Power Density

by Vaclav Smil

  Uranium Mining and the Nuclear Cycle

  All pre-World War II uranium production Qachymov in Bohemia, Colorado, Cornwall, the Congolese Katanga) was done underground, and a deep mine, Cameco's McArthur River Mine in northern Saskatchewan, is the world's largest uranium-producing operation (fig. 5.5). The mine opened in the year 2000, and by 2012 its cumulative extraction (by indirect methods, using freezing and raise bore mining to minimize exposure to radioactive surroundings) reached about 104,000 t and its annual output of about 7,500 t was 13% of the global total (Cameco 2013b). The mine occupies a compact main area of about 1,000 by 400 m, four smaller outlying areas, and an airstrip facility-altogether about 100 ha, which means that only about 10 m2 have been required per tonne of output during 13 years of operation.

  Figure 5.5

  McArthur River uranium mine, Saskatchewan. © STRINGER/CANADA/Reuters/ Corbis.

  As 1 t of natural uranium can generate 42.2 GWh of electricity, the mine's 2000-2012 output produced 4.38 PWh of electricity (roughly 38.5 GW), that is, just 0.023 ha/TWh, and at a very high extraction power density of about 38,000 We/m2 of surface disturbance. The mine's ore is milled at Key Lake, about 80 km southwest of McArthur River, at a former surface mine whose cumulative 1983-2002 production reached about 95,000 t of uranium between 1983 and 2002 and whose total disturbed area (pits, leaching ponds, tailings, roads) amounts to about 15 km2. By using the metal output for both sites (440,000 t in 30 years between 1983 and 2012) and their total disturbed area (some 16 km2), we can calculate a longterm (1983-2012) average of land claims that combines mining (surface and underground) and milling at the world's two premiere uranium deposits: it has averaged less than 40 m2 (0.0036 ha)/t during the three decades, and its extraction plus milling power density has been about 4,400 We/m2.

  But this is an exceptionally high rate because northern Saskatchewan deposits have an extraordinarily high concentration of uranium: at 16.36% U308, McArthur River ore grade is two orders of magnitude above the global average. The Olympic Dam mine in southern Australia is the world's second largest uranium producer from ore (a uranium content of 0.12%) and also yields copper, silver, and gold (WNA 2013a). Its sprawling operations cover about 21 km2 and its cumulative 1988-2012 uranium output was about 65,000 t (WNA 2012): if the entire land claim is attributed to uranium mining, the rate is about 0.03 ha/t, or (at roughly 12 GW) 600 We/m2. The Ranger mine in Australia's Northern Territories (the world's third largest uranium enterprise) produced the metal by surface mining between 1981 and 2012. Its operations yielded about 100,000 t and its pits, ponds, ore stores, tailings, buildings, and roads covered about 900 ha (with about 420 ha disturbed). The overall operation claim prorated to about 0.009 ha/t, that is, a power density of roughly 1,700 We/m2.

  The large open pit and associated mine structures of the Rossing deposit in Namibia, the number six producer globally, cover nearly 25 km2 and the operation produced about 103,000 t of uranium between 1976 and 2012 (WNA 2013a), requiring about 250 m2/t, or less than 550 We/m2. But the share of surface mining has been declining as in situ leaching (ISL) has emerged as the most important method of uranium production. In 2012 that process supplied 45% of all uranium, followed by about 28% from underground seams and 20% from open pits, with the small remainder being a by-product of other extraction (WNA 2013b).

  In large sheetlike deposits ISL is deployed as a gridded well field where injection wells (used to introduce an acid, or alkaline, leaching solution into an aquifer) alternate with extraction wells (from which submersible pumps lift the leachate to the surface) at a spacing of 50-60 m. Additional wells are drilled above and below the aquifers within the well field and around its perimeter to monitor the containment of the leaching solutions (IAEA 2005). In narrower, channel-type deposits the well spacing is as close as 20-30 m. Recovery rates are 60%-80% of the metal present in the deposit during 1-3 years of injection and withdrawal.

  Kazakhstan, now the world's leading producer of uranium, has the largest ISL operations, but there is an increasing number of smaller projects in Australia and the United States. Data from these three countries (IAEA 2005; McKay and Miezitis 2001; WNA 2012a) show annual land requirements on the order of 0.1 ha/t of uranium. Crow Butte in Nebraska has an original license covering 1,320 ha, but the land affected the by mine's structures and 11 well fields has been only 440 ha, and the project yielded about 3,800 t of uranium between 2002 and 2012 (USNRC 2013); this translates to an average claim of about 0.11 ha/t in a decade and an extraction power density of only about 380 We/m2. Australia's Beverly mine contains deposits of about 16,300 t of uranium, of which 10,600 t are recoverable by ISL below an area of 800 ha (McKay and Miezitis 2001). With an annual output close to 1,000 t of uranium (recovery in 10-12 years), this implies an extraction power density of 530-640 We/m2.

  Underground and surface mining and ore milling leave behind voluminous tailings, but the processing of yellowcake claims a minimal amount of space. Cameco's Canadian conversion facility in Port Hope, Ontario, is licensed to convert U3O8 into 12,500 t of UF6 and 2,800 t of U02 (used to fuel Canada's CANDU reactors) but occupies only 9.6 ha on the northern shore of Lake Ontario (Cameco 2013a; Senes Consultants 2009). This means that even if it worked at half its annual capacity, the facility's conversion power density would be on the order of 105 We/m2. The fabrication of fuel rods to reactor specifications is a similarly highly concentrated process with negligible space requirements.

  Additional space should be allocated because of the relatively high electricity requirement of the enrichment process. This depends on the degree of enrichment, an effort measured in separative work units (SWU). The gaseous diffusion process, the original separation method, is highly energyintensive (about 2.3 MWh/SWU), while gas centrifuge plants, now the dominant way of enriching the fuel in the United States, need only about 60 kWh/SWU (FAS 2013). The previously traced nuclear fuel chain material balance would require more than 116,000 SWU and the gaseous diffusion would need 267.7 GWh to enrich enough fuel to operate a 1-GWe reactor at full capacity for a year. If the enriched fuel were supplied solely by gaseous diffusion, then the total annual electricity consumption would be no less than 240 GWh; if it were supplied only by centrifuge plants, the requirement would be as low as 7 GWh.

  If all the fuel were supplied by a combination of the two processes in one country, and if all the requisite electricity were to originate from a single source, it would be easy to calculate a weighted mean. But the United States relies on foreign enrichment services to fuel its nuclear power plants: in 2012, owners and operators of America's commercial nuclear power reactors purchased enrichment services totaling 16 million SWU (USEIA 2013d); the electricity used to enrich it came from different (and gradually changing) national mixtures of sources; and their power densities (as we have seen in preceding sections) range over several orders of magnitude, from large hydroelectric to natural gas-fired generation.

  Consequently, to do as Lovins (2011b) has done, that is, to select gaseous separation as the only choice, estimate that about 10 TWh of electricity are needed to separate the isotopes during the 1-GW plant's four decades of operation, and then assume that all of that energy comes from coal-fired power plants whose annual land requirements add up to 580 ha/TWh, is a questionable procedure. Its outcome would add about 150 ha/year to land claims of a 1-GWe nuclear station. In contrast, Fthenakis and Kim (2009) put enrichment's land requirements at about 3 m2/GWh (assuming 70% centrifugal and 30% diffusion enrichment); for a 1-GWe plant with a 90% capacity factor that would translate to roughly an additional 2.6 ha/year, less than 2% of Lovins's huge total. Even at double that rate (about 5 ha/ year), this would be a small addition that would have only a marginal effect on the aggregate count.

  During the 1970s, the decade of the record expansion of US nuclear capacities, Mielke (1977) put the claims of fissile fuel production (all prorated for one year of a 1-GWe light water reactor operation) as follows: temporarily committed land for mining, 22 ha; for millin
g, 0.2 ha; for UF6 production, 1 ha; for uranium enrichment, 0.2 ha; and for fuel fabrication, just 800 m2. The corresponding totals for actually disturbed area were 6.8, 0.1, 0.08, 0.08, and a mere 160 m2. Mielke considered only direct land claims of enrichment, but otherwise his numbers convey well the relative land demands of nuclear generation: differences for mining claims among underground, surface, and ISL operations will in most cases be far more important in determining the overall requirements than will the aggregates of all postmining operations.

  Finally, there is the matter of the long-term storage of spent fuel. The fuel is removed from reactors to adjacent storage ponds, where it can stay for months or years as its radioactivity decreases; this common practice creates no additional land claims beyond plant boundaries. Cooling ponds now contain about 90% of the world's 270,000 t of all used fuel, with the remainder in dry storage (WNA 2012). In the United States, about a quarter of all used fuel is in interim storage in sealed steel casks or modules at Independent Spent Fuel Storage Installations (USNRC 2012b). The relatively limited mass and volume of these wastes help to explain why there has been no urgency to set up a permanent disposal site (the NIMBY syndrome is another consideration).

  While there is still no permanent national facility for the long-term storage of highly radioactive waste, many years of planning for the Yucca Mountain project in Nevada (now essentially terminated) offer relevant insights into the storage capacities and potential land claims of such depositories (Boisseau 2009). The site's intended capacity was 70,000 t of radioactive waste, and the deposit's footprint would have been 4.27 km2, but its controlled access area would have extended over 230 km2 (Cochran 2008). Using the latter total, the rate would have been about 3,300 m2/t, and hence the waste from a 1-GWe reactor (nearly 29 t of spent fuel a year) would have an overall land claim of roughly 9.5 ha/year.

  Three Representative Plants

  As I did for coal-fired plants I will present two very different but realistic examples of aggregate power densities of nuclear electricity generation, as well as an example of a more commonly encountered facility, all for a standard 1-GWe station with a high (90%) capacity factor. The first plant is a compact operation (much like San Onofre) that occupies just 50 ha and whose fuel comes from the world's most productive Saskatchewan mines (just 40 m2/t U). The fuel is enriched only by centrifugal process (6 GWh/ year), with electricity supplied by coal-fired generation operating with a high power density of 1,000 We/m2 (that is, 8.76 MWh/m2). Spent fuel is eventually stored in a permanent depository whose land claim amounts to 10 ha/year. The plant's land claim is dominated by its site, and it operates with a power density of about 1,600 We/m2.

  The second station is a sprawling enterprise that spreads over 1,000 ha and receives uranium through a low-density ISL recovery (0.2 ha/t U). Enrichment of the uranium is split between a gaseous diffusion process (about 268 GWh needed to enrich fuel for 1 GWe) and a centrifugal process (6 GWh), resulting in an average of 137 GWh required to produce the fuel for one year of the plant's operation. Land requirements may increase even further if that electricity originated in hydrogenation operating with a very low power density of 5 We/m2 (43.8 kWh/m2). This plant's land claim is also dominated by its site, but fuel production will add another 30%, and the plant will operate with a power density of only about 70 We/m2.

  Box 5.5

  Power density of a nuclear electricity generation (high-density variant)

  Box 5.6

  Power density of a nuclear electricity-generating plant (low-density variant)

  The most representative of the three nuclear power plant examples is a station whose site occupies an area right in the middle of the US modal range (200-400 ha). Other assumptions are as follows: the station receives uranium from several sources, including overseas imports; uranium extraction (with a significant share coming from ISL) and ore milling land claims average 0.1 ha/t; and the weighted fuel enrichment cost (80% centrifugal, 20% gaseous process) is about 58 GWh, with electricity coming from a mixture of sources with an average power density of about 500 We/m2 (4.38 MWh/m2). This plant's land claim is also dominated by its site, but fuel production and disposal add roughly 50% to the plant's area, and the station generates electricity with an overall power density of about 300 We/m2.

  Box 5.7

  Power density of a nuclear electricity-generating plant (medium variant)

  These three realistic examples thus span annual claims of roughly 601,360 ha and power densities of 70-1,600 We/m2, attesting again to the fact that the power densities of thermal electricity-generating stations are highly constrained as far their core structures and indispensable infrastructures are concerned but can differ by two orders of magnitude mainly because of differences in fenced-in areas (including undisturbed land), cooling arrangements, and the origins of the fuel supply.

  In closing, here are a few values offered by several studies as averages for the land requirements of nuclear electricity generation. Gagnon, Belanger, and Uchiyama (2002) estimated direct land requirements at 0.5 km2/TWh, that is, about 440 ha for a 1-GWe plant, and a power density of about 230 W/m2. Fthenakis and Kim (2009) concluded that US nuclear generation claims 119 m2/GWh; that implies a total claim of about 104 ha for 1 GWe and a power density of roughly 960 We/m2. The largest component (42% of the total) of their account was the plant itself, but it is obvious that with 42 ha (an area smaller than that controlled by the most compact US nuclear station) they counted only the footprint of its structures, not the total fenced-in area of the facility.

  Their relatively high mining estimate results from the assumed split of extraction methods (50% open pit, 50% underground, no ISL), and their indirect claim for fuel enrichment reflects the 70/30 split between centrifugal separation and diffusion. Lovins (2011b) arrived at an almost identical rate of roughly 120 m2/GWh. McDonald and co-workers (2009) offered a fairly narrow range, 3.02 km2/GW for the most compact and 4.78 km2/GW for the least compact nuclear plant, implying power densities of 210-330 We/m2. These claims, 302-478 ha for a 1-GWe station, are based on a study by Spitzley and Keoleian (2004), whose assumptions were also cited by Lovins (2011b). Jacobson's (2008) calculations for an 847-MW reactor (which included all land for uranium production and a safety zone) ended up with 150 We/m2.

  Transmission

  I could have quantified the land requirements of electricity transmission in the third chapter while assessing the power densities of renewable energies, but placing them here is more apposite. After all, for more than 130 years (Edison's first power plant began operating in 1882) most of the world's electricity has been generated by burning fossil fuels, and since the late 1950s also by nuclear fission: in 2014 thermal electricity generation supplied nearly 80% of the world's total demand-and technical, infrastructural, and economic imperatives make it certain that this primacy will continue for decades to come.

  Much like thermal electricity generation, electricity transmission retains its basics (now more than a century old), but its performance has been much improved as better transformers, higher voltages, taller towers, better wires, and longer unsupported spans have kept pace with the rising demand for larger transfers across longer distances in more difficult terrain.

  Early choices are also reflected in different consumer voltages (100 V in Japan, 120 V in North America, 230 V in Europe), but these do not affect the land claims of high-voltage transmission. Distribution lines to consumers have voltages less than 35 kV, and in modern urban developments they are underground. High-voltage transmission of alternating current (AC) operates at 110 kV and 230 kV, and it steps up to extra-high voltage at 345 kV, 500 kV, and 765 W.Direct current (DC) transmission, a more efficient alternative to connect load centers (cities, industries) with distant sources of power (usually large hydroelectric plants), has higher voltages to reduce transmission losses, but the capacities of high-voltage lines are limited by their heating and by voltage drop.

  The width of transmission rights-of-way increases with voltage
, but extra-high-voltage lines make lower specific claims. In the United States the National Electric Safety Code specifies minimal clearances of 36-46 m for 230-kV lines, 46 m for 345-kV links, and 61 m for 765-kV (all AC) transmission (IEEE 2012). This means that 1 km of high-voltage transmission claims 3.6-6.1 ha of ROW, and the power densities of annual electricity throughput in the United States in 2012 (440 GW) would have been-with 305,000 km of lines 230 kV and higher (USEIA 2014b) and assuming an average ROW of 5 ha/km-just short of 30 W/m2. Because the line conductors (made of aluminum alloys) are not insulated and can fall on the ground, the ROW strip should have no permanently inhabited structures and no tall vegetation (the limit is usually 1.8 m, high enough to grow Christmas trees under high-voltage lines).

  Lines running through forests on flat or gently undulating land need adequate clearing, but many long spans that cross mountain valleys high above the ground do not require cleared strips. And when lines cross barren terrain, natural shrublands and grasslands, or land planted to field crops, there is also no need for ground clearances: existing land use can continue, and land claims, limited to the small areas needed to anchor transmission towers, can be further reduced with new monopole tower designs. Consequently, the high-voltage transmission land claims resemble those of wind turbines as they occupy only tiny shares (even less than 1%) of the total ROW claim. DC transmission has lower losses and narrower ROWs: to transmit the same power single-circuit 500-kV AC line would take 105 m, a double-circuit 500-kV AC line would claim 65-76 m, and 500-kV DC lines would need just 55-60 m of ROW (ATCO Electric 2010).