50 km2, and electricity generated by the combustion of woody phytomass (about 5 GW) comes mostly from burning logging residues and does not create any additional space claims and hence is not included in the aggregate count. Of the three kinds of new renewable energy supply, PV-based solar electricity generation was still minuscule in 2010 (accounting for less than 50 MW), but wind turbines contributed 10.8 GW, equal to slightly more than a third of hydroelectric generation. They were spread over an area of nearly 11,000 km2, but the land actually occupied (mostly by tower pads and access roads) was only on the order of 200 km2 (less than three Manhattans). But in 2010 nearly 29% of corn and sorghum harvests were used to produce ethanol, and their cultivation claimed about 124,000 km2, an area larger than Pennsylvania and three times as large as all land (including all ROWs) claimed by the entire US fossil fuel-nuclear-hydro energy system.

  Once again, these approximate nationwide summations lead to interesting insights. Coal extraction, thanks to highly productive western surface mines, claims less than 1,000 km2, and more land is occupied by ROWs of railroad coal transportation when they are apportioned according to coal's share in the annual mass of rail shipments on high-density lines; unreclaimed coal mining land occupies an even larger area that is impossible to estimate with a fair accuracy. Mainly because of a very large number of old, poorly productive oil wells, the average power density of US crude oil extraction is an order of magnitude lower than the country's coal production, and ROWs for oil and gas pipelines are 2.5 times as large as the land taken by the extraction of all hydrocarbons.