Power Density

by Vaclav Smil

The world's largest coal-fired power plant, with a capacity of 5.824 GWe, is on about 300 ha of reclaimed land on the eastern shore of the Taiwan Strait near Taiwan's city of Taichung (GIBSIN Engineers 2006). The station burns coal imported from several countries (mainly Australia), and with an annual generation of 42 TWh its power density (excluding that consumed in coal extraction) is about 1,600 We/m2. Two of America's three largest coal-fired electricity-generating plants belong to the Southern Company, headquartered in Atlanta: the Robert W.Scherer (four units, 3.52 GW), located on the northern shore of Lake Juliette southeast of Atlanta, and Bowen (four units, 3.499 GW), located in Bartow County, Georgia (fig. 5.1). Scherer is a sprawling installation with a large oval-shaped coal storage yard (36 ha), an ash-settling pond of 120 ha, and an ash disposal pond of 300 ha (designed to last for the plant's life span of some 50 years), and the plant's total operating area covers about 14 km2 (Georgia Power 2013).

  With an average load factor of 75%, the plant generates 23.1 TWh/year, and the power density of its operation is about 190 W/m2. But the entire property, including part of Lake Juliette, is 34 km2. The Bowen plant claims about 3.7 km2 (with main buildings, cooling towers, the coal yard, and the transforming area accounting for only about 10% of the total area), and its annual generation of 22.6 TWh translates into an overall power density of nearly 700 We/m2 The third largest US coal-fired plant is Duke Energy's Gibson, in Indiana (five units, 3.34 GW). The plant owns about 16 km2 of land, of which 12 km2 is a man-made lake used as the plant's cooling pond; the plant itself and its associated coal storage and pollution-control infrastructures take more than 2.5 km2, resulting in a power density of close to 1,000 We/m2. The Tennessee Valley Authority's Bull Run station (with a single 900-MWe unit generating 6 GWh/year) uses once-through cooling and occupies only 65 ha (for a power density of about 1,050 W/m2).

  Belchat6w, Poland's and Europe's largest coal-fired plant (5.298 GWe, 28 TWh annual generation), burns lignite from an adjacent surface mine, and its rather large area results in a generation power density of about 500 We/m2; the entire mining-generation system rates less than 250 We/m2. Drax, England's largest thermal station, has a capacity of 3.96 GW, and in 2012 it generated 27.1 TWh; that implies a load factor of 78%, or 3.09 GW (Drax Group 2013). The plant owns 750 ha, but its structures and storage facilities cover less than 200 ha (with the rest of the site under trees and meadows), resulting in an actual generation power density (exclusive of coal mining) of about 1,500 We/m2.

  Janschwalde, Germany's largest lignite-fueled station, near Peitz in Brandenburg (very close to the Polish border), occupies about 220 ha, has an installed capacity of 3 GWe (six 500 MWe units) and generates 22 TWh/year (Vatenfall 2013). The site's power density is thus about 1,100 We/m2. And the hard coal-fired Westphalen plant near Hamm became the world's most efficient producer of its kind with the addition of two 800-MW units that replaced two smaller units from the 1960s (RWE 2014); the new units claimed only 11 ha, the entire plant occupies about 70 ha, and it has a power density of more than 2,500 Wi/m2 (fig. 5.2).

  Figure 5.1

  Scherer (top) and Bowen power plants, Georgia, United States. Both photos available at Wikimedia.

  The two best conclusions are that when the calculations include only the area actually occupied by plant's structures, then most coal-fired power plants generate electricity with power densities in excess of 1,000 We/m2, and that the inclusion of coal extraction lowers those rate by widely differing margins. Power densities of the entire coal-to-electricity sequence can stay well above 1,000 We/m2 but can be reduced by an order of magnitude. For comparison, Mielke's (1977) average US rate (including that consumed in coal extraction) for uncontrolled generation was nearly 1,000 We/m2, 2,700 We/m2 for emission-controlled western coal-based plants and less than 700 W/m2 for plants burning eastern coal with air pollution controls.

  Hydrocarbons in Electricity Generation

  The consumption of hydrocarbons in electricity generation has seen two shifts: away from liquid fuels (mostly fuel oil in large power plants and diesel fuel in smaller generators) to natural gas, and from gas combustion in large boilers to the widespread deployment of gas turbines (Smil 2010b). As world oil prices rose during the two consecutive OPEC-driven energy "crises" (1973-1974 and 1979-1981), liquid fuels became too expensive to be used for electricity generation, and most of the countries either shut down them down or converted them to burning either coal or natural gas. US statistics show the rapidity of this shift: in 1980, 11% of electricity came from petroleum fuels; a decade later the share was down to 4%, and in 2012 it was only about 0.5% (USEIA 1993, 2014d).

  Natural gas has made up the difference. In 1990 US utilities produced only 9% of electricity from natural gas, but a decade later the share had nearly doubled, to 17%, and in 2012 it reached 30%. Not surprisingly, the world's largest operating station burning fuel oil is in Saudi Arabia, but Japan's Kashima (4.4 GWe burning heavy oil and crude oil) surpasses Russia's Surgut-1 plant in west-central Siberia (3.268 GWe). The Saudi Shoaiba plant (installed capacity of 5.6 GWe), on the Red Sea about 100 km south of Jeddah, is also the kingdom's largest plant; it now operates 14 generating units and combines electricity generation with desalination (Alstom 2013). The plant itself occupies about 250 ha, implying a power density of about 2,200 Wi/m2.

  Figure 5.2

  Westphalen power station, Germany. © Hans Blossey/imagebroker/Corbis.

  Kashima, located on about 60 ha of reclaimed land on the Pacific coast northeast of Tokyo, rates more than 7,000 W;/m2, and Surgut-1 rates about 5,400 W;/m2. One of the most unusual oil-fired stations was the Chavalon plant at Vouvry, Switzerland (Chavalon 2013): it had a capacity of 300 MW;, it operated between 1965 and 1999, and its compact designoccupying just 3.6 ha on a small plateau on a steep slope above the Rhone valley just south of the eastern end of Lac Leman-resulted in a power density of more than 8,000 W;/m2 (its replacement, a gas-fired combined-cycle station of 400 MWe, was approved for the site in 2010).

  Natural Gas in Electricity Generation

  The typical large central power plant fueled by natural gas and supplied by a pipeline claims much less land than other fossil-fueled operations because it has only minimal emergency fuel storage and no need for fly ash or SO2 capture. When new construction of such stations was fairly common in the United States, Mielke's (1977) data put their average power density at about 1,800 We/m2, and six years later the USDOE (1983) estimated that an 800-MWe plant with a 55% load factor would occupy about 36 ha, implying an operating power density of about 1,200 We/m2. Currently the largest group of high-capacity (more than 2 GW,) natural gas-powered stations is in Japan, a country that imports 100% of its natural gas, a choice dictated by the quest for high air quality in Japan's densely populated urban regions.

  Japan has nearly 20 liquefied natural gas (LNG)-based high-capacity power plants, usually located on reclaimed coastal land and close to an LNG terminal, with boilers and generator halls clustered around two or more tall chimneys and with adjacent gas storage tanks and security land buffers. The largest plant is TEPCO's Futtsu, in Chiba prefecture, rated at 5.04 GWe, with an adjacent LNG terminal and ten storage tanks on the eastern shore of Tokyo Bay (fig. 5.3). Japan's second largest natural gas-fired station is Kawagoe (4.8 GWe, about 30 TWh/year, with six large gas storage tanks), in Mie prefecture (Chubu Electric Power 2013). These are highly compact facilities: Kawagoe occupies only about 75 ha, Futtsu about 125 ha, and hence their operating power densities are 2,800-4,500 We/m2.

  Other countries with high-capacity natural gas-fired stations include Australia, Malaysia, and Russia (all burning abundant domestic gas) and China, Taiwan, and South Korea (relying on imported LNG). In 2013 Russia's Surgut-2, in Khanty-Mansyisk region of west-central Siberia, was the world's largest natural gas-fired station (installed capacity of 5.597 GWe1 annual generation of roughly 40 TWh); with an area of about 80 ha its power density was approximately 6,000 We/m2. Ravenswood in Long Island City (Queens, New York) burns
natural gas, fuel oil, and kerosene to power units with a total capacity of 2.48 GW. The plant is situated on a small (just 12 ha) rectangular lot just south of the Roosevelt Island Bridge on the East River, and its power density is about 20,000 W;/m2.

  Figure 5.3

  Futtsu power station and LNG terminal, Japan. Tokyo Electric Power Company (TEPCO).

  Gas combustion in large boilers of central power plants has been eclipsed by burning in gas turbines, efficient and flexible energy converters that now come in capacities ranging from 1 MW to 375 MW, the record rating by 2013, for the Siemens SGT5-8000H model installed in Irsching near Ingolstadt (Siemens 2013c). The Swiss Brown Boveri Corporation pioneered the use of gas turbines just before World War II, but the widespread adoption of these machines came only during the 1960s, especially after the great November blackout in the northeastern United States in November 1965 showed the need for swiftly deployable generators (Smil 2010b). The expansion was also helped by the adoption of a combined gas cycle: hot exhaust gases leaving the turbine are used to produce steam for an attached steam turbine, and the overall conversion efficiency is as high as 60%, a rate unrivaled by other modes of thermal electricity generation.

  Aeroderivative turbines (jet engines adapted for stationary uses), made by GE, Rolls Royce, and Pratt & Whitney (P&W), are also very efficient (about 40%), and some are available as fully assembled units on trailers. A modified P&W FT8 jet engine with a capacity of 25 MWe fits on two trailers and can generate eight hours after arrival (PW Powersystems 2013).The turbine itself occupies no more than 140 m2, and even with its control trailer, access roads, fuel and electricity connections, and a safety perimeter buffer it claims only 600 m2 (a 40-m x 15-m rectangle), implying a power density of nearly 42,000 W;/m2. A larger unit, the 60-MW SwiftPac, is placed on concrete foundations, claims less than 700 m2 (85,000 W;/m2), and could be ready to generate in just three weeks.

  Gas turbines can thus easily be accommodated on small lots in industrial areas. The Delta Energy Center in Pittsburg, California, demonstrates this flexibility (Calpine Corporation 2013a). Calpine Corporation located the plant on an undeveloped 8-ha lot at the Dow Chemical Company facility, and with an 835-MWe combined-cycle capacity its rated power density is nearly 10,500 W;/m2. Gas turbines can be also placed on land that is already occupied by large power plants, a choice that eliminates contentious application and approval processes for new plant sites. The large (1.36-GWe) English Didcot-B plant in Oxfordshire is a perfect example of this option. It was built between 1994 and 1997 within a larger preexisting site of Didcot-A, a 2-GWe coal-fired station whose construction (completed in 1968) had met with a great deal of opposition. The gas-fired plant, with more than two-thirds of the original coal plant's capacity, now takes up less than 10% of the entire site.

  Nuclear Generation

  I called the worldwide achievement in post-1956 nuclear electricity generation a successful failure (Smil 2003)-successful because the fission supplied about 13% of the world's electricity by 2013, or more than was contributed at that time by hydroelectricity after 130 years of developing water turbines and building large dams; a failure because in the UK, and even more so in the United States, the two countries that pioneered its introduction, nuclear generation has fallen far short of the early expectations that it would become the dominant (if not the only) way of electricity supply before the year 2000 (Smil 2010a). Will this great pause-never entirely global, because Japan kept building until the Fukushima disaster of March 2011, and construction continues and expansion plans remain in place for China, India, and Russia-be followed by a long-awaited nuclear renaissance?

  Whatever the outcome, the high power density of nuclear fission has been one of its major appeals. The conversion of nuclear energy to heat proceeds at power densities (used throughout this book to mean power per unit area, not as used by nuclear engineers to mean power per unit of reactor volume) ranging from 50 to 300 MW,/m2 of a reactor's footprint, that is, up to an order of magnitude higher than the power densities for boilers burning fossil fuels. In addition, nuclear stations do not require any extensive fuel-receiving and fuel storage facilities, they have no need for air pollution controls and for land set aside for the storage of captured waste products, and radioactive wastes stored temporarily on-site occupy only small areas. Well-known breaches of reactor containments (Chornobyl in 1985, Fukushima in 2011) should not obscure the fact that the operation record of nuclear stations in the two countries with the largest number of commercial reactors, the United States and France, has been excellent.

  In common with all other thermal plants, nuclear stations have machine halls (housing turbogenerators, steam condensers, waste heat rejection systems, and requisite pumps) and auxiliary buildings (containing water and waste treatment facilities, fuel and maintenance stores, and administrative areas), and road and rail access. Similarly, some nuclear plants use oncethrough cooling, but many have tall cooling towers, and all of them must have appropriately designed switchyards needed to step up the voltage before transmitting the generated electricity to a national grid. And, like other thermal plants, nuclear stations vary greatly in their overall land claims.

  In 2012 almost 60% of America's total nuclear capacity (total of 118 reactors in 74 plants) was located on sites of 200-800 ha, with a modal area of 200-400 ha, of which 20-40 ha were actually disturbed during the plant's construction (USNRC 2012a). But the extreme land claims range over two orders of magnitude, with the largest areas taken up by cooling reservoirs, artificial lakes, or extended buffer areas. California's San Onofre (2.586 GW in three reactors on the Pacific coast in Oceanside, cooled by ocean water) occupies just 34 ha (fig. 5.4), while the William B.McGuire station near Charlotte in North Carolina (2.36 GW) is situated on a site of 12,000 ha.

  Figure 5.4

  San Onofre nuclear power plant, California. © Corbis.

  Power Densities of Nuclear Generation

  San Onofre's power density prorates to about 7,600 We/m2 in terms of installed capacity, and densities between 2,000 and 4,000 We/m2 are common. At the other end of the size spectrum there are power densities of less than 100 We/m2: McGuire rates at just 20 We/m2, Wolf Creek in Kansas (1.17 GW, 3,937 ha, most of it a large cooling lake) at 30 We/m2. Insofar as most of the world's reactors belong to just two dominant types (the most common being the pressurized water reactor, followed by the boiling water reactor) and their installed capacities range mostly between 400 and 1,200 MWe1 it is hardly surprising that the power densities of nuclear electricity stations in other countries are very similar.

  The world's largest station, Japan's Kashiwazaki-Kariwa (like all of the country's plants, it was closed in 2011 after the Fukushima disaster, but some of its reactors are to be restarted) had an originally installed capacity of 7.965 GW in seven reactors located within a 420-ha area along the coast of the Sea of Japan in Niigata Prefecture (Sakai, Suehiro, and Tani 2009). That would be about 1,900 We/m2, but a large part of that area is the surrounding wooded buffer zone, and there is also a broad green strip between two groups of reactors. Plant structures and infrastructures take up less than 200 ha, and the plant's highest output of 60 TWh in 1999 (before it began to experience operation problems) implied an operating power density of at least 3,500 We/m2.

  In contrast, the now infamous Fukushima Daiichi-hit by the March 2011 tsunami and the subsequent crippling of four of the six reactors and the immediate release of radioactivity into the atmosphere and later also into the Pacific Ocean-is located on a much larger piece of costal property; with about 350 ha and an installed capacity of 4.698 GW, its predisaster power density was 1,300 W;/m2. The plant's twin, Fukushima Daini, 12 km south of the crippled station, has a capacity of 4.4 GW in four reactors on a much smaller site of 147 ha, rating about 2,900 W;/m2.

  Gravelines, the largest of 21 French stations and the world's fifth largest nuclear plant (with 5.7 GW in six 900-MW reactors, producing about 38.5 TWh/year) occupies only about 90 ha just west of Dunkerque on t
he Pas de Calais coast; its power density is thus about 6,300 W;/m2. Cruas on the Rhone in southern France (3.6 GW in four 900-MW units) claims 148 ha, for an installed power density of some 2,400 We/m2. My last example is the Swiss Beznau, in 2013 the world's oldest operating PWR plant, with two 365-MW reactors: the plant occupies only 20 ha on an island in the Aarr river, rating about 3,600 W;/m2.

  Land occupied by the structures and infrastructures of most nuclear stations translates into installed power densities of 2,000-4,000 We/m2, and the inclusion of immediately adjacent buffer zones (which do not prevent agriculture or forestry but exclude any permanent habitation) reduces that rate by 40%-60%. And, unlike other thermal stations-some of which, like New York's Ravenswood in Queens, are not only within cities but relatively close to city centers-nuclear station are preferentially and deliberately located in areas of lower population density. In the United States more than half of all nuclear plant sites have an 80-km population density of less than 80 people/km2, and more than 80% have an 80-km density of less than 200 people/km2 (USNRC 2012a).

  As must be expected, land claims for the uranium cycle will significantly lower the densities of the entire fuel generation-disposal sequence. Commercially extracted uranium ores contain as little as 0.15% and as much 20% of the heaviest of all metals (WNA 2010). Their mining is followed by milling (on-site or in a nearby facility) to produce U308 concentrate (yellowcake), which contains more than 80% uranium: about 200 t of the concentrate are needed to produce the fissionable fuel required to run a 1-GWe plant for a year. U308 is refined to UO2 (this unenriched oxide fuels Canada's CANDU reactors) and is then converted to UF6. This gas undergoes isotopic separation (centrifugal or by gaseous diffusion) to produce enriched UO2, and this oxide is formed into ceramic pellets and encased in fuel rods.

  The typical progression (nuclear fuel chain material balance) from mining uranium ore to fabricating the enriched fuel into reactor rods results in a mass reduction of two orders of magnitude. For electricity generation in a 1-GWe plant operating at full capacity and consuming fuel derived from ore with 0.2% uranium by weight, the sequence is as follows (IAEA 2009): about 108,000 t of uranium ore (containing 217 t of uranium) are processed in a uranium mill to yield 245 t of U308 (containing nearly 208 t of uranium), and that compound gets converted to about 306 t of natural UF6; enrichment produces about 38 t of enriched UF6, and the hexafluoride is then converted to almost 29 t of UO2 (containing 25.4 t of uranium), which is fabricated into fuel rods. This reactor fuel will generate 8.765 TWh/year (averaging about 2.9 t U/TWh) and produce 28.8 t of spent fuel.