by Vaclav Smil
But in some extreme circumstances virtually the entire litterfall has been harvested in regions with severe fuel shortages, particularly in long-ago deforested areas of North China, where peasants used to sweep up every piece of plant litter, including leaves, needles, and small twigs, shed by remaining wood groves and carry them in baskets on their backs to their house to stir-fry their meager meals; this practice could be seen in North China as recently as during the 1980s. But because the litter yield in those low-density, low-productivity groves was no more than a 3-4 t/ha, even its complete harvest would prorate to less than 0.2 W/m2. The lowest woodfuel collection rates have been recorded in the poorest and parts of Africa; for example, in Eritrea's scrublands and wooded grasslands the annual yield is as low as 30-50 kg/ha of air-dried (15 GJ/t) wood (Arayal 1999). Even the latter rate is just 5 g of woody phytomass per square meter, just a tiny twig and a minuscule power density of 0.0024 W/m2.
Cutting down mature virgin forests or older succession growth to supply towns and growing cities tapped much richer stores of phytomass but did not proceed with much higher power densities. Clearing a rich temperate primary forest could yield 400 t/ha, but as such a harvest could not be repeated (even after 80 years a secondary growth would yield at least 20% less), a properly prorated annual power density mean was no more than about 0.25 W/m2. Large cities located in temperate regions-such as China's northern capitals Xian and Beijing, London, and Paris-had considerable demand for wood, a function not only of their climate and size but also of the wasteful combustion in open fireplaces (more efficient enclosed stoves came relatively late even in Europe) and the similarly wasteful conversion of wood into charcoal, the preferred (smokeless) fuel for heating.
All preindustrial averages of per capita fuelwood consumption are just approximations, but they suffice to illustrate annual urban claims. My estimate for the early Roman Empire is no more than 600 kg/capita (Smil 2010c). In London of the early fourteenth century it was 1.5 t/capita (Galloway, Keene, and Murphy 1996). During the eighteenth century a typical German mean was close to 3.5 t/capita (Sieferle 2001). In 1830 it was about 4.5 t/capita in Austria (Krausmann and Haberl 2002). In the forestrich United States the mean was nearly 6 t/capita (roughly split between household and industrial uses) during the 1850s (Schurr and Netschert 1960), but in Paris it declined from nearly 1 t/capita in 1815 to less than 300 kg/capita by 1850 (Clout 1983). Even if the average consumption were only 1 t/capita, a preindustrial city of million people would have needed every year wood harvested from about 250,000 ha, the equivalent of a square with sides of 50 km.
The prevailing power densities were further reduced by the conversion of wood to cleaner-burning but inefficiently produced charcoal. As already explained, English charcoal-making during the eighteenth century-when the charcoal to wood ratio was 1:5 in mass and (assuming 29 GJ/t of charcoal and 19 GJ/t of wood) about 1:3.3 in energy terms-operated with a power density of less than 0.1 (just 0.07) W/m2. Iron-makers located in forested regions (resulting, inevitably, in extensive local deforestation), but cities had to import charcoal from increasingly distant sources, even though the easily crushable fuel is not suited for long-distance transport.
An eighteenth-century city consuming roughly half of its fuel demand as fuelwood and half as charcoal would draw on an energy supply produced with power densities of less than 0.15 W/m2, and for every 100,000 of its inhabitants its annual fuel supply would have required wood harvests from about 40,000 ha of forested land. The combination of expanding areas of deforestation in the vicinity of large cities and the high cost of land transport was an important factor limiting the size of preindustrial cities that had to be supplied by wood delivered by heavy carts (pulled by oxen or horses) or on the backs of donkeys or camels (a common practice in pre1900 Beijing) from increasing distances.
Harvests of crop residues had much lower power densities than those of woody phytomass, in temperate regions almost always less than 0.05 W/m2 (less than 1 t/ha). But the two kinds of biofuels are not in the same category, and hence the comparison is flawed: wood is the targeted harvest, while crop residues are just by-products of harvesting cereal, leguminous, oil, sugar, or fiber crops. Moreover, straw and stalks and leaves have always had many more valuable uses than to be burned inefficiently in small household stoves. Such uses range from animal feed to substrates for mushroom cultivation, and the best choice might be to incorporate most, even all, residues into soils to replenish organic matter, retain moisture, and help prevent erosion (Smil 2013a).
Some modern harvests of wood for energy still come from natural forests (mostly from secondary and tertiary growth), but the trend has been to cultivate the trees. The planting of fast-growing tree species grown in short rotations began during the 1960s, mostly with poplar clones (Dickmann 2006). These plantings are dense (some with more than 50,000 stems/ha, to be thinned later) and are harvested after four to six years (rarely, more than 10 years), and their harvest is followed either by coppiced growth or by replanting (West 2013). The recent interest in biomass energy has led to a wave of experiments with fast-growing tree species and to reports of some extraordinarily high yields, even in excess of 50 t/ha. Such claims are often based on a simplistic extrapolation of optimally tended small experimental plots. Care must also be taken to properly convert volumetric yield reports (a common practice in forestry) to mass and energy equivalents.
The specific density of commonly exploited species ranges from less than 350 kg/m3 for some firs to 500-550 kg/m3 for maples, ashes, elms, and oaks. Densities change with age, and vary even among closely related species: for example, young fast-growing plantation poplars in Washington state had a density of 370 kg/m3 during the first three years of growth but averaged 450 kg/m3 six years later (DeBell et al. 2002), while diverse families of loblolly pine have densities ranging between 440 and nearly 510 kg/m3 (Belonger, McKeand, and Jett 1997). The UN Food and Agriculture Organization uses an average density of 500 kg/m3 for conifers and 600 kg/m3 for leafy trees. The energy density (absolutely dry matter) of commonly harvested trees is mostly between 16 and 19 MJ/kg.
Leaving some dubious extreme claims aside, properly conducted recent experiments-including those by Rao, Joseph, and Sreemannarayana (2000), Hytonen (2008), and Sarlls and Oladosu (2010)-and many recent studies of hybrid poplars (Di Matteo et al. 2012; Klasnja et al. 2003; Paris et al. 2011; Truax et al. 2012) confirm the long-standing knowledge of typical yields in tree plantations (Cannell 1989; Dickmann 2006; Mead 2005). In most temperate settings with natural growth or with moderate fertilization (and perhaps some supplemental irrigation), fast-growing large-scale plantings of pines, acacias, poplars, or willows will yield (depending on climate, soils, cultivars, and inputs) between 5 and 15 t/ha (fig. 3.7). In the subtropics and tropics, commonly cultivated trees (different species of Eucalyptus and Acacia, Leucaena, Pinus, and Dalbergia) will yield 20-25 t/ha (ITTO 2009). If a harvested dry mass of 19 GJ/t is assumed, the temperate rates imply power densities of 0.3-0.9 W/m2, the tropical ones range between 1.2 and 1.5 W/m2.
Mechanical harvesters (wheeled or tracked, first developed in Scandinavia during the 1970s and 1980s) perform the entire felling and sawing sequence, from cutting down a tree near the ground to removing its branches and bucking it. Stumps, branches, and treetops remain on-site to recycle nutrients. There is also an option to harvest entire trees (leaving only stumps) by using a sequence of a feller-buncher, chipper, and truck loader, which obviously leads to the recycling only of the nutrients in the stumps. Wood chips are burned (sometimes after preliminary drying) to produce steam for electricity generation, or for both heat and electricity in cogeneration plants.
Figure 3.7
Pine (Pinus taeda) plantation in the United States. USDA.
The combustion of woody phytomass is done most efficiently (close to 90%) in circulating or bubbling fluidized bed boilers (Khan et al. 2009). Gasification (in low - or high-pressure gasifiers) converts as much as 80%85% of energy feed into a gas whose
composition is dominated by CO and H2 and whose energy density is, depending on the procedure used, as low as 5.4 MJ/m3 or as high as 17 MJ/m3 (Worley and Yale 2012). When phytomass-derived gas is used in engines or turbines to generate electricity, the overall efficiency of the sequence could be as high as 35%-40%. Converting wood chips to methanol (CH3OH) can be done with efficiencies of up to 70% (Methanol Institute 2013).
Typical yields of Brazilian eucalyptus plantations rose from about 12 t/ha in 1980 to 21 t/ha in 2011 (CNI 2012), but sustainable yields are much lower in other tropical regions, usually less than 10 t/ha in Africa, India, and Southeast Asia (ITTO 2009). Even when wood from a highly productive tropical plantation (20 t/ha, 1.2 W/m2) is burned, maximum power densities would be close to 1.1 W/m2 for heat generation, 1 W/m2 for gasification, around 0.8 W/m2 for methanol production, and less than 0.5 W/m2 for gas used for electricity generation. The power densities of methanol production and electricity generation based on woody phytomass harvested in tree plantations in temperate climates would be, in most cases, no more than half of the above values. A specific example shows what a temperate-climate harvest of 10 t/ha would imply for generating electricity by burning wood chips in a large power plant. Even with an average energy density of 19 GJ/t the plantation would yield no more than 190 GJ/ha, resulting in a harvest power density of 0.6 W/m2.
Box 3.8
Power density of wood burning
Box 3.9
Land claimed by a tree plantation
To supply a 1-GWe wood-fired power plant operating with a capacity factor of 70% and a conversion efficiency of 35% would require an annual harvest of about 330,000 ha of fast-growing tree plantation, the equivalent of a square with sides of nearly 58 km.
The total area needed by a wood-fired electricity-generating plant is quite negligible when compared to a large area of land claimed by phytomass production: even if the generating station and its associated structures (fuel storage, switchyard, office and maintenance buildings, access roads) were to occupy 10 ha (a square with sides of 316 m) it would still be only 0.003% of the land required to grow trees. Obviously, the very low power densities of wood-based electricity generation prevent it from becoming anything but a very minor contributor to the overall supply, with plants usually burning wood waste generated by wood-processing industries.
If only 10% of the US electricity generated in 2012 (that is, 405 TWh, or 1.46 EJ) had to be produced by burning wood, then (with an average 35% conversion efficiency) the country would require about 4.17 EJ (about 132 GW) of wood chips. With an average power density of 0.6 W/m2, this would claim about 220,000 km2 of wood plantations, or nearly as much land as all of Idaho or Utah. This calculation also shows why even major yield increases (thanks to better hybrids or to entirely new transgenic trees) would not make a fundamental difference as far as the overall land claims are concerned. Even with a doubling of the assumed 10 t/ha mean (not likely on a large scale in temperate regions for decades to come), a woodfueled plant with a 1-GWe capacity would still need annual harvests of all aboveground phytomass grown on a plantation of nearly 170,000 ha, a square with sides of 40 km.
Liquid Biofuels
The power densities of producing liquid fuels from agricultural crops are even lower, although some species saw impressive gains in yields during the second half of the twentieth century. Before the introduction of hybrid varieties the US nationwide corn yield averaged just 1.5 t/ha during the mid-1930s. By 1975 it was 5.4 t/ha, in 2000 it reached 8.6 t/ha, and it peaked at 10.3 t/ha in 2009 (FAO 2014). Similarly, better cultivars and better agronomic management of Brazilian sugar cane helped lift the crop's average yields from 43 t/ha in 1960 to 71 t/ha in 2012 (FAO 2014). By far the most important process used to convert these crops to fuel has been the production of ethanol, and it too has improved over time: for example, Brazil's ethanol yield increased from 59.2 L/t of sugar cane in 1975 to 80.4 L/t in 2008 (Cortez 2011).
Experiments with automotive ethanol predate World War II, but modern large-scale ethanol production began in Brazil with the country's sugar cane-based ProAlcool program in 1975, and in the United States with cornbased efforts in 1980 (Basso, Basso, and Rocha 2011; Solomon, Barnes, and Halvorsen 2007). By 2010 the two programs had expanded to, respectively, about 50 and 20 GL a year. Brazilian production has been stagnating since 2008 (Angelo 2012), and the supply of the US corn-based ethanol is unlikely to grow beyond the 9%-10% share of the automotive market it has reached, thanks to production mandates imposed by the US Congress.
Box 3.10
Power density of US ethanol production
America's ethanol producers have been averaging 2.8 gallons per bushel of wet corn (ACM 2013); in metric units this would be 0.38 kg/kg of dry grain, but Patzek (2006) showed that the theoretical yield of ethanol from corn starch (making up 66.2% of the grain) is 0.364 kg/kg of dry corn. This discrepancy is explained by the industry's practice of counting gasoline denaturant (5% by volume, 8% by energy content) as ethanol. Using the yield of 0.36 kg of ethanol (energy density of 26.7 MJ/kg) per kilogram of dry grain and assuming an average US harvest of 10 t/ha results in a power density of 0.26 W/m2 for US corn-based ethanol.
Sugar cane is a perennial grass, but because its yields decline with successive cuttings the standard Brazilian practice is to have five harvests, followed by replanting (SugarCane.org 2013). According to Crago and co-workers (2010), for the first harvest the ethanol yield is 10,235 L/ha, diminishing to 5,636 L/ha with the fifth harvest: the average is 6,134.4 L/ ha, or 0.41 W/m2, nearly twice as high as for US corn-based ethanol. Sugar cane ethanol has other advantages: the land used to grow the grass is not usually in competition with land used for food crops; the plant's endophytic nitrogen-fixing bacteria eliminate the need for nitrogen fertilizers; and in Brazil's climate, there is no need for irrigation. The production costs are thus cheaper than for the US corn-based ethanol, but the relative competitiveness is changed by including the costs of transporting cane ethanol and the by-product (distiller's grain, corn oil) credits of corn ethanol (Crago et al. 2010).
The conversion of oil extracted from the seeds of oil plants to biodiesel is done by transesterification, that is, by reacting triglycerides in plant oils with alcohol (ethanol or methanol) in the presence of a base catalyst (Gerpen 2005). The transesterification process converts up to 97% of oil into biodiesel, and rapeseeds contain about 40% oil, which means that nearly 39% of the crop yield can end up as fuel. The crop yields mostly between 2 and 3.5 t/ha (the EU-27 mean is about 2.5 t/ha), but the average is 4 t/ha in the Netherlands. The Dutch biodiesel yield thus averages close to 1.5 t/ha, and its energy equivalent (with 37.8 GJ/t) of 56.7 GJ/ha translates to 0.18 W/m2. By contrast, the best rapeseed-based performance by a Spanish company was 0.22 W/m2 (Gonzalez-Garcia, Garcia-Rey, and Hospido 2013).
The average EU yield translates to only 0.12 W/m2, and, as in the case of the US corn-based ethanol, it is obvious that this low power density alone precludes this fuel ever supplying a significant share of the EU's large diesel demand. The rapeseed required to cover the EU-27 diesel demand of roughly 260 GW would have to be planted on nearly 220 Mha, while the EU-27 arable land adds up to only about 103 Mha (Eurostat 2012). In any case, owing to the high energy cost of farming and processing inputs, about a quarter of the EU's area could produce rapeseed biodiesel only with a net energy loss (Firrisa 2011). The biofuel with the greatest promise-but with a constantly deferred large-scale production-is cellulosic ethanol made by enzymatic hydrolysis of any phytomass high in cellulose and hemicellulose (any woody material, cereal straws, and intensively cultivated high-yielding grasses).
The latter group includes switchgrass (Panicum virginatum), a North American native tolerant of summer heat; reed canary grass (Phalaris arundinacea), an up to 2-m-high creeping C3 species from temperate Eurasia and North America; the giant reed (Arundo donax), a common Eurasian rhizomatous plant that can grow up to 8-9 m; and miscanthus (Miscanthus giganteus), originally an Asian C4 species whose
rapid growth can reach 3-4 m (4F CROPS 2011; Singh 2013). In experimental plots and in small field settings these grasses yield mostly between 10 and 20 t/ha, and dry matter yields of up to 50 t/ha have been reported for reeds, but high yields could not be maintained in large-scale plantings without adequate fertilization and necessary irrigation.
The potential of cellulosic ethanol has been chronically overestimated: completion dates of large commercial plants have been slipping for years, production costs remain uncompetitive, and the conversion of hemicellulose, which makes up 25%-36% of grass tissues (Lee et al. 2007), remains a challenge. Not surprisingly, even as the first large-scale US cellulosic ethanol plant began operating in 2014, the overall appraisal of the fuel's prospects was titled "Cellulosic ethanol fights for life" (Peplow 2014). And no future technical improvements can change the fundamentals. Even when very high yields (15 t/ha of dry matter) and an average conversion of 330 L of ethanol/t of grass are assumed (Schmer et al. 2008), the power density of cellulosic ethanol would be about 0.4 W/m2, no higher than that of Brazilian ethanol made from sugar cane, while the NREL's design of a thermochemical pathway by indirect gasification and mixed alcohol synthesis assumes a yield of 8.2 GJ/t of dry feedstock (Dutta et al. 2011), and that would translate to just 0.26 W/m2 for production based on harvesting 10 t of grain corn stover.
Finally, the power density of biogas generation, an energy conversion technique that has been pioneered in rural areas of China and India to turn organic waste into a low-energy-density gas for household use. The conversion is now used on a large commercial scale to produce gas for subsequent electricity generation. Among the affluent countries, Germany has by far the most extensive national biogas program: in 2011 it had 7,000 biogas plants with an installed capacity of about 2.7 GWe, and their feedstock was about equally divided between energy crops and livestock excrements (FNR 2012). Because of its very high water content, cattle and pig slurry generates only a small share of biogas, with most of it coming from the fermentation of phytomass. German data show an equivalent of about 100 m3 CH4/t of fresh corn silage (FNR 2012), and with 50 t/ha this would produce 5,000 m3 of methane, or 0.6 W/m2, and after conversion to electricity (producing about 18.5 MWh) the final power density would be slightly above 0.2 W/m2.
