Vaclav Smil. Making the modern world: materials and dematerialization

Preface.  I can’t believe I read this book, it is just a long litany of the  gigantic amounts of materials we exploit, with no analysis, implications, or the meaning of what impact this will have on the planet.

I certainly don’t expect anyone to read even this shortened version of his book, but it might be worthwhile to skim for an idea of how much material we’re consuming.

As I point out in my review of the United Nations 2016 report Global material flows and resources productivity” here, in order to accommodate an additional 2 billion people in 2050, material consumption will need to nearly triple to 180 billion tonnes of materials, almost three times today’s amount. If 180 billion tonnes grows in the future at a 5% compound rate, in 497 years the entire earth will be consumed, all 5.972 x 1021 tonnes of it, and we’ll be floating in outer space.

After reading this book, it’s hard to believe there’s anything left to exploit, though here it is 5 years later and the earth is still being pillaged.  But from Smil’s gargantuan numbers and the exponential exploitation of just about everything, clearly this will end badly.  The issue of peak sand has been in the news more frequently lately, which is essential for civilization to make concrete, computer chips, solar PV, and fracking.

Smil covers a wide range of materials that are essential to civilization that you may not have thought much about, and all the myriad uses of silicon, plastics, nitrogen, aluminum, steel, hydrogen, ammonia, cement, and more.  All of them made possible by oil.  All of them essential for civilization, so if one fails….(Liebig’s law of the minimum).

Nor can we avoid our predicament by recycling. Smil states that While some metals can be reused indefinitely (albeit with some mass losses) recycling of most materials often entails considerable loss of quality and functionality”.

Alice Friedemann   www.energyskeptic.com  author of “When Trucks Stop Running: Energy and the Future of Transportation”, 2015, Springer and “Crunch! Whole Grain Artisan Chips and Crackers”. Podcasts: Derrick Jensen, Practical Prepping, KunstlerCast 253, KunstlerCast278, Peak Prosperity , XX2 report

***

Vaclav Smil. 2013. Making the Modern World: Materials and Dematerialization.  Wiley.

An overwhelming majority of people lived in pre-modern societies with only limited quantities of simple possessions that they made themselves or that were produced by artisanal labor as unique pieces or in small batches – while the products made in larger quantities, be they metal objects, fired bricks and tiles, or drinking glasses, were too expensive to be widely owned. The principal reason for this limited mastery of materials was the energy constraint: for millennia our abilities to extract, process, and transport biomaterials and minerals were limited by the capacities of animate prime movers (human and animal muscles) aided by simple mechanical devices and by only slowly improving capabilities of the three ancient mechanical prime movers: sails, water wheels, and wind mills.

An updated inventory, with data for aggregate categories extending until 2006, was published in 2009 (Matos 2009) and data on individual elements, compounds, and materials are updated annually (USGS 2013).

the series does not include materials contained in traded finished goods: given their mass and variety their tracking would be very difficult.

The ships that made the first Atlantic crossings were remarkably light: a Viking ship (based on a well-preserved Gokstad vessel built around 890 CE) required the wood of 74 oaks (including 16 pairs of oars).

The Egyptian pyramids at Giza are unique: Khufu’s pyramid not only remains the largest stone structure ever built (195 m high, it required 2.5 million stones whose average weight was 2.5 t) but this mass of more than 6 Mt of stone

Romans are credited with the invention of concrete, but this is an inaccurate attribution. Concrete is a mixture of cement, aggregates (sand, pebbles), and water and cement is a finely ground mixture of lime, clay, and metallic oxides fired in kilns at a high temperature. There was no cement in Roman opus cementitium and hence this sturdy mixture, strong enough to build large vaults and domed structures, was not the material now known as concrete. Opus cementitium contained aggregates (sand, gravel, stones, broken bricks, or tiles) and water but its bonding agent was lime mortar (Adam, 1994). The combination of slaked lime and volcanic sand from the vicinity of Puteoli near Mount Vesuvius (pulvere puteolano, later known as pozzolana), produced a superior mixture that could harden even under water and that could be used to build not only massive and durable walls but also spectacular vaults.

The most consequential material development in antiquity was the ability to smelt and to shape a growing array of metals. All of this devastated local and regional wood resources, and copper smelting was a leading cause of Mediterranean deforestation, particularly in Spain and Cyprus.

[We still live in] the Iron Age, with the total consumption of other metals adding up to a small fraction of iron use.

Global population increased by less than 60% during the 500 years between 1000 and 1500 but then more than doubled (from about 460 million to nearly a billion) by 1800 – but remained overwhelmingly rural, with cities accounting for less than 5% of all humanity

Fuel-wasting fireplaces and braziers resulted in a huge demand for fuelwood and charcoal to heat the expanding cities of the pre-coal era. In Paris, the demand rose from more than 400,000 loads of wood in 1735 to more than 750,000 loads in 1789 (about 1.6 Mm3) and the same amount of charcoal, prorating to more than a ton of fuel per capita (Roche, 2000).

Wood remained indispensable not only for building houses and transportation equipment (carts, wagons, coaches, boats, ships) but also—as iron smelting rose in parts of Europe—for charcoal production for blast furnaces (substitution by coke began only during the latter half of the eighteenth century and was limited to the UK). And as Europe’s maritime powers (Spain, Portugal, England, France, and Holland) competed in building large ocean-going vessels—both commercial and naval—the increasing number of such ships and their larger sizes brought unprecedented demand for the high-quality timber needed to build hulls, decks, and masts.

With wooden hulls, masts, and spars being as much as 70% of the total mass (the remainder was divided among ballast, supplies, sails, armaments, and crew) these pioneering vessels contained 60–75 t of sawn timber (Fernández-González, 2006).

Iron production in small blast furnaces required enormous quantities of charcoal and combined with inefficient wood-to-charcoal conversion this led to widespread deforestation in iron-smelting regions: by 1700 a typical English furnace consumed 12 000 t of wood a year (Hyde, 1977).

Only during the mid 1950s that Alastair Pilkington introduced the molten tin bath that allowed production of very large pieces of flat glass with near-perfect uniformity

By 1900 the railroads on five continents added up to 775,000 km, with about 250,000 km in Europe, more than 190,000 km in the USA, 53,000 km in Russia, and 30,000 km in the UK (Williams, 2006). Given the wide range of terrains covered by rail tracks it is impossible to estimate a typical volume of bulk construction materials – earth displaced and replaced to create cuts or embankments, stone cut to create tunnels or incision in mountainsides, and stone quarried to produce gravel for access roads and rail beds – that had to be handled for an average kilometer of new track. Even a highly conservative assumption of 3000 m3/km would result in nearly 2.5 Gm of bulk materials associated with the global railway construction of the second half of the nineteenth century. A similarly conservative assumption of at least 2000 t of ballast (crushed stones packed underneath and around ties) per kilometer would translate to at least 1.5, and more likely to 2 Gt, of coarse gravel applied to hold in place the tracks built between 1830 and 1900. Mineral aggregates were also needed in unprecedented volumes for the building of new factories, for the expansion of ports, and for the construction of hard-top roads.

All ties (sleepers) installed during the nineteenth century were wooden; concrete sleepers were introduced only around 1900 but remained uncommon until after World War II. Standard construction practice requires the placement of about 1900 sleepers per km of railroad track, and with a single tie weighing between roughly 70 kg (pine) and 100 kg (oak) every kilometer needed approximately 130–190 t of sawn (and preferably creosote-treated) wood. My calculations show that the rail tracks laid worldwide during the nineteenth century required at least 100 Mt of sawn wood for original construction and at least 60 Mt of additional timber for track repairs and replacements (Smil, 2013).

Rails used during the nineteenth century weighed between 20 and 30 kg/m and, assuming an average of 25 kg/m, the railway construction between 1850 and 1900 would have required about 20 Mt of steel, while replacement would have more than doubled that total. Steel became the favorite material for railway bridges:

Because of their renewability, annually harvested crop residues used to be indispensable materials in all traditional agricultural societies. In many deforested regions they were the only source of household fuel, straw–clay mixtures were made into bricks and straw bundles were used for roof thatching, in some countries peasants wore straw sandals and coats, and cereal straws were used as both feed and bedding for domesticated ruminants

There are no reliable data about the final fate of crop residues: in many agroecosystems they should be directly recycled to maintain soil organic matter and to prevent erosion, but often their mass is judged to be excessive and they are simply burned in fields. This undesirable practice is particularly common in rice-growing regions of Asia. Straw continues to be burned even in some affluent countries, most notably in Denmark where about 1.4 Mt of wheat straw (nearly a quarter of the total harvest) is used for house heating or even in centralized district heating and electricity generation (Stenkjaer, 2009).

A global aggregate of around 40 EJ in 2000 is thus a good consensus value and implies a nearly 70% increase in biomass fuel demand between 1950 and 2000 and a doubling of wood and crop residue harvests during the twentieth century. But the intervening high population growth greatly reduced the average per capita consumption and the huge expansion of fossil fuel extraction cut the biofuel share from 50% in 1900, to less than 10% of global primary energy supply in the year 2000, and (because of inferior efficiencies of wood and straw combustion) to less than 5% in terms of useful final energy supply. Among the major economies, wood has the highest national share of primary energy supplies in Brazil, at about 10%, while its share in affluent nations ranges from negligible values (just 1% in the UK and Spain) to about 20% in Sweden and Finland, with the US share falling from about 4.5% in 1950 to just 2% in 2010

Wooden railway ties, that quintessential nineteenth-century innovation, maintained their high share of the global market throughout the twentieth century. During the 1990s, 94% of America’s ties were wooden.

 

Better treatment of ties prolonged their average lifespan from about 35 years in 1940 to 40–50 years by the year 2000 (James, 2001). European and North American tie markets have been basically limited to replacements, mostly reinforced concrete

But most reinforced concrete has not gone into iconic structures but into ever-increasing numbers of nondescript or outright ugly (or brutal looking) apartment buildings, high rises, factories, garages, roads, overpass bridges, and parking lots.

Much more steel (in the form of sheets and rods) has gone into cars and trucks and new transportation infrastructures on land (ranging from multi-lane highways and bridges to new airports) and into the construction of large oil tankers, bulk carriers (transporting anything from grain to ores), and, starting in the 1960s, container ships and ports. Steel allows particularly captivating design of long suspension bridges with woven cables supporting lengthy road spans:

The transportation sector also became the leading user of aluminum: the combination of light weight and durability made the metal, and its alloys, an ideal choice for applications ranging from cooking pots to rapid train cars,

The fourth most important metal has been zinc, with a consumption of 12.6 Mt in 2010; but the steadily rising demand for lead has brought this formerly more distant number five close to the zinc total: in 2011 the global refined lead supply surpassed, 10 Mt for the first time, to reach 10.6 Mt, with about 45% being primary metal and the rest coming from recycled material

With a total of just over 1 billion cars and light and heavy trucks, and with an average mass of 10 kg Pb in automobiles and 13 kg Pb in truck batteries, there was nearly 11 Mt of lead on the world’s roads in 2010.

Silicon makes nearly 28% of the Earth’s crust, and while it is abundantly present as SiO2 (silica) in sand, sandstone, and quartz and in many silicates ranging from hard feldspars (rock-forming minerals) to soft kaolinite (a layered clay mineral), it is never found in pure, unbound elementary form. But the purest crystalline silicon is the material foundation of modern electronics: intricate webs of semiconductors

Global production of all plastics

265 Mt in 2010:

we could not have supported the twentieth century global increment of 4.5 billion people consuming increasingly better diets without a huge increase in nitrogen applications.

Global output of synthetic fertilizers (in terms of pure N): 85.13 Mt in and 2000, an increase of two orders of magnitude (roughly 570 times) in 80 years.

Global output of synthetic fertilizers (in terms of pur nitrogen) rose from just 150,000 tonnes in 1920 to 3.7 million tonnes in 1950 and 85.13 million tonnes in 2000, an increase of two orders of magnitude (roughly 570 times) in 80 years.

Remarkably that was not even an exceptionally large gain, as the global production of other new materials saw even greater increases over the course of the 20th century:

  • three orders of magnitude for aluminum (roughly 3600 times, from just 6800 t in 1900 to 24.3 Mt in 2000)
  • four orders of magnitude for plastics (from about 20,000 t in 1925 to 150 Mt in 2000).
  • 30 times more production of paper and steel (from 28.3 to 850 Mt)
  • 27 times more copper (from 495 000 t to 13.2 Mt)

In comparison, the global population increased 3.8 times between 1900 and 2000, and the gross world product (in constant monies) rose about 20-fold,

The annual output of bovine (cattle and water buffalo) hides surpasses 6 Mt, that of sheep and lambskins over 400,000 t, and some 300,000 t of goat and kidskins are turned into leather product annually (FAO, 2011). Production of wool, the most important animal fiber, rose from about 960 000 t in 1950 to 2.9 Mt in 1970, fluctuated afterwards (peaking at 3.3 Mt in 1990), and declined to just below 2 Mt in 2011 (FAO, 2013). In contrast, production of silkworm cocoons has more than doubled during the past 50 years, to about 500 000 t in 2010.

30% of humanity continues to live in structures whose material, locally available clay, has not undergone any elaborate processing and that can be made without any modern energy inputs.

Production of all durable soil- or earth-based materials requires firing in kilns, with temperatures ranging from less than 500 °C for low-quality bricks to as much as 1100 °C for ceramic tiles, 1300 °C for vitrified bricks, and 1400 °C for glass, while the pyro-processing of Portland cement requires 1400–1450 °C (Berge, 2009).

Sequential washing, screening, crushing, and dewatering eliminate any organic matter and clay and produce a specific coarseness of material with low moisture. The best available estimates indicate that, in the USA, 41% of construction sand and gravel ends up as concrete aggregates, a quarter of the total is destined for road building, 13% for construction fill, and 12% for asphaltic concrete and similar mixtures (USGS, 2012). The small remainder is used for filtration, snow and ice control on roads (some municipalities also use salt), railroad ballast and golf courses, as well as for replenishment of eroding beaches

The construction of the US Interstate Highway System was a major component of this rising demand (USGS, 2006). About 60% of these multi-lane highways are paved in concrete whose standard thickness is 28 cm and hence 1 km of a four lane highway (each lane is 3.7 m wide) requires about 4150 m3. This adds up to roughly 10,000 t of concrete for every kilometer and the entire system of 73,000 km embodies about 730 Mt of concrete in driving lanes, with more emplaced in shoulders, medians, approaches, and overpasses.

Global compilations of CO2 emissions from the cement industry show its contribution almost 5% in 2010 (CDIAC, 2013).

Concrete (particularly its reinforced form) is now by far the most important manmade material both in terms of global annual production and cumulatively emplaced mass.

While this material provides shelter and enables transportation and energy and industrial production, its accumulation also presents considerable risks and immense future burdens. These problems arise from the material’s vulnerability to premature deterioration that results in unsightly appearance, loss of strength, and unsafe conditions that sometimes lead to catastrophic failures, and whose prevention requires expensive periodic renovations and eventually costly dismantling. Concrete, both exposed and buried, is not a highly durable material and it deteriorates for many reasons (AWWS, 2004; Cwalina, 2008; Stuart, 2012). Exposed surfaces are attacked by moisture and freezing in cold climates, bacterial and algal growth in warm humid regions (biofouling recognizable by blackened surfaces), acid deposition in polluted (that is now in most) urban areas, and vibration. Buried concrete structures (water and sewage pipes, storage tanks, missile silos) are subjected to gradual or instant overloading that creates cracks, and to reactions with carbonates, chlorides, and sulfates filtering from above. Poor-quality concrete can show excessive wear and develop visible cracks and surficial staining due to efflorescence in a matter of months. Alternations of freezing and thawing damage both the horizontal surfaces (roads, parking) that collect standing water, as well as vertical layers that collect water in pores and cracks. While concrete’s high alkalinity (pH of about 12.5) limits the corrosion of the reinforcing steel embedded in the material, as soon as that cover is compromised (due to cracks or defoliation of external layers) the expansive corrosion process begins and tends to accelerate. Chloride attack (on structures submerged in seawater, from deicing of roads, in coastal areas from NaCl present in the air in much higher concentrations than inland) and damage by acid deposition (sulfate attack in polluted regions) are other common causes of deterioration, while some concretes exhibit alkali-silica and alkali-carbonate reactions that lead to cracking. Unsightly concrete blackened by growing algae embedded in the material’s pores is a common sight in all humid (especially when also warm) environments. Given the unprecedented rate of post-1990 global concretization, it is inevitable that the post-2030 world will face an unprecedented burden of concrete deterioration.

 

This challenge will be particularly daunting in China, the country with by far the highest rate of new concrete emplacement, where the combination of poor concrete quality, damaging natural environment, intensive industrial pollutants, and heavy use of concrete structures will lead to premature deterioration of tens of billions of tons of the material that has been poured into buildings, roads, bridges, dams, ports, and other structures during the past generation. Because maintenance and repair of deteriorating concrete have been inadequate, the future replacement costs of the material will run into trillions of dollars. To this should be added the disposal costs of the removed concrete: some concrete structures have been recycled but the separation of the concrete and reinforcing metal is expensive. The latest report card on the quality of American infrastructure gives poor to very poor grades to all sectors where concrete is the dominant structural material:

with an estimated investment of at least $3.6 trillion needed by 2020 in order to prevent further deterioration (ASCE, 2013).

Transposed to post-2030 China, this reality implies the need for an unprecedented rehabilitation and replacement of nearly 100 Gt of concrete emplaced during the first decade of the twenty-first century, at a cost of many tens of trillions of dollars.

The world’s impervious surface area (built-up, paved) at about 580 000 km2: that is less than 0.5% of ice-free surface, but an area equal to Kenya. In per capita terms, high-income countries in northern latitudes had the largest areas of impervious surfaces (Canada 350 m2, USA 300 m2, Sweden 220 m2)

Of course, not all impervious surfaces are concrete but the material accounts for their largest share.

 

In 2010, humanity put in place close to 40 Gt of them (dominated by 33 Gt of concrete and 4.5 Gt of bricks), an equivalent of at least 17 km3. For comparison, the volume of one of the world’s best known mountains, Japan’s Fuji, is about 400 km3

By the year 2000 the global output of iron ore, pig iron, and steel had reached new global records: at 1 Gt/year iron ore extraction was surpassed only by the output of fossil fuels and bulk construction materials, pig (cast) iron production rose to nearly 600 Mt, and at roughly 850 Mt/year steel output was about 30 times higher than in 1900. That total was also almost 20 times larger than the aggregate smelting of aluminum, copper, zinc, lead, and tin, and in per capita terms it rose from less than 20 to about 140 kg/year. Demand for copper increased by a similar rate (27-fold, to 13.2 Mt) and zinc production rose almost 20-fold, from about 480 000 t to 8.77 Mt (Kelly and Matos, 2013).

Gold output rose nearly 7-fold, but in absolute terms it amounted only to about 2600 t in the year 2000, compared to 18,100 t for silver,

Polyethylene (PE) is by far the most important thermoplastic (it accounted for 29% of the world’s aggregate plastic output, or roughly 77 Mt, in 2010), polypropylene (PP) comes next (with about 19% or 50 Mt in 2010), followed by polyvinyl chloride (PVC, about 12% or 32 Mt in 2010).

In 2010, packaging consumed almost 40% of the total (mostly as various kinds of PE and PP), construction about 20% (mostly for plastic sheets used as vapor barriers in wall and ceiling insulation), the auto industry claimed nearly 8% (interior trim, exterior parts), and the electrical and electronic industry took about 6% (mostly for insulation of wires and cables).

All of these products begin as ethane. In North America and the Middle East ethane is separated from natural gas, and low gas prices and abundant supply led to surplus production for export and favored further construction of new capacities: in 2012 Qatar launched the world’s largest LDPE plant and, largely as a result of shale gas extraction, new ethylene capacities are planned in the USA (Stephan, 2012). The dominant feedstock for ethane in Europe, where prices of imported natural gas are high, is naphtha derived by the distillation of crude oil.

Transparent or opaque bags (sandwich, grocery, or garbage), sheets (for covering crops and temporary greenhouses), wraps (Saran, Cling), and squeeze bottles (for honey), HDPE garbage cans, containers (for milk, detergents, motor oil), and toys (including Lego bricks). Among a myriad of hidden PE applications are HDPE for house wraps (Tyvek) and water pipes; PEX for water pipes and as insulation for electrical cables; and UHMWPE for knee and hip replacements.

Other plastic uses:

  • massive LDPE water tanks
  • indoor–outdoor carpeting to lightweight fabrics woven from PP yarn and used particularly for outdoor apparel,
  • insulated wires, water, and sewage pipes to food wraps and her car’s interior and body undercoating
  • disposable and surgical gloves, flexible tubing for feeding, breathing and pressure monitoring, catheters, blood bags, IV containers, sterile packaging, trays, basins, bed pans and rails, thermal blankets, lab ware (Smil, 2006, p. 131)
  • construction (house sidings, window frames), for outdoor furniture, water hoses, office gadgets, toys,

Plastics have a limited lifespan in terms of functional integrity: even materials that are not in contact with earth or water do not remain in excellent shape for decades. Service spans are no more than 2–15 years for PE, 3–8 years for PP, and 7–10 years for polyurethane; among the common plastics only PVC can last two or three decades and thick PVC cold water pipes can last even longer (Berge, 2009).

[In conclusion, then, it is clear] plastics, [and the fossil fuels they are derived from], are indispensable for the functioning of modern civilization.

Industrial Gases

The three most important elements – oxygen, hydrogen, and nitrogen – deserve such ranking because without them we could not produce steel in the most efficient way, and could not have our modern petrochemical and nitrogen fertilizer industries. Other elements and compounds classified as industrial gases include acetylene, argon, carbon dioxide, helium, neon, and nitrous oxide.

Without the synthesis of ammonia (predicated on large-scale supply of pure nitrogen) we would not be able to feed billions of people, and without oxygen we could not produce most of the world’s most important alloy. Ammonia synthesis is the world’s largest consumer of nitrogen: in 2010 it required 130 Mt of the gas (about 112 Gm3 of N2). Nitrogen’s other key uses as a feedstock include ammonia for the synthesis of nitric acid, hydrazines, and amines.

 

Nitrogen cooling of metal parts enables tight assembly fits and, in reverse, it allows the taking apart of closely-fitted parts. With the expansion of modern electronics, nitrogen found a new market in those instances (particularly during soldering) when it is necessary to reduce the presence of oxygen and to maintain a clean atmosphere (by 1985 this use claimed 15% of US consumption).

Ferrous metallurgy is by far the largest user of oxygen: the gas is blown into blast furnaces, EAFs, and BOFs

Chemical syntheses (above all ethylene oxidation) are the second largest market, and oxygen is also used in smelting color metals (lead, copper, and zinc furnaces), in the construction material industries (producing a more intense flame and reduced fuel use in the firing of glass, mineral wool, lime, and cement),

Argon, the cheapest truly inert gas, goes into incandescent and fluorescent lights

Hydrotreating, hydrodesulfurization, and hydrocracking used to process roughly 3.7 Gt of oil in 2010 claimed (assuming that H2 demand averaged 0.5% of the total crude input, or roughly 60 m3/t) about 20 Mt of the gas.

Hydrogen

Industrial gases are used in sectors that account for more than half of the world’s economic output and the value of their production has been growing faster than the growth rate of the global economy: in 2000 their global market was worth about $34 billion, a decade later it had nearly doubled as it exceeded $60 billion, and it is heading to about $80 billion by 2015

Liquid hydrocarbons (principally naphtha) are the feedstock for hydrogen production in crude oil refineries where the gas is needed for the catalytic conversion of heavier fractions to lighter fuels, and also in order to comply with ever stricter environmental regulation and to desulfurize the refined products.

Synthesis of ammonia remains the leading user of hydrogen, followed by refinery needs

Post-1950 expansion was rapid, with global ammonia synthesis rising from less than 6 Mt in 1950, to about 120 Mt in 1989, 164 Mt in 2011 (USGS, 2013).

Two-thirds (65–57%) of all synthesized NH3 has been recently used as fertilizer, with the total global usage more than tripling since 1970, from 33 to about 106 Mt N in 2010. Because ammonia is a gas under ambient pressure, it can be applied to crops only by using special equipment (hollow steel knives), a practice that has been limited to North America. The compound has been traditionally converted into a variety of fertilizers (nitrate, sulfate) but urea (containing 45% N) has emerged as the leading choice, especially in rice-growing Asia, now the world’s largest consumer of nitrogenous fertilizers; ammonium nitrate (35% N) comes second.

Compared to traditional harvests, the best national yields of these three most important grain crops have risen to about 10 t/ha for US corn (from 2 t/ha before World War II), 8–10 t/ha for European wheat (from about 2 t/ha during the 1930s), and 6 t/ha for East Asian rice (from around 2 t/ha).

 

High-yielding US corn now receives, on average, about 160 kg N/ha, European winter wheat more than 200 kg N/ha, and China’s rice gets 260 kg N/ha, which means that in double-cropping regions annual applications are about 500 kg N/ha. According to my calculations, in the year 2000 about 40% of nitrogen present in the world’s food proteins came from fertilizers that originated from the Haber–Bosch synthesis of ammonia (Smil, 2001).

The rising use of nitrogen had to be accompanied by a rising use of the other two essential macronutrients

Agricultural phosphate consumption: 20.3 Mt P in 2010.

Potassium is obtained mostly by underground mining of sylvinite, a mixture of about a third KCl and two-thirds NaCl; Saskatchewan has the largest reserves of the rock and is the leading global producer. Worldwide extraction (expressed in terms of K2O equivalent) rose to nearly 34 Mt by 2010, with Canada (nearly 10 Mt) and Russia (more than 6 Mt) being the largest producers and worldwide exporters. About 85% of all KC ends up as fertilizer.

Silicon

The raw material for producing silicon is abundant, but an energy-intensive high-temperature deoxidization with carbon – SiO2 + 2C + Si + 2CO (using graphite electrodes in electric furnaces) – is required to yield element that is 99% pure. But even 99% purity is quite unacceptable for solar and electronic industries, and hence the metallurgical-grade Si has to undergo elaborate and costly processing that makes it many orders of magnitude purer in order to meet the specifications for producing semiconductors, solar cells, and optical fibers (Föll, 2000).

In 1965, when the number of transistors on a microchip had doubled to 64 from 32 in 1964, Gordon Moore predicted that this rate of doubling would continue,

By 2012 the count reached 5 billion in Xeon Phi Coprocessor (Intel, 2012). Mass deployment of these increasingly powerful microprocessors in conjunction with increasingly capacious memory devices has transformed every sector of modern economies thanks to unprecedented capacities for communication, control, storage, and retrieval of information.

Wafer shipments for semiconductor applications rose from just $4 billion in 1977 to  $292 billion in 2012 (SIA, 2013).

During the first decade of the twenty-first century, electronics ceased to be the major consumer of high-grade silicon as most of that material now ends up in PV cells.

There are hundreds of PV-powered satellites used for weather and Earth monitoring, telecommunication, and spying;

The best commercially available models are rated at 19–22% (NREL, 2013; Solarplaza, 2013). For decades, PV cells were made with off-grade polycrystalline material that was not good enough for electronic applications, but as the heavily subsidized market for PV installation rose from less than 100 MW/year in 1995 to more than 10 GW/year in 2009, it was necessary to divert increasing amounts of purified polycrystalline metal into the solar cell industry. In 1997 the industry used only 800 t of such metal, by 2009 it required 69,100 t, three times as much as consumed by electronics, to produce about 44,500 t of solar cells, mostly by the casting of polycrystalline metal (Takiguchi and Morita, 2011).

While some metals can be reused indefinitely (albeit with some mass losses) recycling of most materials often entails considerable loss of quality and functionality.

Increasing burdens of environmental pollution and the critique of economic thinking that tended to ignore such matters. Ayres et al. (1969, pp. 283–84), describing the reality in clear physical terms, noted that such omissions “may result in viewing the production and consumption processes in a manner that is somewhat at variance with the fundamental law of the conservation of mass,” and pointed out the obvious consequences for the environment, namely that in the absence of trade and net stock accumulation “the amount of residuals inserted into the natural environment must be approximately equal to the weight of basic fuels, food, and raw materials entering the processing and production system, plus oxygen from the atmosphere.” But it took nearly two decades before this admonition was transformed into the first fairly comprehensive studies of material requirements on a national level, as it was only during the late 1990s that several research teams began to reconstruct direct material inputs (DMIs) as well as outflows, and total material requirements (TMRs) of the world’s leading affluent economies. Fischer-Kowalski et al. (2011)

There are other approaches to the investigation of material flows; one attempts to trace the life-cycles of individual commodities on a national, regional, or global level; another looks at the energy costs of commodities and products; and yet another traces the environmental impacts of their production, use, and abandonment (or recycling). Life-cycle assessments (or analyses, in either case the acronym is LCA) have been performed at different scales for many elements and compounds – for example, chlorine by Ayres (2008) and polyvinyl chloride (PVC) by the European Commission (EU, 2004) – and for products ranging from aluminum cans (Mitsubishi, 2005) to steel truck wheels (Alcoa, 2012).

Limiting the account to DMI will greatly underestimate the overall resource demand in all modern economies engaged in intensive international trade, and particularly in such major powers as the USA, Germany, or Japan that rely on imports for large shares of many materials. Correcting this by the inclusion of net imports of all raw materials is only a partial (and increasingly deficient) solution, because many metals and other minerals are not imported in the form of ores or concentrates or bulk shipments but are instead embodied in finished products. Identifying the specific material content of these products (even their limited inventory would run to many hundreds of individual machines, tools, components, and consumer items) presents a major challenge – but the adjustment should not end there, as many items imported from a particular country contain components made of materials in a number of other countries that, in turn, imported parts or raw materials from yet another country or, more likely, a set of countries.

The global on-line database and most of the global and national studies of material flows have been produced by a small group of researchers from Austria and Germany and that most of them have been published in just two sources, in the Journal of Industrial Ecology and in Ecological Economics.

I question the utility of constructing these all-encompassing national or global flow accounts because I am not sure what other revealing conclusions to derive from these summations of disparate input and output categories besides the obvious confirmations of substantial differences in national aggregates and in the rates of long-term growth. Of course, the maximalist aggregates of the all-encompassing variety also have an undoubted heuristic and curiosity value and they do convey the truly massive scale of global mobilization of raw materials.

Half a dozen studies of global material extraction at the beginning of the twenty-first century, that include all harvested biomass, all fossil fuels, ores and industrial minerals, and all bulk construction materials (but exclude hidden flows, water, or oxygen), cluster fairly tightly around 50 Gt/year. This is hardly surprising given the fact that these studies derive the flows from the same sets of data:

with roughly 18 Gt coming from biomass, 10 Gt from fossil fuels, nearly 5 Gt from ores and other minerals, and more than 17 Gt from bulk construction materials.

given the uncertainties in estimating the mass of bulk construction minerals (above all for the extraction of sand and gravel) that account for at least two-thirds of the all material flows, the mass of 0.5 Gt is well within the minimal range of estimation error

and the total of roughly 25 Gt thus remains my preferred aggregate of directly used global materials in the year 2000. That total prorates to just over 4 t of materials per person (the global population was 6.08 billion in the year 2000), with at least 2.5 t (and perhaps as much as 3 t) accounted for by bulk construction materials and only about 0.8 attributed to all metals and nonmetallic minerals. These rates compare to nearly 1 t of food and feed crops (fresh weight), close to 0.5 t of wood (excluding fuelwood), and about 1.7 t of fossil fuels (roughly 0.8 t of coal, 0.6 t of crude oil, and 0.3 t of natural gas) extracted for every inhabitant of the world in the year 2000.

 

pre-1950 global totals are nothing but questionable estimates, and even the recent aggregates depend critically on what is included. For example, Krausmann et al. (2009) put the worldwide biomass extraction (crops, their residues, roughages, and wood) at 19.061 Gt in 2005, while in my detailed account of phytomass harvest (Smil, 2013) I showed that in the year 2000 the total for woody phytomass alone could be anywhere between 2 and 13.4 Gt depending on the boundaries chosen for the analysis.

Consequently, there can be no single accurate total, as the search for global totals will be always determined by assumptions, and even if everybody agrees on common boundaries the basic results will be largely predictable. Physical realities dictate that the mass of sand and gravel used to emplace and maintain modern concrete-based infrastructures must be substantially greater than the mass of metallic ores; and that the mass of iron, a metal of outstanding properties produced from abundant ores with a moderate energy intensity, must be orders of magnitude higher than the mass of titanium, an even more remarkable metal but one derived from relatively rare ores with a great energy expense. At the same time, it must be kept in mind that data for inexpensive, readily available bulk construction materials (particularly for sand and gravel) that are usually sold not far from their points of extraction are generally much less reliable than the statistics for metal ores and industrial nonmetallic minerals that are globally traded.

the world now consumes in one year nearly as much steel as it did during the first post-World War II decade, and (even more incredibly) more cement than it consumed during the first half of the twentieth century.

little guidance for future decision-making (beyond the obvious point that the recent high growth rates cannot continue for many decades).

Useful insights can be gained from two kinds of finer focus: through closer examination of material flows on the national level, and by putting more restrictive analytical boundaries on the set of examined materials and tracing the flows of individual commodities with some clear goals in mind. This can be done by detailing their uses, dispersal, and persistence in a society, and by attempting life-cycle analyses of those materials that circulate on a human timescale, that is by quantifying their direct and indirect requirements for energy or by identifying and assessing the environmental impacts of their production and use.

During the twentieth century, natural growth potentiated by immigration increased the US population nearly 4-fold (3.7), and the country’s GDP (expressed in constant monies) was 26.5 times higher in 2000 than in 1900: not surprisingly, the combination of these two key factors drove absolute consumption increases in all material categories, with the multipliers ranging from 1.7 for materials produced by agriculture to more than 90 for nonrenewable organics (and 8 for primary metals, 34 for industrial minerals, and 47 for construction materials). The importance of renewable materials (wood, fibers, leather) fell from about 46% of the total mass (when bulk construction materials are included) or 74% (with stone, sand, and gravel excluded) in 1900 to just 5% (or 22%) for analogical rates in the year 2000, a trend that was expected given the increasing reliance on light metals and plastics. Aggregate wood demand rose less than 1.4-fold during the twentieth century, but consumption of primary paper and paperboard multiplied about 19 times and was supplemented by rising quantities of recycled paper: when the data collection in the latter category began in 1960, recycled paper accounted for about 24% of all paper and paperboard use, but by the year 2000 its share was up to 46% even as large quantities of waste paper are exported (in 2000 this amounted to about 22% of all domestic collections), primarily to China (FAO, 2013).

The fact that bulk minerals used in construction (crushed stone, sand, and gravel) have increasingly dominated America’s annual flows during the twentieth century – in 1900 they accounted for 38% all materials, by 2006 their share reached 77% – is not surprising given the enormous expansion of material-intensive transportation infrastructures after World War II. Construction of the Interstate system began in 1956 and required the building of many new bridges (USDOT, 2012), while the introduction of commercial jetliners led to the rapid expansion of airports, a process recently repeated in China. Large demands for bulk construction materials also came from the building of new container ports, stream regulation (above all in the Mississippi basin), electricity generation (hydroelectric dams, nuclear power plants), new factories, commercial real estate (warehouses, shopping centers), and housing. The mass of construction materials used in the USA rose about 7-fold between 1900 and 1940 and then doubled between 1945 and 1951, doubled again by 1959 to 1.1 Gt, but the next doubling, to 2.26 Gt, took until 1997.

End-use data indicate that the largest identifiable category of sand and gravel consumption (about a fifth of the total surpassing 1 Gt/year) is as aggregate added to cement in the production of concrete, followed by road base and coverings, fill, and as aggregate added to asphaltic and bituminous mixtures; but unspecified uses make up the largest category, accounting for about a quarter of the total. Differently-sized aggregates used in the production of concrete are also the leading final uses for crushed stone, and railroad ballast is another indispensable application. With a ballast minimum depth of 15 cm and up to 50 cm for high-speed lines, and an overall width of roughly 4.5 m, this amounts commonly to more than 1000 m3/km or (with density of 2.6 t/m3) to around 3000 t of crushed stone per kilometer.

In comparison to construction sand, the total use of industrial sand is minuscule but qualitatively very important. Annual consumption has recently fluctuated around 25 Mt/year: about 40% of this total is pure silica used in glassmaking, and a fifth goes to foundries to make moldings and refractories as well as silicon carbide for flux and metal smelting. Smaller but functionally irreplaceable uses include abrasives used in blasting and sanding, sands for water filtration, and sands for creating artificial beaches and sporting areas. A new, and rapidly rising, market is for the special kinds of sands used in hydraulic fracturing of gas- and oil-bearing shales, well-packing, and cementing.

 

390 Mt in 2010. Its largest constituents include salt (about 55 Mt in 2010), phosphate rock (about 30 Mt), nitrogen (about 14 Mt), and sulfur (about 11 Mt). America’s salt consumption is remarkably high (20% of the world total in 2010); the two dominant uses (each about 18 Mt in 2010) are production of alkaline compounds and chlorine, and for road deicing; amounts an order of magnitude smaller (both about 1.8 Mt) are used in food production and in animal feed, and more than 1 Mt/year is used in water treatment (in water-softening to remove mineral ions).

Consumption of 10.3 Mt of primary metals in 1900

by the year 2000, with total metal consumption at nearly 144 Mt, they supplied 44%.

I must reiterate that actual domestic US consumption of virtually all metals is, often significantly, higher than shown by the USGS balances because substantial amounts of various metals reach the country embedded in products, and that presence is excluded from nationwide aggregates of apparent domestic consumption. Major components of these unaccounted flows include not only such leading metals as steel and aluminum in cars, airplanes, machinery, and appliances, but also such toxic heavy metals as lead in automotive lead-acid (PbSO4-H2SO4) batteries and cadmium in rechargeable Ni-Cd batteries.

In 1900, the total consumption of nonrenewable organics (mostly paving materials and lubricating oils) was less than 2 Mt, but subsequent extension of paved highways, mass ownership of cars, the rise of the trucking industry, and, above all, rapid expansion of crude oil- and natural gas-based synthetic materials made this the fastest growing material category in the USA: by 1950 the flow surpassed 30 Mt and by 1999 it had reached 150 Mt, with nearly two thirds being hydrocarbon feedstocks (naphtha and natural gas) used to make ammonia, the starting compound for all synthetic nitrogen fertilizers. The second largest input by mass is asphalt and road oil; consumption of these paving and surfacing materials rose from less than 10 000 t in 1900 (when few paved roads existed outside cities) to 100 times that mass in less than two decades, it reached more than 10 Mt by 1950 and, until the 2008 recession, it was on the order of 30 Mt/year.

In aggregate terms, the USGS accounts translate to a domestic consumption of about 1.9 t/capita in 1900, 5.6 t in 1950, and 12 t/capita in the year 2000; after leaving out bulk construction materials these rates are reduced, respectively, to 1.2, 2.3, and 3 t/capita, which means that the use of construction materials rose from about 0.7 t/capita in 1900 to 3.3 t in 1950, and 9 t in the year 2000. Wood is the only material category showing a century-long decline of per capita consumption, from about 800 kg in 1900 to about 400 kg by 1950 and about 300 kg/capita in 2000. Materials produced by agriculture rose slightly from 40 to 47 kg/capita during the first half of the twentieth century, but afterwards they declined to just 18 kg

Consumption of all metals has shown a similar pattern, rising from 135 kg in 1900 to 515 kg/capita in 1950, but by the year 2000 were essentially the same at 510 kg/capita (once more a somewhat misleading rate given the country’s large post-1970 net imports of cars, airplanes, and machinery).

In comparison with the USA, the EU-27 has similar metal consumption (0.4 vs. 0.5 t/capita) but a much lower demand for construction minerals (4.6 vs. nearly 10 t/capita), a difference that is due mostly to the continent’s much higher population density and more compact transportation infrastructure.

 

According to official statistics, between 1980 and 2010 China’s annual rate of economic growth was only below 5% three times (1981, 1989, and 1990) and was above 10% 16 times, while the average for the three decades was 9.6% (IMF, 2013). This implies a doubling every 7.3 years resulting in a 2010 GDP (in constant prices) 17.8 times higher than in 1980. In per capita terms, the multiple was still roughly 13-fold (NBSC,

The pace of China’s frenzied concretization and its overall scale has been stunning. In 1980 the country produced just short of 80 Mt of cement, a decade later it had more than doubled the total to about 210 Mt, by the year 2000 it rose to 595 Mt and by 2010 that total had tripled and reached 1.88 Gt (nearly 24 times the 1980 total and 57% of the global production for less than 20% of the world’s population), and it rose further to 2 Gt in 2011 (NBSC, 2013).

such a pace of construction guarantees that a substantial share of newly poured concrete will be of substandard quality, a conclusion confirmed by the obvious dilapidation of China’s concrete structures built during the late 1980s and the early 1990s,

The quality of concrete used to construct many of China’s new dams (by 2010 the total stood at more than 87,000 structures of all sizes including the world’s largest dam, Sanxia) is of particular concern, even more so as thousands of them are located in areas of repeated, vigorous seismic activity.

raw steel output rose 17.2 times between 1980 and 2010, from 37.1 Mt in 1980 to 637.4 Mt in 2010, when it accounted for nearly 45% of the global output (WSA, 2013). But as the extraction of iron ores increased about 14 times (from 75 000 t to 1.07 Gt) an increasing share of this output has come from imported materials. In 2010, China imported 618 Mt of iron ore, more than a third of the total input into its blast furnaces, and it has been by far the largest iron ore importer (nearly 60% of the global total and close to 70% of the domestic demand) with Australia and Brazil being the major suppliers. And while the country has been the near-monopolistic exporter of rare earths and a major exporter of molybdenum and magnesium (also of graphite), it has been also the world’s largest importer of bauxite (44 Mt in 2010) and, at nearly 1.2 Mt in 2010, of copper ores and concentrates.

the material category that has seen the greatest production increase has been the synthesis of plastics, with a nearly 70-fold rise between 1980 and 2010. Of course, that large multiple is due to a rapid development from a very low base (less than 900 000 t in 1980) but the absolute output of 62 Mt in 2010 was larger than the production of about 57 Mt in EU27 (Europe Plastics, 2011).

The need to secure more food and better nutrition for a still-growing population has led to substantial gains in the production (and imports) of fertilizers. New Haber–Bosch plants were added to raise the output of nitrogenous fertilizers from 10.3 Mt N in 1980 to 45.2 Mt in 2010, but in 2009 the record output was 48.6 Mt N, a nearly 5-fold increase in three decades, while production of phosphate fertilizers posted a roughly 8-fold increase to 19 Mt. Disparity between N and P growth rates is explained by China’s attempt to move away from excessive nitrogen uses toward more balanced fertilization with N:P:K ratios improving the efficiency of applications. As a result, China has been buying record amounts of potash from Canada. China has also become a prominent importer of materials for recycling, and the USA has been their greatest supplier. In 2010, Chinese imports of waste paper were nearly 25 Mt/year, with the USA as the leading exporter (Magnaghi, 2011). Similarly, in 2010 China bought almost 6 Mt of scrap steel – becoming the world’s third largest importer of the material after Turkey and South Korea (WSA, 2013) – with the USA again as the leading supplier. This trade is certainly one of the most remarkable indicators of changing national fortunes, as the world’s largest affluent economy has become the primary supplier of waste materials to the second largest economy experiencing a rapid rate of growth. In 2011 the USA exported more than $11 billion of waste and scrap (materials belonging to the 910 category of the North American Industry Classification System) to China. This was less than the exports of transportation equipment or agricultural products – but more than the exports of all nonelectric machinery and more than five times as much as the shipments of all electrical equipment and appliances (Smil, 2013). China is also the world’s largest importer of plastic and electronic waste.

Before any materials can start flowing through economies, energies must flow to power their extraction from natural deposits or their production by industrial processes ranging from simple mechanical procedures to complex chemical reactions. These energies belong to two distinct streams: direct flows of fuels and electricity used to energize the production processes (producing mechanical energy, heat or pressure and lighting, and electronically controlling a process) and indirect flows (embedded energies) needed to produce the requisite materials, machines, equipment, and infrastructures.

the energy needed to smelt a ton of iron from its ore in a blast furnace (as coke and supplementary coal, gas, or oil) will be vastly greater than energy embedded in the furnace’s steel, lining, and charging apparatus and prorated per unit of output. Modern blast furnaces can operate without relining for two decades, and during that time can produce tens of millions of tons of hot metal. Similarly, the energy needed to create the combination of high temperature and pressure that is required by many chemical syntheses will be far greater that any prorated energy embedded in the initial construction of reaction vessels, pipes, boilers, compressors, and computerized controls. This explains why the second category of flows is almost always neglected

Most appraisals of energy costs have followed one of two distinct approaches: either a quantification based on input–output tables of economic activities, or a process analysis that traces all important energy flows needed to produce a specific commodity or manufactured item. In the first instance, relevant prices are used to convert values of energy flows in a matrix of economic inputs and outputs (for major industrial sectors or, where available, disaggregated to the level of product groups or individual major products) to energy equivalents in order to assemble direct and indirect energy requirements. In contrast to this aggregate approach, process analysis can focus on a particular product in specific circumstances as it identifies all direct energy inputs and as many relevant indirect needs as possible, a process that is in itself quite valuable as a management tool. As with all appraisals that deal with complex inputs and encompass sequential processes, the setting of analytical boundaries will affect the outcome of process analysis. In most cases, the truncation error inherent in counting only direct energy inputs (purchased fuels and electricity) will be small, but in some instances it could be surprisingly large. For example, Lenzen and Dey (2000) found the energy costs of Australian steel to be 19 GJ/t with process analysis, but 40.1 GJ/t with input–output analysis.

Another complication is introduced due to increasing shares of globally traded commodities and products: in some cases the additional energies required for import of raw materials and export of finished products will be a negligible share of the process energy cost, in other cases their omission will cause a serious undercount. For example, two identical looking steel beams used at two construction sites in New York may have two very different histories: the first being a domestic product made by the scrap-EAF-continuous (electric arc furnace) casting route in an integrated operation in Pennsylvania, the other coming from China where Australian iron ore and coke made from Indonesian coal were smelted in a blast furnace in one province and the beams were made from ingots in another one before loaded for a trans-Pacific shipment and then transported by railroads across the continent. Approximate energy costs of long-distance transportation can be easily calculated by assuming the following averages (all in ton-kilometers for easy comparability, ranked from the highest rates to the lowest): air transport 30 MJ, diesel-powered trucks (depending on their size) mostly between 1 and 2.5 MJ, diesel-powered trains 600–900 kJ, electricity-powered trains 200–400 kJ, smaller cargo ships 100–150 kJ, and large tankers and bulk cargo carriers just 50 kJ/tkm (Smil, 2010). Obviously, energy-intensive air shipments will be restricted to high value-added products, while bringing iron ore by a bulk carrier from a mine 3000 km from a Chinese blast furnace would entail energy expenditure equal to less than 10% of overall requirements for steel production – while the energy cost of shipping construction stone from Europe or Asia to the USA may be equal to 25–50% of the energy used to cut and polish it. These realities should be kept in mind when examining and comparing the values reviewed in this section. Energy costs – presented here in a uniform way as gigajoules per ton (GJ/t) of raw material or product –as with any analytical tool, they alone cannot be used to guide our choices and preferences of material use without concurrent considerations of affordability, quality, durability, or esthetic preference; if the latter were to be ignored concrete, a material of low energy intensity, would rule the modern world even more than it actually does.

The energy cost of market-ready lumber (timber) is low, comparable to the energy cost of many bulk mineral and basic construction materials produced by their processing. Tree felling, removal of boles from the forest, their squaring and air drying will add up to no more than about 500 MJ/t, and even with relatively energy-intensive kiln-drying (this operation may account for 80–90% of all thermal energy) the total could be as low as 1.5 and more than 3.5 GJ/t (including cutting and planing) for such common dimensional construction cuts as 2 × 4 studs used for framing North American houses.

The low energy cost of wood is also illustrated by the fact that, in Canada, the energy cost of wood products represents less than 5% of the cost of the goods sold (Meil et al., 2009). Energy costs on the order of 1–3 GJ/t are, of course, only small fractions of wood’s energy content that ranges from 15 to 17 GJ/t for air-dry material. Obviously, the energy cost of wood products rises with the degree of processing (FAO, 1990). Particle board (with a density between 0.66 and 0.70 g/cm3) may need as little as 3 GJ/t and no more than 7 GJ/t, with some 60% of all energy needed for particle drying and 20% for hot pressing.

The energy cost of papermaking varies with the final product and, given the size and production scale of modern papermaking machines (typically 150 m long, running speeds up to 1800 m/min., and annual output of 300 000 t of paper), is not amenable to drastic changes (Austin, 2010). Unbleached packaging paper made from thermo-mechanical pulp is the least energy-expensive kind (as little as 23 GJ/t); fine bleached uncoated paper made from kraft pulp consumes at least 27 GJ/t and commonly just over 30 GJ/t (Worrell et al., 2008). Most people find it surprising that this is as much as a high-quality steel.

Recycled and de-inked newsprint or tissue can be made with less than 18 GJ/t, but the material is often down-cycled into lower quality packaging materials.

Construction aggregates whose production requires only extraction and some physical treatment (sorting, sizing, crushing, milling, drying) have generally very low to low energy costs, and higher fuel and electricity use comes only with the pyro-processing required to make bricks, tiles, glass, and, above all, cement. The energy cost of natural stone products is low, usually just around 500 MJ/t for quarried blocks, somewhat less for crushed stone, but twice as much for roughly cut or split stones,

The energy costs of sand extraction and processing can easily vary by a factor of 2, but even the higher costs leave them in the category of the least energy-intensive materials when compared in mass terms. The simplest mining and preparation sequence to produce fairly clean sand and uniformly-sized sand may require no more than 100 MJ/t, and even more costly gravel sorting (or crushing as needed) should have an energy cost well below 500 MJ/t. The highest energy input is required for the preparation of the industrial sand that is used in glassmaking, ceramics and refractory materials, metal smelting and casting, paints, and now also increasingly in hydraulic fractioning of gas- and oil-bearing shales: its moisture must be reduced (in heavy-duty rotary or fluidized-bed dryers) to less than 0.5%, and this may consume close to 1 GJ/t. Bricks fired in inefficient rural furnaces in Asia may require as much as 2 GJ/t, just 1.1–1.2 GJ/t is typical for Chinese enterprises (Global Environmental Facility, 2012; Li, 2012) while US production of high-quality bricks requires 2.3 GJ/t (USEPA, 2003). Cement production is fairly energy-intensive because of high temperatures required for the thermo-chemical processing of the mineral charge. Limestone supplies Ca and other oxides, and clay, shale, or waste material provide silicon, aluminum, and iron; in order to produce a ton of cement about 1.8 t of raw minerals are ground and their mixture is heated to at least 1450 °C.

This sintering process combines the constituent molecules, and the resulting clinker is ground again with the addition of other materials to produce 1 t of clinker that is then ground to produce fine Portland cement. Fly ash (captured in coal-fired power plants) or blast furnace slag can be used to lower the amount of clinker. Additional energy is to needed to rotate large kilns. These inclined (3.5–4 °) metal cylinders are commonly around 100 m, up to 230 m, long, with diameters of 6–8 m, and they turn typically at 1–3 rpm, with the charged raw material moving down the tube against the rising hot gases (Peray, 1986; FLSmidth, 2011). Disaggregation of all energy inputs shows that the extraction of minerals (limestone, clay, shale) and their delivery to cement kilns is a minimal burden. Kiln feed preparation is electricity-intensive as crushing and grinding of the charge consumes about 25–35 kWh/t and the grinding and transportation of the finished product (clinker) claims at least 32–37 kWh/t (Worrell and Galitsky, 2008). This leaves the bulk of energy consumption for the pyro-processing, a sequence of water evaporation, decomposition of clays to yield SiO2, decomposition of limestone or dolomite (calcination) that releases CaCO3, formation of belite (Ca2SiO4, making up about 15% of clinker by mass), and finally sintering, production of alite (Ca3O·SiO4, that makes up some 65% of the clinker mass) (Winter, 2012). Total energy use in cement production varies with the principal fuel used, the origin of the electric supply, and the method of production. Average specific energy consumption in the cement industry has declined as a more efficient dry process replaced the old wet method. The highest electricity consumption in the dry process is for the grinding of raw materials and clinker and for the kiln and the cooler, in aggregate more than 80% of the total that averages mostly between 90 and 120 kWh/t of cement (Madlool et al., 2011). Heating of dry kilns (mostly with coal, petroleum coke, and waste materials in the USA, and with coal in China) consumes between mostly 3 and 4 GJ/t; the range is 3.0–3.5 GJ/t for kilns with four or five stages of preheating, while a six-stage process could work with as little as 2.9–3 GJ/t (IEA, 2007; Worrell and Galitsky, 2008).

World best practice can now produce Portland cement with total primary energy inputs of 3.3–3.5 GJ/t, while the rates for fly-ash cement and blast furnace slag cement can be as low as, respectively, 2.4 and 2.1 GJ/t (Worrell et al., 2008). In contrast, many plants in low-income countries still need around 4.5 GJ/t for Portland cement.

Energy requirements for glass production range mostly between 4 and 10 GJ/t, with about 7 GJ/t being a typical value

the energy cost of ceramic products rises with the degree of pyro-processing and the quality of items: unglazed tiles need only 6 GJ/t, glazed tiles up to 10 GJ/t, fine ceramics as much as 70 GJ/t,

The usual approach for quantifying the energy costs of the iron and steel industry is to include the energy costs of coke, pelletizing and sintering of ore, iron and steel making, cold and hot rolling, and galvanizing and coating; this leaves out the energy costs of coal and ore mining and transportation, of such energy-intensive inputs as electrodes and refractories, as well as the embodied energy cost of scrap metal. Analyses performed (more or less) within these boundaries show that average energy consumption in the global steel industry was about 20 GJ/t by the year 2000 (Yellishetty et al., 2010).

A review of best industry practices for the entire iron–steel sequence ended up with 16.3–18.2 GJ/t for the blast furnace-BOF-continuous (basic oxygen furnace) casting route, 18.6 GJ/t for direct iron reduction followed by EAF steelmaking and thin-slab casting, and 6 GJ/t for melting scrap metal in EAF and thin-slab casting (Worrell et al., 2008).

A comparative analysis of the energy costs of the iron and steel industry in the USA and China illustrates this reality: it shows that the aggregate input in 2006 was, respectively, 14.9 GJ/t and 23.11 GJ/t of crude steel (Hasanbeigi et al., 2012).

electricity’s share is 20% of the total primary energy used in the US steelmaking, but only 10% in China’s industry. Taking 25 GJ/t as a mean would suggest that in the year 2010 the global iron and steel industry needed roughly 36 EJ of energy, or about 6% of the worldwide consumption of primary commercial energy. For comparison, Allwood and Cullen (2012) put the global energy use in steelmaking at 38 EJ. Aluminum production is much more energy intensive than making steel. Fuel and electricity consumption in the Bayer process, between 10 and 13 GJ/t of alumina, is a small share of the overall cost that is dominated by the electrolysis that is done, preferably with the cheapest kind of electricity produced in large hydro stations (it supplies about 60% of the industry’s needs worldwide).

The IEA put the weighted energy cost of the entire sequence at 175 GJ/t in 2004, and a review of best industry practices came up with a nearly identical rate of 174 GJ/t of metal (Worrell et al., 2008). That is nearly twice the energy intensity of copper and almost 10 times as much as the least energy-intensive production of steel using the blast furnace-BOF-continuous casting route. Global 2010 production of 40.8 Mt of Al would have thus required about 7.1 EJ, less than 1.5% of the world’s total primary commercial energy supply. Metal’s high electricity requirements steer the location of primary aluminum production to countries with abundant hydro resources, and the four such largest producers (China, Russia, Canada, and the USA) account for half of the world output. Secondary aluminum requires only 7.6 GJ/t (for remelting only). As already noted, titanium has the highest energy cost among the other relatively commonly used metals (400 GJ/t), followed by nickel at about 160 GJ/t and copper (global average of 93 GJ/t), while chromium, manganese, tin, and zinc have a very similar energy cost of about 50 GJ/t (IEA, 2007). Not surprisingly, very low metal concentrations of even the best exploited deposits raise the energy intensities of silver and gold orders of magnitude above common metals: the average for silver is about 2.9 TJ/t (30 times higher than for copper) and for gold it is 53 TJ/t, roughly 300 times the energy cost of aluminum.

A World Bank review of PE energy costs found ranges of 87.4–107.8 GJ/t for high density polyethylene (HDPE) and 74.4–116.3 for low-density polyethylene (LDPE), with the processing energy being as low as 25–28 GJ/t and as high as 45 GJ/t (Vlachopoulos, 2009).

The heavy dependence of modern production of plastics on hydrocarbon feedstocks has not been, so far, a major burden, as the industry still claims less than 5% of the world’s natural gas and crude oil output. Rising demand for plastic materials and higher costs, particularly of crude oil, will change this, and in the long run plant-based bioplastics appear to be the only practical answer both to the eventually less abundant petrochemical feedstocks and to the presence of nonbiodegradable materials in the environment.

 

the lignin carbon fiber used in reducing the weight of passenger cars and other vehicles costs 670 GJ/t and carbon-fiber reinforced polymer (polyacrylonitrite) fiber requires just over 700 GJ/t (Das, 2011), more than three times that of aluminum.

synthesis of ammonia from its elements, the Haber–Bosch process

as little as 27 GJ/t NH3 in the year 2000. When Worrell et al. (2008) reviewed the best commercial practices, they rated natural gas-based synthesis at 28 GJ/t (roughly a third higher than the stoichiometric minimum), and coal-based process at 34.8 GJ/t. Naturally, typical performances are higher, around 30 GJ/t NH3 for gas-based plants, 36 GJ/t for heavy fuel oil feedstock, and more than 45 GJ/t NH3 for coal-based synthesis (Rafiqul et al., 2005). The IEA (2007) used regional means ranging from 48.4 GJ/t in China to 35 GJ/t in Western Europe, resulting in a global weighted mean of 41.6 GJ/t for the year 2005.

treating the insoluble rocks with sulfuric and nitric acids in order to produce water-soluble phosphorus compounds is much more energy intensive. Overall energy costs range from 18 to 20 GJ/t for superphosphates (single superphosphate with just 8.8% P, triple superphosphate with 20% P) to 28–33 GJ/t for diammonium phosphate containing 20% of soluble P (Smil, 2008). The energy cost of potash (sylvinite) extraction is low: in Saskatchewan, conventional underground mining followed by milling needs only 1–1.5 GJ/t, and surface mining and milling averages only about 300 MJ/t (NRC, 2009).

the entire production chain – starting with Si made from quartz and carbon through trichlorosilane, polysilicon, single crystal ingot, Si wafers, and actual fabrication and assembly of a microchip – consumes about 41 MJ for a 2-g chip. This implies a total electricity cost for shipped wafers of at least 2100 kWh/kg. Even if using only hydroelectricity this would prorate to about 7.6 GJ/kg, and energizing the entire process by electricity generated from fossil fuels would push the total primary energy to more than 20 GJ/kg for finished Si wafers, 2 orders of magnitude more than aluminum made from bauxite, and 3 orders of magnitude more than steel made from iron ore.

The typical rates presented in this section can be used (after rounding, to avoid impressions of unwarranted accuracy) to assess the global energy needs of major material sectors and to calculate their fractions of TPES (whose total was just over 500 EJ) in 2010 (BP, 2013). Not surprisingly, steel’s relatively high energy intensity (25 GJ/t) and its massive output (1.43 Gt in 2010, 1.5 Gt in 2011) make it the material with the highest total energy demand that dominates the total of about 50 EJ (or 10% of TPES) required to produce all metals in 2010. Plastics are next (assuming 80 GJ/t and an output of 265 Mt in 2010) with roughly 20 EJ (4% of TPES), well ahead of construction materials (cement, bricks, glass) with about 15 EJ, or 3% of TPES. Paper production required about 10 EJ and fertilizers added less than 8 EJ for the grand total of just over 100 EJ or 20% of the world’s TPES in 2010. For comparison, the IEA (2007) estimated energy input for the entire global industrial sector at almost 88 EJ for the year 2005. Paper (and paperboard) and aluminum each end up with very similar totals of close to 10 EJ (2% of TPES) as a result of aluminum’s much higher energy intensity (175 vs. 25 GJ/t) but much lower total output (53 vs. 400 Mt in 2010). Perhaps the most interesting result concerns the energy cost of inorganic fertilizers: given their truly existential importance it is reassuring to realize that the energy needed to produce them adds up to a surprisingly small share of global supply. Assuming averages of 55, 20, and 10 GJ/t for, respectively, N, P, and K (all including the cost of final formulation, packaging, and distribution) would result in a total demand of a bit more than 5 EJ in the year 2010 (with nitrogenous fertilizers accounting for about 90% of the total) – or only about 1% of the TPES.

 

the steadily increasing crowding of transistors has limited the annual mass of wafers needed to produce all of the world’s microchips to only about 7500 t in 2009 and to an aggregate energy expenditure of just 150 PJ, or about 0.03% of TPES.

These calculations also make it clear that modern civilization can afford all this steel and fertilizers and microchips because scientific discoveries and technical advances have greatly reduced their energy intensities.

LCA is now a mature analytical discipline that has its own periodical, International Journal of Life Cycle Assessment,

Varieties of LCA include complete cradle-to-grave sequences

LCAs of housing in cold climates show embodied energies as a small fraction of life-time total.

A 200 square meter Canadian house that cost 1.5 TJ to build, heating and lighting (averaging about 25 W/m2) will claim about 9.5 TJ in 60 years, reducing the construction share to just 14% of the overall cost. By coincidence, that share is nearly identical to the construction share of a medium-sized American car: it takes about 100 GJ to produce and (at about 8 l/100 km and 20 000 km/year) it will need about 550 GJ of fuel and oil in 10 years, while the initial construction will claim only about 15% of the overall cost, and even less once repairs and garaging are included (Smil, 2008). Embodied energies make up even lower shares in the life-cycles of machines that are in nearly constant operation: only 6–7% for jetliners, freight trains, and cargo ships (Allwood and Cullen, 2012).

Williams (2004) ended up with the reverse ratio. According to his analysis, the energy used to make a desktop with a Pentium III processor, 30 GB hard drive, and 42.5 cm monitor added up to 6.4 GJ, while during its relative life of three years the desktop would consume about 420 kWh of electricity or roughly 1.5 GJ of primary energy, yielding a manufacturing:usage energy split of 81:19.

For a Swiss desktop computer, the split was much closer at 46:54 (Ecoinvent, 2013).

LCAs also make it clear that over their life-time many infrastructures will cost nearly as much, or more, to maintain than their initial construction. A Canadian LCA for a high-volume two-lane concrete highway shows initial construction costs of 6.7 TJ/km and a rehabilitation cost of 4.1 TJ/km or 38% of the total 50-year cost of 10.8 TJ/km, and the burdens are actually reversed for a roadway made of flexible asphalt concrete that needs 15 TJ/km to build and 16% more (17.4 TJ/km) to rehabilitate over a 50-year life-cycle (Cement Association of Canada, 2006).

The LCA of repeatedly washed garments is yet another excellent illustration of the boundary problem noted at the beginning of the energy costs section, as products made from different materials have different durabilities and maintenance requirements and a complete account of these realities may shift the overall advantage from a material that requires less energy to produce to one that is more energy-intensive to make but whose life-long energy cost may be lower.

polyester production required twice as much energy as producing cotton lint (clear advantage cotton) but because of a higher cost of cloth manufacturing, the total energy cost of a cotton shirt was about 20% higher than that of its pure polyester counterpart (slight advantage polyester), and after including the energy costs of maintenance (washing, drying, ironing) the cotton shirt was about 3.6 time more energy intensive (clear advantage polyester).

Cotton appears even more disadvantaged once nonenergy impacts are compared: the water requirements of the blend are less than a third of those for pure cotton, and global warming and acidification potentials are 38% lower than those for producing and laundering two pure cotton sheets. Going even further, we can consider the long-term cost of excessive soil erosion in cotton fields, soil quality decline due to salinization in irrigated cotton fields in arid regions, and the presence of pesticide residues in soil and water (Smil, 2008). All of these are avoided by using a synthetic fiber – but its production depends on a nonrenewable feedstock. But so does the cultivation of cotton: PE synthesis consumes about 1.5 kg of hydrocarbons per kilogram of fiber but growing a kilogram of cotton requires nearly 500 g of fertilizers and 15 g of pesticides made from hydrocarbon feedstocks, as well as liquid fossil fuels for farm machinery.

But what LCAs have done is to allow comparisons within the same category of environmental consequences (when a choice of materials is possible, which one will have the lowest effect on water use or water pollution?) as well as more comprehensive rankings of materials according to several categories of environmental impact. This is important because material use is one of the three dominant ways that humans have been changing the biosphere: food production and energy supply (dominated by extraction and combustion of fossil fuels) are the other two great interventions. And when considered in its entirety, the intricate system of extraction, processing, transportation, use, reuse, and disposal of materials encompasses every major environmental interference, from land use changes (ranging from deforestation due to lumber and pulp production to destruction of plant cover and disruption of water cycle due to massive surface ore mines) to atmospheric emissions (ranging from acidifying gases to being a major contributor to anthropogenic warming).

 

there can be no ranking of anthropogenic environmental impacts. There is no unifying metric that would allow us to conclude that soil erosion should be a greater concern than photochemical smog, or that tropical deforestation is more worrisome than the enormous water demand of modern irrigated agriculture.

With CO2 usually being the leading contributor, this indicator will have a very high correlation with the mass of fossil fuel used in the initial production. For example, production of a kilogram of cement will release about 0.5 kg CO2 per kg of but making a kilogram of hot-rolled sheet steel will release about 2.25 kg of CO 2 liberated from fossil fuels (NREL, 2013b). The other two commonly assessed variables are acidification and eutrophication impacts. Emissions of sulfur and nitrogen oxides from combustion of fossil fuels, smelting of ores, and other industrial processes are the precursors of atmospheric sulfates and nitrates whose wet and dry deposition acidifies waters and soils. In LCAs this acidifying potential of products or processes is calculated in terms of grams of hydrogen ions per square meter (g H +/m2) or per volume of water (g H +/l). Eutrophication is the process of nutrient enrichment of fresh and coastal waters (most often by releases and leaching of nitrates and phosphates) that leads to excessive growth of algae whose decay deprives waters of dissolved oxygen and creates anoxic zones that either kill or impoverish heterotrophic life in affected waters

As just explained in the previous section, production of materials claims roughly a quarter of the world’s primary energy supply. Because the combustion of fossil fuels provides most of this energy – in global terms about 87% in 2010, with the rest coming from primary, that is mostly hydro and nuclear, electricity (BP, 2013) – production of materials is a leading source of emissions of particulate matter (including black carbon), SOx and NOx (whose conversion to sulfates and nitrates is the primary cause of acidifying precipitation), and GHGs. The water needed for processing, reaction, and cooling ends up contaminated or is released at elevated temperatures by many industries, and most of them also share the necessity of disposing of relatively large volumes of solid waste and small but potentially worrisome volumes of hazardous waste.

 

Production of a kilogram of typical construction steel sections generates about 1.5 kg of CO2 equivalent, 50 g of SO2 equivalent of acidification potential, a negligible amount of eutrophication potential (0.36 g), and just 0.8 g of the photochemical smog-inducing ethane (WSA, 2011b). Production of a kilogram of PVC (consuming around 60 MJ) requires about 10 kg of water (excluding cooling demand), produces 1.9–2.5 kg of CO2 equivalent, 5–7 g of acidification potential (as SO2), 0.6–0.9 g of nitrification potential (as PO4), nearly 0.5 of ethane (measure of photochemical ozone creation), 0.4–0.8 g of total particulate waste, and 5–8 g of hazardous waste (Sevenster, 2008).

Natural materials can be a better choice, and often resoundingly so in the case of biomaterials. For example, a comprehensive LCA by Bolin and Smith (2011) showed that a wood/plastic composite decking results in 14 times higher fossil fuel use, 3 times more GHG, almost 3 times more water use, 4 times higher acidification potential, and about 2 times more smog potential and ecotoxicity than a wooden deck built with lumber treated with alkaline copper.

Wooden floors are much less energy intensive than the common alternatives: the total energy per square meter of flooring per year of service was put at 1.6 MJ for wood (usually oak or maple) compared to 2.3 MJ for linoleum and 2.8 MJ for vinyl.

Remarkably, many life-cycle assessments take a cavalier approach to life-spans and just assume what seems to be a reasonable length or use (often unexplained)

There are no LCAs taking into account the great longevity of some plastics (on the order of hundreds of years) and their now ubiquitous, and clearly highly disruptive, distribution in aquatic environments. Their buoyancy, their breakdown into progressively smaller particles, and their eventual sinking through the water column to the sea bottom combine to make them a truly global and omnipresent environmental risk to marine biota: they are now found on the remotest islands as well as in the abyss, but their highest concentrations are in surface water and on beaches (Moore, 2008; Barnes et al., 2009).

Great Pacific Garbage Patch (Moore and Phillips, 2011). Later studies estimated that at least 6.4 Mt of plastic litter enters the oceans every year; that some 8 million pieces are discarded every day; that the floating plastic debris averages more than 13 000 pieces per km2 of ocean surface; and that some 60% of all marine litter stems from shoreline activities (UNEP, 2009). Additionally, the latest summaries show that, despite many efforts to limit this, now planet-wide, degradation the accumulation is still increasing (STAP, 2011). Dangers to ocean life are posed by every size unit of discarded plastics: among the largest items are abandoned or damaged fishing nets that can ensnare fishes, dolphins, and often even whales; while aquatic birds often mistake small-size pieces of plastic for small fish or invertebrates and regurgitate them to their fledglings: the stomachs of many species show a distressing collection of such objects. And microplastics – the smallest pieces (sizes less than 5, 2, or 1 mm) that are manufactured for cosmetics, drugs, and industrial uses and that arise from abrasion and photodegradation of larger pieces – can be ingested by marine biota and can have serious metabolic and toxic effects (Cole et al., 2011). Inevitably, masses of plastic microparticles have been also accumulating on shorelines where they endanger more organisms (Browne et al., 2011). This important example shows how incomplete and uncertain are even our best analytical procedures tracing the requirements and consequences of material production, use, and abandonment; it also provides a strong argument for much better management of materials and, obviously, recycling should be a key component of these efforts.

 

with most materials recycling is more accurately described as down-cycling: high-quality paper becomes packaging stock or cardboard, expensive plastics are turned into cheap items.

Production of recycled steel will require roughly 75% less energy and the LCA of common steel products (sections, hot-rolled coil, and hot-dip galvanized steel) showed recycling benefits of up to about 50% for both GWP and acidification potential (WSA, 2011b).

Delacquering (removal of any coatings) needs about 7 GJ/t, melting consumes 7 GJ/t, addition of pure aluminum to adjust the alloy composition takes about 8 GJ/t, and the production of containers (casting, rolling, blanketing, forming) adds 30 GJ/t for the total of about 52 GJ/t, a saving of 74% rather than 96% (Luo and Soria, 2008).

Benefits go beyond energy savings, as recycling lowers GWP, claims less water, and generates less water and air pollution: Grimes et al. (2008) provide many comparisons for steel, aluminum, and paper production. Recycled paper can be made with 40% less energy while generating 45% less waste water and 50% less solid waste (EPN, 2007). Energy savings for recycled high-density polyethylene are 45–50% and 40–45% for PVC.

recycling is also a quest that is often very difficult to pursue (be it because of logistic challenges, excessive costs, or negligible energy savings) and one that, unlike the flows of carbon or nitrogen atoms in grand biogeochemical cycles, often amounts to a fairly rapid down-cycling as the reused materials appear in less valuable guises.

The greatest challenge is to recycle the increasing amount of electronic waste that is, in mass terms, dominated by a few plastics, a few metals, and screen glass, but that also contains small, but in aggregate quite substantial, amounts of more than a dozen elements.

Of the 60 metals and metalloids examined by a UN report, more than half (including all rare earths, as well as germanium, selenium, indium, and tellurium) have recovery rates of less than 1%, for only five elements is the rate between 1 and 25%, and 18 common (or relatively common) metals have end-of-life recycling rates above 50%, but rarely above 60% (Graedel et al., 2011).

Unfortunately, some toxic heavy metals, whose release into the environment is particularly undesirable, have very low recycling rates. The only relatively common way of recycling cadmium is by returned Ni-Cd batteries, but their collection rate remains low. The recycling rate of fluorescent lights containing mercury is also too low.

Correct separation is difficult but imperative: a single PVC bottle in a load of 10 000 PET bottles can ruin the entire melt (ImpEE Project, 2013). Sorting is followed by cleaning (but removing print and labels cannot be 100% successful) and reduction to uniform pellet sizes ready for reuse, still mostly only in lower-grade applications such as cheap carpets, garbage cans, or park benches.

Collection of household waste paper is expensive, and a thorough processing of the material is needed to produce clean fibers for reuse. This includes defibering of paper, cleaning and removal of all nonfiber ingredients (most often adhesive tapes, plastics, and staples), and de-inking is needed if the fibers are to be reprocessed into white paper. Reprocessing shortens the cellulose fibers and this means that paper can be recycled no more than 4 to 7 times.

Recycling of e-waste is particularly challenging, because the devices commingle many compounds and elements that must be separated by a sequence of mechanical and chemical operations. Silicon is the core of modern computers, but transistors and microchips could not function without the presence of a multitude of materials, including many heavy metals.  Elements used to dope silicon include arsenic, phosphorus, boron, and gallium. Printed wire boards, disk drives, expansion cards, electrical supplies, and connections add up to a hoard of materials, small (and for many elements even miniature) per unit but highly consequention in the aggregate. Most of the mass in electronic devices is steel, glass, plastics, copper, and aluminum, but more than a dozen other metals are also present in tiny amounts, eight of them classified as hazardous (As, Cd, Cr, Co, Hg, Pb, Sb, and Se).

Lead is of the greatest concern. The lead in computer wire boards has 30 to 100 times the level of lead (5 mg/l) that classifies waste as hazardous.

In 1997 100 million cellphones were sold a year, a billion by 2009, and at the end of 2012 more than 6.5 billion devices in use, plus tablets, notebooks, and netbooks. These only last 1.5 to 2 years, so up to 1.2 billion cellphones a year are discarded.  In the USA recycling rates for all mobile devices was a dismal 8% in 2009 (and 17% for TVs, 38% for computers).  130 million discarded devices in the USA collectively have 2000 tons copper, 45 tons silver, 4 tons gold, 9,000 tons plastics, 2,500 tons ceramics and glass. Multiply by 6 or more for global amounts.

 

The USA buries nearly 80,000 tons of plastic in its landfills every day, with only 8% of discarded plastics recovered (23% for PET bottles but less than 1% for polypropylene waste).

ENERGY EFFICIENCIES

Late 19th to early 20th century hand-stoked coal stoves converted no more than 20-25% or less of the fuel’s chemical energy to useful heat, though that’s good compared to the less than 10% efficiency of wood-burning fireplaces before that. Oil-fired furnace efficiency can be up to 50%, natural gas home furnaces 70-75%.

Steam engines with top efficiency of 15% used to power small freight ships (less than 5,000 dwt). No bulk carriers (100,000 dwt+) are powered by diesel engines whose best efficiencies are close to or slightly above 50%.

The world’s copper resources are about 1.6 Gt.  To provide all 10 billion people in the future with the average per capita mean (170 kg) would require 1.7 Gt, more than the estimated resource in the crust (Gordon 2006).

Gordon, R.B., et al. 2006. Metals stocks and sustainability. Proceedings of the National Academy of Sciences (USA) 103:12-9-1214

Please follow and like us:
This entry was posted in Infrastructure, Life Before Oil, Limits To Growth, Peak Resources, Vaclav Smil and tagged , , . Bookmark the permalink.

10 Responses to Vaclav Smil. Making the modern world: materials and dematerialization

  1. hugh owens says:

    Alice I can’t believe you read the whole book either. I am working my way thru your reader’s digest version and it doesn’t seem to be any better than Chris Clugston’s book some years back which details it all I think better. And he does lay out implications. In case you haven’t seen it:https://www.amazon.com/gp/product/1621412504/ref=dbs_a_def_rwt_hsch_vapi_taft_p1_i0. In case this doeasn’t link, it’s call Scarcity: Humanitys final chapter. kind regards.

  2. John A. Joslin says:

    Holy Crap. Ulp. Uh… well, at least I finally got Elon Musk figured out. He’s a small-time , 3-D ‘materials’ cartoonist.
    Chirpily turns out shiny little artifacts for us ammonia nourished rubes to amuse ourselves w/ until closing time or the concrete rubble comes tumbling down…
    Whichever comes first!

  3. Jeff W. says:

    @Alice: thanks for trying to digest another bone dry Smil tome. I find it odd that he will crank out these huge volumes of alarming factoids and draw few if any conclusions from it all. He’s like some modern age scribe jotting down the myriad ways humans are driving full-speed into civilizational collapse.

    @Hugh: thanks for the book reference. Looks like none of the libraries here have it either..

  4. NJF says:

    Another issue that comes up with trying to combat overpopulation is that there isn’t some sort of world government that holds any power.

    China may be able to enact a one-child policy, and has, but it cannot control the actions of Brazil, Mali, or Indonesia.

    Thus they can optimize the population in China, but their population reduction is not going to mean anything if Africa tacks on 2 billion more. And the US just keeps letting in more people despite already being likely 100 million “over budget” in terms of population. And the current trend is, the less soluble the ethnic group, the more “moral” it is to resettle them here.

    Unfortunately, resource scarcity inevitably intensifies tribalism, so expect more violence if not outright warring factions post-collapse.

    • Ken Barrows says:

      China gave up that one child policy.

      • NJF says:

        I am pretty sure it is still 2 child limit. Still no growth.

        But my comment is a thought experiment. The current status of the one-child policy in China has no bearing on the point being made.

        • Ken Barrows says:

          My only point is that we need a one child policy (or stricter ) everywhere and we have it nowhere

  5. A 200 square meter Canadian house that cost 1.5 TJ to build, heating and lighting (averaging about 25 W/m2) will claim about 9.5 TJ in 60 years, reducing the construction share to just 14% of the overall cost. By coincidence, that share is nearly identical to the construction share of a medium-sized American car: it takes about 100 GJ to produce and (at about 8 l/100 km and 20 000 km/year) it will need about 550 GJ of fuel and oil in 10 years, while the initial construction will claim only about 15% of the overall cost, and even less once repairs and garaging are included (Smil, 2008). Embodied energies make up even lower shares in the life-cycles of machines that are in nearly constant operation: only 6–7% for jetliners, freight trains, and cargo ships (Allwood and Cullen, 2012).

    This partiality has played havoc with humans’ understanding of Energy, since 1800’s Carnot and Jevons;

    Smil doesn’t recognise here, too, the energy-cost of mining-for, manufacturing and sustaining the industrial lines and processes that have created the Car.

    He inspires that those systems are found existing by default, from ‘nothing’, and they will run forever, unattended and without any repairs or replacements – as long as you put fuels into them.

    Wrong!

    Engineers know how energy-intensive keeping the best of our CNC machines calibrated, for instance, and their worn-out parts replaced, or they become one of the most destructive tools man has ever created!

    Shouldn’t Smil have added the energy-cost of all those subsystems only, but adding the energy-cost of extracting the energy-supplies that have fueled the process that brought the lines into existence, too.

    That is, for example, all the energy expended since 1930’s in creating the oil extraction industrial base in the US that has been then shipped to and installed later in Saudi Arabia to produce and export Saudi Arabia national oil (very little industrial base was in Saudi then, if at all. The country remains to date poorly industrialised).

    Then, the Saudi oil, along with (maybe) some Australian natural gas, after being collected and shipped in tankers, have created an industrial line and an infrastructure in Germany to create the ‘car’.

    Added to it, all the energy cost of sustaining humans in the process, including running the industrial lines in the US and Netherlands that have manufactured combined harvesters to provide food, ships that have transported the food and fertilizers, supermarkets for selling the food, and cookers cooking the food – for all other people involved!

    The car will never last long enough in one piece to pay its share of all that very long trail of total energy before wear and tear takes it down into the junkyard – not to mention the energy expended in its maintenance and repairs.

    Alice best shines when font-bolding such statements and touches in her works on Externalities and the fact that in each and every piece of an energy-producing device we run today, we must see traces of energy supplies that were burned in 1600’s Britain, and earlier, a piece of road that’s still contributing into the making of the car, a piece of rail track – if we like to entirely dismiss the fact that the steam engine remains the master design of all ICEs!

    Without forests in 1600’s Britain, the steam engine wouldn’t have likely became ready, just-in-time, later in the US for the oil extraction in Pennsylvania.

    Without the oil and trains in Pennsylvania, the Saudi Oil wouldn’t be possible to produce in the 1940’s.

    And without the Saudi Oil, starting the extraction of Britain’s North Sea oil wasn’t possible in the 1990’s.

    Not Smil or anybody else can claim they can cut-clear where the total energy that made the Car has started from!

    Energy is continuous, always and only comes flowing from the past into the future (The Fifth Law).

    When a nuclear plant today generates power, it draws from that long line of energy-burning legacy, giving an impression it generates excess energy (see the number of cars parking at any nuclear power plant site. Take them away and the plant will stop when operators fail to arrive to run it. Calculate the grand total energy behind what brought the cars into existence and you’ll find out the plant is an energy sink!).

    To that extent, humans became so confused by the massive amount of fossil fuels supplies that we have sacrificed since the early steam engine, they think a fusion power plant now will do even better than a nuclear plant, or solar panels or wind turbines will do better generating more excess energy than the total energy put into constructing them!

    Start building a nuclear power plant without the existing industrial base by growing forests and charcoaling trees – and the process will take 500+ years to finish, if ever.

    In the process, you’ll create first a charcoal steam engine, and likely stay there, unless you enslave many nations, depopulating them, in the process.

    Run the nuclear power plant and it will not last for a fraction of the centuries-long time taken to build it, before wear and tear causes it to fail beyond repair.

    “Time taken in stocking energy to build an energy-producing device added to the time taken to build the device will always be longer than the entire useful working lifetime of the device”.

    Space colonisation remains challenged, at core, and that’s probably why we have not seen Aliens – due to what we understand now that Energy is not a playground for humans (or Aliens) on earth and beyond, but likely a prison!

  6. Fascinating information here, the book review and ensuing comments. I’m thinking of my recent visit to Hovenweep Nat’l Monument, the remains of stone age pueblos near Mesa Verde. Starting with what they had, and a complete “world knowledge encyclopedia” you’d have such a long journey to the axe, much less the telephone, fighter jet and internet. It’s been wonderful living at the acme of civilization. I never though about any of this back in the 60s. Just watched the Ed Sullivan show and decided what to wear to school. Thanks everyone.