by Vaclav Smil
Injection of pulverized coal is a method that was first used by Armco in Ashland, KY, in 1963, but (with the exception of Shougang mill in China) it found little acceptance elsewhere as long as inexpensive oil and natural gas could be used as supplementary fuels to replace part of the coke charge, a situation that changed after two OPEC-led oil price increases (McManus, 1989). Nippon Steel licensed the Armco process in 1981 as the record high oil prices led to a rapid Japanese adoption of coal injection: by 1986 half of the Japanese blast furnaces used it. The first British Steel installation came in 1982, and by the late 1980s there were nearly 50 furnaces using the process, replacing up to 40% of coke with coal (at 1:1 rate), whose cost was only 35–45% as much per unit of weight. During the 1980s, most furnaces received less than 100 kg/t, but during the 1990s many furnaces began to work with as much as 175 kg of coal per tonne of hot metal, and in 1998 US Steel Gary Works began testing simultaneous injection of natural gas and pulverized coal, replacing unprecedented shares of coke.
The cooling effect of injecting pulverized coal makes it possible to use higher hot blast temperatures and higher concentrations of injected oxygen, and this results in reduced total fuel consumption and higher furnace productivity (Danieli Corus, 2014). Nomura and Callcott (2011) made a theoretical investigation of the maximum pulverized coal injection rates and concluded that they are between 190 and 210 kg/t of hot metal. Although some large furnaces operated for short periods of time with higher injection rates (up to about 260 kg/t), totals close to 200 kg/t are in agreement with the highest reported rates used for long periods of time in stable blast furnaces. They found that the replacement of coal by coke increases the generation of CO and H2 and lowers the bosh gas temperature. Coal injection has been a superior option to investing in new coking batteries, but there are clear limits both because of the structural support and permeability provided by coke as well as due to the fact that high rates of coal injection require the enrichment of the hot blast with oxygen, offsetting the coke savings.
In 1996, NKK began the injection of coarse grains of used plastics as an alternative to coal at its No. 2 Keihin Works furnace south of the capital. The company was charging annually as much as 120,000 t of shredded plastic per blast furnace, saving 1.1 tonnes of coke per tonne of plastics, reducing annual energy use by about 1.5%, and reducing CO2 emissions by about 3.5 kg C/t of hot metal (Ogaki et al., 2001). JFE Steel (established in 2003 by the merger of NKK and Kawasaki Steel) continues the practice at Keihin Works (Fig. 5.2). Nippon Steel Corporation had also developed a new process for waste plastic recycling in coke ovens (Kato et al., 2006).
Figure 5.2 JFE’s No. 2 blast furnace at Keihin Works, south of Tōkyō. Reproduced by permission from JFE Steel.
And the late 1990s also saw the first trials of direct hot oxygen injection (as opposed to further oxygen enrichment of the hot blast) into blast furnaces: Praxair developed a thermal nozzle injecting high-temperature oxygen, and this technique should improve coal burnout by increasing oxygen levels in the vicinity of the injected powdered coal plume (Halder, 2011). Another source of increased efficiency, to be described in some detail in the next chapter, has been the charging of beneficiated (sintered or pelletized) ores.
Direct Reduced Iron
Before moving to a brief review of state-of-the-art steelmaking practices, I must devote a few pages to describing the only major practical alternative to BF ironmaking, direct reduced iron (DRI). This group of techniques obviates the use of metallurgical coke as it reduces iron ores in their solid state at temperatures well below the metal’s melting point. In principle, DRI could be thought of as a modern, efficient replication of the preindustrial, artisanal production of spongy iron masses in bloomeries, the process described in this book’s first chapter. Between 1869 and 1877, William Siemens experimented with a variant of DRI by attempting to reduce a mixture of crushed high-quality iron ore and coal in rotating cylindrical furnaces, and during the 1920s two Swedish processes were employed in a small-scale local production of iron powders. One of them (Höganäs process) is still used for that purpose by an eponymous Swedish company and by other enterprises (Höganäs, 2015).
Commercialization of DRI began only during the late 1960s, and by the mid-1970s there was a choice of nearly 100 designs combining different reactors (furnaces, kilns, retorts, fluidized-bed reactors) with a number of reducing agents, including coal, graphite, char, liquid and solid hydrocarbons, and gases (Anameric & Kawatra 2015; Hasanbeigi, Price, & Arens, 2013). DRI processes can be classified according to the kind and source of reducing gas or type of reactor used, and the DRI process that has been so far the most commercially successful relies on natural gas and hence it has been most commonly installed in locations and in countries where this fuel has been inexpensive and readily available. Natural gas is reformed by a catalytic steam process (CH4 + H2O → CO +3H2) and it reduces iron pellets or fines as it ascends, mixed with crushed limestone, in a shaft (Anameric & Kawatra, 2015). The Mexican Hojalata y Lamina was the pioneering design (with later versions marketed as HYL-III), and the most successful US contribution, MIDREX process, uses shaft furnaces, while Fior/Finmet, Iron Carbide, and Circored use fluidized-bed reactors.
MIDREX has earned its leadership in direct reduction because its plants are the industry’s most productive and most reliable (often operating for more than 8000 hours a year) and can use a range of reductants and raw materials (MIDREX, 2015). Plants can produce reducing gas from the energy source that is either locally, or most readily, available or that is most competitively priced, be it natural gas (reformed to yield CO), syngas from coal, petroleum coke ore heavy refinery residue (processed in a gasifier), or coke oven gas. Furnaces can work either with lump ore or iron oxide pellets (or their mixture), and the process itself is simple (involving the countercurrent descent of iron-bearing material loaded at the top of a cylindrical vessel lined with refractories and ascent of reducing gases).
MIDREX DRI is in the form of a solid sponge, vulnerable to reoxidation and unsuitable for transportation, and since 1984 this sponge iron has been first converted into hot-briquetted iron (HBI), produced by discharging hot DRI into roller presses that mold it into dense, pillow-shaped briquettes highly suitable as EAF charge. The other products are cold DRI (cooled before discharging) and hot DRI in the form of dense (bulk density of 2.5–3 t/m3) briquettes containing 90–94% Fe and each weighing up to 3 kg. When compared to blast furnace smelting, the process has three other key advantages besides doing away with coke. First, its specific energy requirement (GJ/t) is approximately only half of those for BF operation (for details see the penultimate section of the next chapter). Second, as a result, its specific carbon emission is also much lower (see the last section of the next chapter). Third, there is a much greater flexibility in designing plant capacities: while the economies of scale for the standard BF–BOF choice demand annual output of at least 2 Mt, DRI plants associated with a mini-mill can produce as little as 0.5 Mt/year.
Corex, the most successful smelting-reduction process, was originally developed by Voest-Alpine Industrieanlagenbau (VAI), and it is now marketed by Siemens under the name SIMETAL Corex (Siemens VAI, 2011). It requires two separate process reactors, the first being a reduction shaft charged with lump ore or pellets and additives to produce DRI in a counterflow reaction, and the second being a melter gasifier where the reduction is completed and hot metal and slag are tapped, much as in conventional BF. The first Corex plant began to operate in Pretoria in 1989, and subsequent installations included Saldanha in South Africa, Pohang in South Korea, at China’s Baosteel (two units), and five units in India (JSW Steel and Essar Steel).
Rotary hearth furnaces (RHFs) have been used for decades in heat treating metals and high-temperature recovery of nonferrous metals, and hence their use in ironmaking has been a matter of specific application and appropriate process control. RHFs—developed and marketed under proprietary labels of Fastmet, Fastmelt, Redsmelt, Sidcomet, Primus, and lTm
lk3—now constitute the largest class of new direct reduction processes but account for a minority of global DRI output (Anameric & Kawatra, 2015; Guglielmini & Degel, 2007; McCelland, 2002; Sohn & Fruehan, 2006). Their flat refractory hearths rotate inside high-temperature circular tunnel kilns lined with refractories, with iron ore and reductant in a single- or multi-layer bed, and with temperature controlled by burners (using natural gas, fuel oil, or pulverized coal) along the wall and on the roof. Mixed ore-and-coal pellets are subjected to temperatures up to 1300 °C, mostly for just 6–12 min.
The product is not liquid and, inevitably, it contains relatively large amounts of ash and nonmetallic residues from the processed coal–ore mixture, and it must be melted in EAF. The process requires constant feeding and has limited productivity. The first commercial RHF, designed to recycle wastes containing Ni and Cr at INMETCO in Elwood, PA, went into operation in 1997 (INMETCO, 2015). MIDREX began developing an RHF process concurrently with its countercurrent reactor, using reformed natural gas, abandoned the quest in favor of the latter technique, and returned to it by 1992 with the development of the Fastmet process. The first Japanese plant has been operating since the year 2000 at Hirohata Works, and there are now six plants with combined annual capacity of nearly 1 Mt (Kobelco, 2015a). The process is particularly suitable for converting such iron- and steelmaking wastes as blast furnace dusts and sludges and EAF dust and mill scale.
Obvious drawbacks are the need for constant feeding and a limited productivity. Fastmelt is Fastmet with an added electric iron melting furnace to produce hot DRI. Currently the most advanced RHF design (which can be seen as a variation of Fastmet) is the ITmk3 (Ironmaking Technology Mark 3, BF being the first, and gas-fueled DRI the second generation) process developed by Kobelco Steel (Harada & Tanaka, 2011; Kobelco, 2015b).
Iron ore concentrate (magnetite or hematite or their mixtures) and noncoking coal mixed in pellets are processed for 10 min by a single-stage heat treatment in an RHF, yielding high-quality (96–97% Fe), slag-free iron nuggets. These nuggets are easier to handle than DRI or hot-briquetted iron and are processed in EAF or (after remelting) in BOF. Because nearly all combustible gases generated by the reactions are burnt within the furnace, the process can reduce specific CO2 emissions by 8–17% compared to BF. The first plant, at Hoyt Lakes, MN, began operating in 2010 and it had an annual capacity of 500,000 t.
Not surprisingly, during the early years of its commercialization DRI received enthusiastic reception and there were high expectations for its continuing strong expansion, if not an eventual dominance of primary iron production. Global DRI production rose nearly 10-fold during the 1970s (to 7.1 Mt), and Miller (1976) forecast the output of 120 Mt by 1985—but the global capacity in that year was just over 20 Mt, and actual sales were only 11 Mt (MIDREX, 2014a). Expectations shifted, as slow but steady growth came to represent DRI success. Global output rose to 31 Mt in 1995, 43 Mt in 2000, and 75.22 Mt in 2013, and DRI’s share of the global primary iron production (excluding the output of RHFs using recovered mill wastes) rose from 1% in 1980 to 4.7% in 2013.
In 2013, there were more than 120 DRI plants worldwide and MIDREX (now owned by Kobe Steel) remains the premiere DRI licensor as its plants have been supplying 60% of all DRI for more than two decades and produced 63% of all DRI in 2013 (MIDREX, 2014b). In 2013, the largest of the 56 plants were in Mobarakeh in Iran (capacity of 3.2 Mt in five modules) and in Corpus Christi, TX (2 Mt). The other leading shaft reduction process (HYL/Energiron) supplied about 15% of all DRI in 2013, while coal-based RHFs delivered about 21%. Maximum annual capacity of MIDREX installations grew from less than 0.5 Mt (Series 400) to more than 3 Mt (MEGAMOD series), and the oldest MIDREX reactors have been in service for more than 40 years (MIDREX, 2014b). The Middle East is the leading (natural gas-based) producer with nearly 40% of the global total; India, with more than 40 (mostly coal-based) plants, is the leading nation (nearly 25% of the total), and both the United States and the EU are only negligible producers.
Basic Oxygen Furnaces
Because open hearth furnaces dominated global steel production until the 1960s, their capacities had to increase to accommodate the rising demand. The largest furnace owned by US Steel in 1910 had a hearth area of about 40 m2 and capacity of 42 t, while the record-sized furnace introduced during WW II had a hearth of 85 m2 and produced 200 t of steel in a single heat (King, 1948).
OHFs had benefited from the post-WW II expansion of steel production when they reached their greatest scale in both the United States and in the USSR, with new units capable of producing more than 600 t of steel in a single heat, and the world’s largest furnace in the USSR had a capacity of 900 t. But by the time it went into operation the dominance of OHFs was over: in the early 1950s, it appeared that they would remain in demand for decades to come, but by the end of the twentieth century they produced just 4% of all steel, with Russia and Ukraine being the only two major producers with high shares of open hearths, respectively 27% and 50% (WSA, 2001).
This was the result of a veritable technical revolution, but because it unfolded during the same time as the advances of the electronic world (that have always received disproportionate media attention), its impressive achievements have received hardly any public appreciation. As a result, Ohashi (1992, p. 540) had to conclude that
in the popular conception, steelmaking is the quintessential example of an outmoded manufacturing technology, a 19th-century dinosaur that is about to lumber obliviously over the threshold of the 21st century. But in fact, steelmaking has been transformed in the past 20 years by a flood of innovations. With little fanfare, it has become as impressive as that acme of modern manufacturing practice, integrated-circuit processing.
This flood of innovations, which brought the end of the open hearth furnace era, began during the early 1950s when the European steelmakers pioneered the adoption of BOFs, fundamentally nothing but improved variants of the old Bessemer converter: there is no external source of heat and the process relies on the exothermic reaction between iron and oxygen. In fact, Bessemer obtained a British patent (UK 2,207, on October 5, 1858) that envisaged the blowing of oxygen to decarburize iron. That idea could not be transformed into commercial reality as long as oxygen was not available in large volumes and at an affordable cost. The road to that was opened by Mathias Fränkl’s tubular regenerator (heat exchanger), patented in 1928 and incorporated into the Linde-Fränkl process, whose first commercial installation came in 1934. New regenerators were very efficient, and they were a major reason why the price of oxygen produced by the new Linde process offered to industrial consumers in 1950 was reduced by an order of magnitude.
As is the case with Bessemer or open hearth furnaces, the adjective basic does not refer to any essential quality or the simplicity of design but to the pH level of the slag: magnesium oxide (MgO) in the furnace’s lining is used to remove and retain trace quantities of P and S from the molten metal. Added slag also prolongs the lining’s life, and after every competed heat the vessel is rocked to distribute the slag up its sides. The development and commercialization of BOFs were remarkable for the absence of any of the world’s major steelmakers: they were driven by the dedication of a Swiss metallurgist and the foresight of managers in two small Austrian companies.
After Swiss metallurgist Robert Durrer (1890–1978) graduated from the Aaachen University in 1915, he remained in Germany, and in 1928 he became a professor of Eisenhüttenkunde at Berlin’s Technische Hochschule, where he began to experiment with oxygen in steel production. When he returned to Switzerland in 1943, he joined the board of the country’s largest steel company, von Roll AG, and, helped by German metallurgist Heinrich Hellbrügge, he continued his research at a steel plant in Gerlafingen (Starratt, 1960). As is often the case with major inventions, Durrer had competitors: C.V. Schwarz obtained a German patent (735,196) for a top-blown oxygen converter in 1943, and John Miles received a Belgian patent (468,316) in 1946 (Adams & Dirlam, 1966).
But ne
ither of these ideas was translated into a working prototype before Durrer made a breakthrough in April 1948 when a small converter bought during the previous year in the United States was used to produce steel by blowing with pure oxygen. Moreover, Durrer and Heinrich Hellbrügge, the metallurgist in charge of the experiment, established that cold scrap could amount to as much as more than half of the hot metal weight. Soon afterwards, Hellbrügge notified Herman Trenkler at VÖEST (Vereinigte Österreichische Eisen- und Stahlwerke AG, the country’s largest steelmaker) that the oxygen process was ready to be tried in a commercial setting, and von Roll, VÖEST, and Alpine Montan AG quickly concluded an agreement for such testing at VÖEST’s Linz plant and at Alpine’s Donawitz plant.
Initial operation problems were overcome in a matter of months; construction of two 30-t oxygen furnaces began in December 1949, and VÖEST produced its first steel on November 27, 1952, and Alpine in May 1953 (Starratt, 1960; Stubbles, 2015). That is why the new method became commonly known as the Linz-Donawitz process (das LD-Verfahren) and why many people considered it, wrongly, an Austrian invention. Given its intellectual and practical origins, the method should be known as the Bessemer–Durrer process. VÖEST eventually became the sole proprietor of the L–D process, but the world’s major steelmakers were in no hurry to license new oxygen furnaces from a small Austrian company whose aggregate capacity was less than a third of the output of a single plant of US Steel (Adams & Dirlam, 1966). Although the advantages of the process were widely recognized by US steelmakers, corporate inertia among large companies and their high sunk costs in open hearth furnaces meant that all major US producers stayed far behind the oxygen wave (ASME, 1985; Emerick, 1954; Hogan, 1971).