by Vaclav Smil
There has been, of course, no absolute dematerialization as far as the use of steel in the global economy is concerned: in 2013 crude steel output was 2.3 times larger than in 1973. Similarly, it is obvious that all populous, rapidly modernizing economies (above all China, India, and Brazil) have seen nothing but substantial absolute long-term increases in steel consumption. Finding out if there has been any notable absolute steel dematerialization in major mature, affluent economies requires several adjustments. In 2013, steel production in those countries was, without exception, lower than it was two generations ago: 64% lower in the United Kingdom, 38% lower in France, 36% lower in the United States, 23% lower in Germany, and 10% lower in Japan. During those 40 years apparent crude steel consumption, taking into account all exports and imports of steel mill products, has shown absolute declines in all of these countries: 61% in the UK, 42% in France, 29% in the United States, 15% in Germany and 19% in Japan.
But given the extent of international trade in products whose mass is dominated by steel (vehicles, ships, mechanical and electrical machinery), it is necessary to adjust the apparent crude steel consumption by the net value of such indirect exports and imports. As expected, in 2013 Japan and Germany were large absolute net exporters of steel embedded in the products (mainly cars and heavy machinery) they sold abroad: the Japanese net value was 20 Mt and the German total was 8.9 Mt, while the United States indirectly imported an additional 15.3 Mt of steel, and the analogical 2013 totals were 3.2 Mt for France and 3.7 Mt for the United Kingdom (WSA, 2014). When the adjustment for indirect trade is done for these countries for the year 1973, 40-year declines in absolute steel consumption within national boundaries look like this: 17% in the United States, 26% in Germany and France, about 30% in Japan, and 40% in the United Kingdom.
The conclusion is thus clear: both apparent domestic consumption totals of crude steel and values adjusted for indirectly traded steel-intensive products show substantial absolute declines. Absolute reductions in steel consumption in affluent, mature economies have resulted from the combination of the following factors: better, less massive, product redesigns; substitution of the metal with lighter or cheaper alternatives (mainly by aluminum and plastics); shifts of economic activity from steel-intensive resource extraction (especially obvious in the United Kingdom, with abandoned coal mining) and metal-based manufacturing (compare the dominance of the US carmaking or Japanese shipbuilding during the 1970s to the recent status of those diminished industries) to services; reduced need for building construction in economies with very slowly growing (Germany, UK) or even declining (Japan’s case) populations; and lower material requirements needed just to maintain rather than to expand basic infrastructures.
New Processes
Forecasting the specifics of technical innovations is a notoriously counterproductive exercise, but the exceedingly low rate of success has not had any discernible effects on the frequency of this futile effort. Hence the next example is not to point out a particularly wrong forecast: I am using it merely as an apposite specific illustration of a common problem. In 1988, Iron Age examined the prospects of iron- and steelmaking, and the article quoted Egil Aukrust, the technical director of LTV Steel, who saw an early end of the era of oxygen steelmaking as a new process in which BOF and EAF get together for an in-bath smelting reduction (McManus, 1988b). And the same article cited the AISI estimate that it would take 10 years before Klöckner-CRA Technologies’ direct smelting process would become commercial. Neither of these have become a reality more than a quarter century later.
Today BOFs are more important than ever—in 2013 they produced 71% of all steel compared to 57% in 1988, and their total output was roughly 2.6 times higher, 1.14 Gt compared to 442 Mt (WSA, 1990, 2015)—and in 2015 there is no credible sign (not even a plausible inkling) of their early demise. And, not surprisingly, the expected direct smelting process never became commercial, while DRI processes that have been producing iron since 1988 have continued to diffuse, but their penetration has been, as already noted, much slower than initially expected: global DRI output rose from 14 Mt in 1988 (1.8% of the total) to 75.2 Mt in 2013, still only 4.7% of the total (WSA, 1990, 2015).
None of this is surprising: developing new, commercially acceptable ironmaking and steelmaking processes is a great challenge and the progress has been commensurably slow. Perhaps the best illustration of this slow progress has been the history of the HIsmelt (High Intensity) process, promoted by its proprietor as “the world’s first commercial direct smelting process for making iron straight from the ore” (Rio Tinto, 2014). Its history goes back to the development of bottom-blown oxygen converters and combined steelmaking by Klöckner Werke in the early 1970s, and its key feature is the injection of coal and fine ore into the molten bath through water-cooled lances. The ensuing reduction produces liquid iron and CO; oxygen-enriched hot blast comes from a top lance and burns the generated gas.
Advantages of the HIsmelt process include direct injections of iron ore fines (hence no sinter, no pellets) and crushed coal (cokemaking is eliminated) into the smelt reduction vessel and the use of a wide variety of possible feed materials, including hematite ore fines (even those high in P), magnetite concentrate, titano-magnetite ores, noncoking coals, and steel mill wastes. Elimination of sintering, coking, and hot blast stoves should reduce capital costs as well as space requirements and simplify operations. Trials of the process began in 1981, Klöckner tested a small pilot plant between 1984 and 1990, and a demonstration project at Kwinana (Western Australia) operated with a horizontal vessel between 1993 and 1996. A new vertical smelt reduction vessel was tested between 1997 and 1999, and its success led to an international joint venture (Rio Tinto, Nucor, Mitsubishi, Shougang) to build, at Kwinana, a HIsmelt plant with an annual capacity of 800,000 t.
At that time, Goldsworthy and Gull (2002) thought that successful operation of that plant would lead to a scale-up and that a larger HIsmelt unit, or a combination of units, would be able to replace a BF. The new plant operated between 2005 and 2008 before it was closed amid the economic downturn (Rio Tinto, 2014). According to a 2011 agreement, the plant was to be dismantled and moved to India (Jindal Steel & Power in Orissa), but the deal was canceled and the plant was bought by Molong company in China and intended for operation by 2014, with a larger pilot plant to be ready for operation by 2016 (Steel Times International, 2013). In Europe, a HIsmelt reactor has been combined with a cyclone pre-reducer in a demonstration plant at Tata Steel Ijmuiden (Netherlands), and the process, known as HIsarna, is a part of the European ULCOS project seeking methods for low CO2 emission steelmaking (Birat, 2010).
Reviews of emerging ironmaking processes are available in Manning and Fruehan (2001), Fruehan (2005), Harada and Tanaka (2011), and Fischedick et al. (2014)—but all of them are conspicuous for their lack of truly new alternatives as they mostly review the accomplishments and potential of existing DRI techniques. In its Technology Roadmap Research Program, the American Iron and Steel Institute defined an ideal ironmaking process as one that eliminates the need for coal and coke ovens (and hence reduces the emissions of CO2), that is able to use low-quality iron ores, that requires lower capital investment than the combination of coking oven and BF, and that is able to produce 5000–10,000 tonnes of hot metal a day in order to support the capacity of existing steel mills (AISI, 2010). The AISI roadmap considered six potential ironmaking alternatives and concluded that three of them could greatly reduce CO2 emissions: suspension reduction of iron ore concentrates, molten oxide electrolysis, and paired straight hearth furnace.
The first option, whose bench-scale testing has been done at the University of Utah, is now known as Novel Flash Ironmaking (AISI, 2014). This process would use fine iron oxide concentrates directly sprayed into the furnace chamber to be reduced by gaseous agents (natural gas, now so abundantly available in the United States from hydraulic fracturing of shales, syngas, hydrogen, or a combination of these gases): no pelletized or sintered products and
no coke would be needed. The process would require nearly 40% less energy, and CO2 emissions would be reduced by 96%, about 6%, and 30% with, respectively, hydrogen, natural gas, or coal when compared to BF ironmaking and could eventually replace that traditional route.
Molten oxide electrolysis produces molten iron and oxygen as electricity passes between two electrodes immersed in a molten salt that contains dissolved iron oxide (AISI, 2010). This process, obviously, replicates the well-known and massively commercial electrolytical aluminum smelting, and its eventual (as yet poorly defined) costs would be considered along with its much-reduced CO2 emissions. The third American innovation under investigation is the paired straight hearth furnace (PSH). The furnace is charged with cold self-reducing pellets (mixture of iron oxide and coal) whose reduction produces 95% metallized pellets that can be used in EAFs. Unlike conventional rotary hearth furnaces, whose bed height is just two to three pellets, the PSH has a bed of eight pellets (or 12 cm) deep to minimize reoxidation and to allow more efficient combustion. Eventually this furnace could be coupled with an oxy-coal melter to produce hot metal for steelmaking, a combination that would cut energy use by a third and CO2 emissions by two-thirds compared to the standard BF route (AISI, 2014).
A techno-economic evaluation of innovative steelmaking techniques concluded that the most likely scenario is that the standard BF–BOF route, as well as BF smelting combined with carbon capture and sequestration, will become unprofitable by the middle of the twenty-first century, and that a high share of renewable energy sources and high cost of carbon will make hydrogen direct reduction and electrolysis economically attractive (Fischedick et al., 2014). This conclusion is consistent with the scenarios constructed by the authors, but the assumptions they used to build them are arguable (most notably, and not surprisingly, the German team of authors assumed a mass penetration of inexpensive wind and solar electricity).
In any case, even if successfully demonstrated, all of the new ironmaking techniques would still have to make those critical shifts from demonstration to pilot plants to mass-scale production, going against a formidable target. The combination of the thermal and chemical efficiency of modern BFs and their large working volumes, high productivity, and remarkable longevity make it very difficult to come up with a mass-reduction technique of similar performance.
Future Requirements
Perhaps the best way of assessing the future of steel production is to separate what we know for certain from what we know fairly well in general terms but are unable to assess or quantify satisfactorily, and then to stress the areas of major uncertainties. We can say with confidence that any rationally conceivable increase of steel production during the first half of this century (and almost certainly also in its second half) will not be limited by the availability of primary resources. The Earth’s crust contains plenty of the iron ore, fluxing materials (limestone, dolomite), and coal needed to produce coke and to be used directly as pulverized fuel for injecting into BFs.
Obviously, every expansion of resource extraction has to end and what follows depends on the endeavor’s scale. Locally, a high annual output from a particulate mine (if not its actual peak output) may be followed by a rapid decline and a complete abandonment of the site, and a similar sequence can affect even an entire mining region or a specific resource extraction that was going on in a number of sites in a small country. Resource extraction has seen many of these production peaks or brief plateaux followed by precipitous retreats and creation of economically depressed towns or regions. Among the greatest reversals are the demise of British coal mining—from 130 Mt in 1980 to the closure of the last two mines in 2015—and the emergence of the United Kingdom as a major coal importer: “bringing coals to Newcastle” is now a mundane reality, with 50 Mt imported in 2014!
This development was not caused by running out of a resource but rather by the rising costs of local and regional production, by resource substitutions (North Sea oil and gas displacing British coal), and by imports of easily available and cheaper foreign supplies. For major, globally shared resources, and for the most commonly used materials there are three key questions to ask. First, are there realistic prospects of large-scale substitution or even complete displacement by equally satisfying, or even better, alternatives? Second, if a resource looks largely irreplaceable, what is the most likely future trajectory of use? Third, how do the likely future requirements compare to the best assessment of available raw materials?
In the case of steel, the first question is easily answered in negative. When looking half a century ahead, the conclusion based on our best engineering, scientific, and economic understanding must be that there is no realistic possibility that our civilization could do without steel. The scale of the global dependence on the metal is too large to be marginalized rapidly: we use about 33 times more of it than aluminum, and nearly 6 times more of it than all plastics combined. The industry’s amortization spans are several decades long: blast and oxygen furnaces and continuous casters are not built to be discarded in a few years. There is no doubt that, based on the historical experience, we will use less steel per unit of economic output or per mass of a specific durable product (benefits of relative dematerialization) and that in mature economies with near-stationary or declining populations we will see significant declines of absolute steel consumption (although in the longer term this process may be reversed by a major influx of migrants and renewed population growth).
Steel will remain the most commonly used metal of modern civilization. Even on a planet with a stationary population and minimal economic growth there would be considerable demand for steel needed to maintain and upgrade the existing infrastructures whose state ranges in most high-income countries from unsatisfactory to parlous. But the global population is set to increase during this century to at least 9 billion, and there are contradictory opinions regarding its eventual stabilization before the century’s end or its continuing growth after 2100. What is indisputable is that the largest share of this growth will take place in Africa and Southeast Asia where per capita in-use steel stocks are still minuscule and where developmental needs are immense.
Nearly a billion people in Asian and African countries are malnourished, more than a billion of them still have no access to electricity, and nearly twice as many do not have a proper supply of clean water and adequate sanitation. Removing these deficiencies calls for massive increases in steel consumption to generate electricity (all kinds of central and distributed plants, high-voltage transmission), to raise food production (new fertilizer factories, farm machinery, irrigation pumps and pipes, pest-proof grain storage), and to improve water supply and treatment (dams, trunk pipes, distribution pipes for in-house delivery, wastewater treatment plants, and in an increasing number of countries also new desalination capacities). And continuing urbanization, industrialization, and expanding international trade will further add to higher steel demand for decades to come. On this basis alone, it is easy to see the potential for the 2050 global steel output to be 50% above the 2015 level.
But, as with any resource, it must also be expected that the future trajectory of steel use will have to experience, and perhaps sooner rather than later, lower growth rates, and those will eventually be followed by a global production plateau—but the time of this growth deceleration and the subsequent developments are much harder to predict. The only certain conclusion (in the absence of any affordable mass-scale substitute) is that a global production peak or a brief output plateau cannot be followed by a rapid decline without triggering enormous economic and social consequences. For example, should the global extraction of iron peak in 2050 at a level 50% higher than in 2010, it could not fall to 15% or 25% of that record level in a matter of years, or within a decade, without bringing on a massive derailment of the global economy.
And “How long can our massive steel output continue?” is the most difficult question to answer as it depends on how fast we scale up, how much we scale down, how much we recyc
le, and how fast we are able to come up with affordable substitutes. The common practice of relating reserve and resource estimates to recent production levels provides some useful insights, but it does not allow us to come up with the numbers for absolute future limits. The latest USGS estimate puts the global iron ore reserves at 190 Gt, containing 87 Gt of iron (USGS, 2014). At the current rate of iron ore production (3.2 Gt in 2014), the R/P ratio is just 27 years. But the global iron ore R/P has a specific shortcoming: the largest output component (China’s extraction at 1.5 Gt/year) refers to crude ore rather than (as do the totals for all other countries) to usable ore, and properly adjusting that national total lowers the global output to 2.4 Gt and lifts the R/P ratio to 36 years.
R/P ratios are commonly cited, but they are not particularly illuminating: investment and innovation keep transferring usable materials from the resource category to economically exploitable reserves, and hence we can be absolutely certain that there will be plenty of iron ore in 27 or 36 years. Looking at the resource to production ratio is more revealing, and the USGS puts the total iron content of global ore resources at 230 Gt: with annual output at the 2014 level of 1.6 Gt, those resources would last just over 140 years, far beyond any rational planning horizon: think of the industrial planners worrying in 1871 (just 6 years after the end of the US Civil War and during the year of France’s defeat by Prussia, when the global steel output was less than 1 Mt) about the metal’s production in 2015. Obviously, concerns are different at the national level. Perhaps most notably, if iron ore exports from Brazil and Australia were to continue along their rising post-2000 trajectory, their stocks could be depleted rather rapidly (Yellishetty & Mudd, 2014).