Power Density

by Vaclav Smil

  And how does the total of 2.5 Gt of charcoaling wood (or, with a 10% markup for losses, 2.75 Gt) compare to today's global roundwood harvest? According to the FAO, that total wood harvest reached 3.4 Gm3 in 2010 or (assuming 0.65 t/m3) about 2.2 Gt (FAO 2014), and a generous addition of 15% for illegal logging would bring it to 2.5 Gt. Consequently, a global charcoal-based iron smelting that would replicate Brazilian practices would alone claim as much (or slightly more) wood than the world's total (legal and illegal) 2010 wood harvest for fuel, lumber, and pulp. Another way to look at this is that in 2010 (assuming an average 25% conversion efficiency), less than 10% of the world's wood harvest was converted to charcoal, while charcoal-based iron smelting at the 2010 level would require converting all of the world's harvest of wood into charcoal, leaving nothing for other uses-and would require a doubling of recent wood harvests to keep the world supplied with timber, veneer, plywood, fuel, and paper pulp.

  The extent of the harvested area would depend on prevailing yields, and those, in turn, vary with wood species, soils, climate, and plantation subsidies. With an annual mean increment of 10 t/ha, the global pig iron output in the year 2010 would have required harvests of at least 250 Mha (2.5 Tm2), an area equal to 60% of Brazil's Amazon rain forest of 4.1 Tm2 or to nearly half of the total forested area of 5.5 Tm2 in the entire Amazon basin. With 19 MJ/kg of dry wood-the average value for eucalyptus clones, according to Pereira and co-workers (2012)-that would imply an average power density of 0.6 W/m2. A large-scale transition to charcoal-based iron smelting would lead to many improvements, and hence it is realistic to make more optimistic assumptions.

  For example, Piketty and co-workers (2009), assuming an average (dry matter) wood yield of 16 t/ha and a 30% carbonization yield (for 80% C content with 10% handling loss), put the land requirement at 1,290 km2 for 1 Mt of hot metal, and hence the global iron output fully energized by charcoal would require roughly 130 Mha of high-yielding tropical plantations. Even better performance is conceivable. Modern continuous charcoaling methods (retorts) should have average conversion yields of 35%-40% (Rousset et al. 2011), and cultivation of high-yielding clones in eucalyptus plantations should bring annual wood increments as high as 25 t/ha (Pfeiffer, Sousa, and Silva 2012). In combination, those increments would boost the charcoal yields per hectare fourfold when compared to my initial assumptions, and the global iron smelting at the 2010 level could be supported by less than 70 Mha of tropical plantations.

  But all of these calculations are questionable because even the existing Brazilian plantations have serious environmental impacts, because it is most unlikely that all charcoal would come just from Brazil, and because the global iron smelting will continue to increase in order to supply the enormous demand for steel, mainly in Asia and Africa. Extensive areas of tropical eucalyptus (and pine) plantations have been already described by Brazilian environmentalists as deserto verde, destroying biodiversity and maintaining monocultures through intensive applications of herbicides (Reporter Brasil 2011). A common practice in Brazil is to harvest trees in a five-year rotation and coppiced for three cycles: after 15 years of growth, glyphosate is applied to kill the remaining rootstock, and new seedlings are planted (Bailis et al. 2013).

  Monocultural plantations also increase the rate of soil erosion, divert water from nearby farms, and contaminate runoff with agrochemicals, and their expansion would compete even more extensively for land that would otherwise be used for crops or pasture. In any case, it is extremely unrealistic to expect that all charcoal needed for global pig iron smelting could be made only from harvesting intensive plantations of Brazilian eucalyptus hybrids-and if a significant share of charcoal were to come from temperate species, then the average wood yields would be substantially lower. The short-term cultivation of small plots has impressively high yields, but studies that look at productivities in various environments most likely to be used for tree cultivation (abandoned farmland, former forest land) show a predictably wide range of outcomes.

  Truax and co-workers (2012) found that for the productivity of hybrid poplar plantations on abandoned farmland in Quebec, the site effect (elevation, climate, soil fertility) was far more important than the clone effect, with annual yields as high as 22.4 m3/ha (about 11 t/ha) in bottomlands and as low as 1.1 m3/ha (about 0.5 t/ha) on the poorer soils of hill slopes. In northern Italy, Paris and co-workers (2011) harvested 15-20 t/ha/year, but only with fairly heavy nitrogen applications (300 kg N/ha), while achieving means between 10 and 14 t/ha elsewhere, while in southern Italy (Latium and Molise), Di Matteo and co-workers (2012) recorded annual poplar yields of 10-13 t/ha. And a review of 21 studies of poplar and willow plantations in Europe and the United States found yields between 5 and 16.8 t/ha (Djomo et al. 2011). Consequently, assuming a future average global wood increment of 15 t/ha would not be too conservative.

  The global output of pig iron doubled between 1990 and 2010, and even if it were to grow only half as fast in the next two decades it would reach about 1.6 Gt in the year 2030. Using the more optimistic assumptions regarding future wood and charcoal productivities it would thus require at least 2.9 Gt of charcoal and annual harvest of wood from more than 190 Mha in 2030. This would be an area of forest or tree plantations only slightly smaller than in the first scenario, and equal to nearly half of Brazil's Amazon rain forest or to more than China's total forest land. And this massive spatial claim would have enormous environmental repercussions, particularly when most of this wood would have to be grown in high-yielding monocultural plantations that would require large inputs of fertilizer, pesticides, herbicides, and, in many drier climates, at least supplementary irrigation.

  The world whose dominant metal would be smelted with charcoal produced by the annual logging of an area equal to half of the Brazilian Amazon is conceivable, but it is hardly desirable, and we simply cannot appreciate all of its eventual consequences. But a fundamental concern is clear: could such massive tree harvests (required to produce a billion tonnes of charcoal a year) offer a truly renewable alternative, given their impacts on biodiversity and soil erosion and their constant requirements for water, nutrients, and protection against pests? What would be the real cost of this enormous enterprise, and how practical would it be even if the costs were a secondary matter?

  Even if it turned out to be more practical than we think today, the industry based on harvesting wood on a semicontinental scale would be a very different one compared to our current arrangements, in which iron production is based on carbon-rich coke made from coal that is extracted in just a few thousand large mines in a dozen major coal-producing countries and that is easily distributed to large industrial centers for coking. Of course, our current practice taps a finite energy resource, but in light of its importance for iron smelting, we could accord it a highly preferential status and keep relying on it for many generations to come. That would not be difficult to do because we could replace coal's largest use, for electricity generation, with other energy sources.

  Wind and Solar Electricity

  Both of these flows differ fundamentally from biofuel harvests and conversions. On a positive side, they have substantially higher power densities, and wind turbines can share other productive (agricultural, pastoral, silvicultural) activities in the area required for their optimized spacing. But staggered wood or crop harvesting and phytomass storage (entire trees, wood chips, baled straw) can ensure a continuous supply, while intermittent radiation and wind have, at best, moderately high and often fairly low capacity factors.

  Wind

  Conversion of wind to electricity has been undoubtedly helped by its light footprint, as the surfaces actually occupied by wind turbine foundations, access roads, and transmission towers amount to a small fraction of the area that contains properly spaced machines. In acknowledgment of this reality I have not used the power densities of a wind farm when quantifying the aggregate claims of energy systems but have counted only actual footprints approximated by a high power density of 50 W/m2
.

  And yet using this apparently rational choice is misleading because the power densities of machine spacing are highly relevant for the expansion of the industry. In the United States, where relatively fast and fairly persistent winds prevail across large parts of the Great Plains between northern Texas and North Dakota, there is no imminent prospect of running out of windy sites, but if the wind projects in the region keep expanding they will run into a fundamental power density imperative, namely, the limits of power production by wind farms larger than about 100 km2. As shown by Adams and Keith (2013), wind turbine drag on local winds would limit electricity generation by these large installations to no more than 1 W/m2.

  Germany faces a different problem. The earliest stages of wind power development took place in localities that combined high wind speeds with persistent flows, that is, in the coastal regions of the windy northern states of Niedersachsen and Schleswig-Holstein. This first phase was to be followed by massive offshore projects, but the costs of such projects-and the costs of and delays in connecting them to land by undersea cables, and then transmitting electricity to the south-refocused attention on the country's central and southern regions, to sites with lower wind speeds and lower wind frequencies. As a result, there are now plans to build 60,000 new wind turbines in orchards, vineyards, and forests (which will require extensive tree felling), on mountaintops (which will require new access roads for the trucks and heavy cranes used to transport and assemble turbines), and even in protected areas in states ranging from Nordrhein-Westphalen to Bayern and from Baden-Wurttemberg to Sachsen (Schulz 2013).

  Such an expansion would greatly alter the appearance of German landscapes, particularly with the invasion of many forested areas, a dubious quest in light of the inherently low-capacity factors in many of these central and southern locations. Still, to meet the country's ambitious goal of 35% of energy coming from renewable sources by 2020, many proponents of massive wind power development offer no relief. Winfried Kretschmann, minister-president of the state of Baden-Wurttemberg and the first Green Party president of the German Bundesrat (in 2012-2013), insists (as quoted in Schulz 2013) that "es fiihrt kein Weg daran vorbei, die Landschaft auf these Weise zu verschandeln" (there is simply no other way but to disfigure the countryside like this).

  How much this will be opposed in Germany remains to be seen, but the UK has already seen a great deal of opposition to what is perceived as defacement of the country's remarkable landscapes. Wind farms "desecrate our national heritage" (McMahon 2011, 18), "as if some malevolent creature from mythology shed its spawn over the land" (Etherington 2009, 10). Again, the insult would be more tolerable if much higher power densities of wind machines allowed concentrating electricity generation into a much smaller number of locations. And disfiguration of landscapes is not the only consequence of the limited power densities of wind power.

  The need to install large numbers of machines tends to reduce the width of noise exclusion corridors, to increase the chances of large-scale bird fatalities, and to affect many terrestrial species as a consequence of the fragmentation of their habitat. In windy and sparsely inhabited Scotland, the rule is to allow 2 km between wind farms and the edge of cities and villages, while in densely populated German states the minimum distances from houses are just 500 (in Bayern), even 300 m (in Sachsen). And, obviously, large wind farms with hundreds of smaller machines in forested and mountainous terrain will kill more birds-particularly raptors, which tend to use mountain slopes and ridge saddles between hills (Subramanian 2012)than single machines or a small grouping of large wind turbines located on flatlands.

  PV Generation

  Germany is also the prime example of forcing (through high subsidies) massive installations of PV panels in the country where solar electricity generation has inherently low power densities not only because of the relatively high latitudes (roughly 47°-55°N) but also because of very low capacity factors in climates governed by the prevailing frontal flows from the Atlantic: in 2012 the country had nearly 40% more PV capacity (32.4 GW) than sunny Spain, Italy, and Greece combined (23.7 GW). But in 2013 Germany's Siemens, one of the key benefactors of the country's solar boom, published a study whose conclusion was that building and expanding Europe's solar and wind installations in wrong locations was costing €45 billion (about $60 billion) in unnecessary investment (Siemens 2013b).

  I read this with astonishment: one of the world's largest engineering firms had apparently discovered the power of power densities! Details are telling: if Europe's new PV capacities that are to be built by 2030 (about 140 GW) were located in the sunniest Mediterranean sites, the EU could save 39 GW of installed capacity, even after accounting for the cost of additional south-to-north transmission. At the same time, Siemens and several other major German companies (since 2009 grouped into Dii GmbH with other EU partners) was a member of a consortium pushing for a solution that is the very opposite of PV expansion in Germany. The DESERTEC Project (or EUMENA-EU and the Middle East and North Africa) is to make PV-based electricity generated in Sahara and the Middle East the principal source of European demand (DESERTEC 2014; fig. 8.2). The scheme would take advantage of the highest possible power densities and would rely on concentrated solar power (CSP), rather than PV cells, to provide a more evenly distributed supply.

  Figure 8.2

  DESERTEC. Carl De Torres Graphic Design. Based on DESERTEC 2014.

  Post-2010 upheavals in North Africa and the Middle East have accentuated the project's major weakness: costs and the need to construct highvoltage links of unprecedented capacity aside, would it be wise to put large CSP facilities in Morocco, Algeria, Tunisia, Libya, Egypt, Syria, Saudi Arabia, and Iraq when most of these countries are not only politically unstable but also have been subject to violent conflicts, and then to transmit all electricity through a few choke points (Gibraltar, Sicily, Bosporus)? In this case the highest possible solar power densities are easily trumped by the inherent riskiness of the endeavor. In any case, Siemens (and Bosch) left the consortium in 2012, and there are no imminent prospects for DESERTEC becoming a major source of EU electricity. DESERTEC also highlights the challenges of servicing very large areas of solar collectors or reflectors in and environments full of airborne dust.

  Regular cleaning (at least two to three times a year) of these exposed surfaces with power washers or brushes would obviously be both water - and labor-intensive, and in many locations the cost of water would surpass the cost of electricity. The best solution appears to be the deployment of robots that glide along rows of PV panels, an innovation by Greenbotics of California that also brings large water savings (SunPower 2013). But this option requires higher inputs of energy-intensive materials to make the devices and more electricity to operate them, thus reducing the net power generation of a solar facility. Robotic cleaning might be a good solution for China with its extraordinarily high air pollution and chronic shortages of water, but more would be required for the DESERTEC projects, where many near-ground structures could be buried under sands in a matter of months.

  Another complication whose true dimension cannot be fully answered is the longevity of PV panels mounted on rooftops. They make no new spatial claims as they are merely additions to built-up (impervious) areas, but the length of their service may be rather short in many rapidly growing Asian cities, where their optimal placement may also be fairly limited owing to common shading and where their performance could be heavily degraded by extraordinarily high levels of air pollution (with airborne particulate matter in China's largest cities reaching maxima an order of magnitude higher than the WHO air quality standards).

  Finally, a few clarifications are in order regarding the often cited claims that imply extraordinarily high power densities of solar generation. Calculations that divide national electricity demand by peak solar power in sunny regions are quite misleading. For the United States, the 2010 electricity demand of 422 GW could be supplied by an area of some 2,800 km2 (a square with sides of 53 km) in Ar
izona-but only if that surface could generate year-round, with a noontime peak power density of 150 We/m2. The National Renewable Energy Laboratory offers a somewhat less misleading number when it claims that "a 100-by-100 mile area of Nevada could supply the United States with all of its electricity" (NREL 2013, 1). In reality, that area (25,600 km2) could-with an average insolation of 220 W/m2 (Las Vegas mean), a capacity factor of 25%, and a high average conversion efficiency of 15%, and hence with an average annual power density of 8.25 We/m2-generate about 211 GW, or half of the 2010 US demand.

  Realistic theoretical estimates are highly dependent on the assumptions used to construct a massive imaginary nationwide PV system. Denholm and Margolis (2008a) used long-term data on average irradiation for the 48 contiguous states assumed a combination of 25% of rooftop and 75% ground-based modules (40% fixed arrays with a 25-degree tilt, 25% singleaxis and 10% two-axis tracking) coupled with appropriate long-term storage and ended up with total requirements of 181 m2/capita-compared to per capita averages of 35 m2 for airports (or golf courses), 65 m2 for roofs, 162 m2 for major roads, and nearly 840 m2 for urban areas. This means that in absolute terms, entirely PV-based US electricity generation would require about 55,000 km2, slightly larger than the combined area of Massachusetts and Vermont and about 0.7% of the total area of the 48 states, and the average power density would be a realistic rate of less than 8 We/m2.