The 2004 Brimacombe Memorial LecturePublished with Permission of AIST

Research on Sustainable Steelmaking

The international steel community is faced with the challenge of developing processes that willmake steel production more sustainable in the future. Specifically, processes that produce lessCO2 and less net waste materials and emissions and that consume less energy are required.This article outlines where energy consumption and CO2 emissions are high and can bereduced. Reductions can be achieved by incremental improvements to existing processes or by a‘‘break-through innovative process’’; both strategies are examined. Since most of the energyconsumption and CO2 generation occur in ironmaking, research in this area is emphasized.Research on controlling the cohesive zone in the blast furnace, improving the final stages ofreduction in direct reduction processes, the use of biomass, and other innovative processes forironmaking are reviewed. In oxygen steelmaking, improved postcombustion (PC) to allow formore scrap melting is examined. Postcombustion and slag foaming in the electric arc furnace(EAF) in order to reduce energy is reviewed.

DOI: 10.1007/s11663-008-9223-x� The Minerals, Metals & Materials Society and ASM International 2009

I. INTRODUCTION

IT is not only a great honor to give the BrimacombeLecture but also a large responsibility. It is a responsi-bility to be sure that the Lecture is on a significantsubject, makes our community sensitive to long-termcritical issues, and is consistent with Professor Brima-combe’s vision of what our industry should be doing inthe future. With this goal, I have chosen to discusssustainable steelmaking and examples of research aimedat this goal.

A. Current Status

Before discussing specific research projects, the over-all goal of sustainability will be reviewed. Sustainable

R.J. FRUEHAN

Institute of Japan and served as President of the Iron and Steel Societyof AIME from 1990 to 1991. He was elected a Member of the NationalAcademy of Engineers in 1999.

R.J. FRUEHAN, U.S. Steel University Professor and Co-Director,is with the Center for Iron and Steelmaking Research, MaterialsScience and Engineering Department, Carnegie Mellon University,Pittsburgh PA 15213. Contact email: fruehan@cmu.edu

R.J. Fruehan is currently the U. S. Steel University Professor atCarnegie Mellon University. He received his B.S. and Ph.D. degreesfrom the University of Pennsylvania and was an NSF Scholar atImperial College, University of London. Dr. Fruehan organized theCenter for Iron and Steelmaking Research, and is currently theCo-Director. He was Director of the Sloan Steel Industry Study, whichexamines the critical issues impacting a company’s competitiveness andinvolves numerous faculty at several universities from 1992 to 2002.Dr. Fruehan has authored over 250 papers, two books on steelmakingtechnologies, co-authored a book on managing for competitiveness,and is the holder of six patents. He has received several awards,including the 1970 and 1982 Hunt Medal (AIME), the 1982 and 1991John Chipman Medal (AIME), 1989 Mathewson Gold Medal (TMS-AIME), the 1993 Albert Sauveur Award (ASM International), the1976 Gilcrist Medal (Medals Society UK), the 1996 Howe MemorialLecture (ISS of AIME), the 1999 Benjamin Fairless Award (ISS ofAIME), the Brimacombe Prize (ISS, TMS, CSM) (2000), the 2004Bessemer Gold Medal (Institute of Materials, Minerals & Mining(UK); an IR100 Award for the invention of the oxygen sensor andthe TMS Science Award (2008). He is a Distinguished Member of theIron and Steel Society, an Honorary Member of the Iron and Steel

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—123

Development has been defined as ‘‘development thatmeets the needs of the present without compromisingthe ability of future generations to meet their needs.’’

Specifically, for sustainable steelmaking, the goalsinclude

(1) conserving resources such as ore, coal, etc.;(2) reducing the amount of CO2 emitted;(3) reducing other gaseous emissions such as SO2 and

Nox;(4) reduce materials to landfills; and(5) reduce the amount of hazardous wastes.

Steel production remains high and continues to grow.Steel production occurs in virtually every major countrywith about 50 pct produced in five countries. Much ofthe future increases will be in developing countries suchas China, India, and those in Eastern Europe. Energyconsumption is about 18 to 20 GJ per tonne. The steelindustry in the United States consumes about 18 to20 GJ/t or about 2 to 3 pct of the total energy used inthe United States.

In order to determine where technical resourcesshould be targeted, it is useful to examine the energyused in the individual processes and what the practicaland theoretical limits are. Normally, CO2 emissions arerelated to the energy used in a process. As seen inTables I and II, hot metal, or ironmaking, is by far thelargest energy consumer and CO2 emitter. Therefore,ironmaking has the greatest potential for improvement.The electric arc furnace (EAF) steelmaking hasimproved significantly and is close to its practicalminimum. In steelmaking, the major recoverable energyis the energy in the off-gas (CO). For hot rolling, startingwith section sizes such as thin slab or strip reduces

energy and CO2. However, substantial savings withconventional slab are possible by direct charging, sincereheating is the major energy consuming part of theprocess. In fact, 75 pct of the potential savings of thin oreven strip casting can be achieved with conventionalslabs if they are directly charged.There has been a dramatic reduction in energy

consumption in the past 30 years, from about 35.8 to18.0 GJ/tonne of shipped steel in the United States. Thishas been accomplished primarily due to the implemen-tation of continuous casting, other yield improvements,a shift to scrap-based EAF production, and a decreasein the blast furnace fuel rate. Future reduction will bemuch more difficult, since most of the ‘‘low hanging fruithas been picked.’’

B. Strategy for Research on Sustainability

Since CO2 abatement and energy reduction are closelyrelated and the major challenge, I will concentrate onthis issue. Other emissions such as SO2 and possibleparticulate matter will also be generally reduced ifenergy consumption is reduced.There are at least four possible alternatives for

lowering the amount of CO2; reducing the amount ofCO2 emitted in the processes, using renewable energysources such as wood charcoal, using hydrogen, andsequestering or capturing the CO2. Hydrogen will not beavailable in large quantities at a reasonable price. Theavailable hydrogen will be primarily used for othertechnologies such as fuel cells and transportation.Sequestering the CO2 is a reasonable partial solutionformany industries including power generation. The steelindustry should not use ‘‘its’’ meager resources on thissolution but rely on development from other industries.However, the industry should examine processes, whichare consistent with sequestering technologies.For example, sequestering CO2 for steelmaking may

be simply accomplished by injecting CO2 into deepwells. This requires, as well as other sequesteringtechniques, a relatively pure CO2 stream. Research onan oxygen blast furnace with a hydrogen shift reactorand gas circulation can produce a nearly pure CO2

off-gas. Circofer (Lurgi, Germany), which uses coaland ore in a fluid bed–type process, can accomplish thesame goal. The resulting gas is nearly pure CO2, whichis more easily sequestered. Research in these areasshould be considered.In the iron and steelmaking processes, one strategy is

continuous incremental improvements with modestgains, but with the total being significant (10 to15 pct). Most likely this new technology will be iron-making, because 75 pct of the energy is consumed inmaking iron. The new process should eliminate coke-making and use fine ores, and in special circumstances, arenewable energy source.Which strategy should be adopted? Both. We need to

use the capital invested in our existing technologieswhile reducing energy in these processes in the shortterm, 3 to 5 years. However, for the longer term, 10 to20 years, a new technology, or technologies, may benecessary.

Table I. Actual, Practical Minimum, and TheoreticalMinimum Energy Consumed in Various Processes, Energy

(GJ/t Product)

Process ActualPracticalMinimum

TheoreticalMinimum

Hot metal 13 to 14 10.4 9.8Steelmaking (BOF) 0 (0.5) (1.0)Steelmaking (EAF) 2.1 to 2.4 1.6 1.3Hot rolling 2.0 to 2.4 0.09 0.03Cold rolling 1.0 to 1.4 0.2 0.02

( ) Indicate energy available.

Table II. CO2 Emissions for Various Steelmaking ProcessesCO2 Emissions (kg/t)

Process TypicalPracticalMinimum

TheoreticalMinimum

Hot metal 1500 1160 1090Steelmaking (BOF) 200 144 144Steelmaking (EAF) 380 280 225Rolling 320 60 10

124—VOLUME 40B, APRIL 2009 METALLURGICAL AND MATERIALS TRANSACTIONS B

II. RESEARCH IN EXISTING PROCESSES

In this section and Section III, research beingconducted at the Center for Iron and SteelmakingResearch (CISR) at Carnegie Mellon University aimedat reducing energy and CO2 emissions is discussed.There is extensive excellent research being done world-wide including Europe, Japan, and other institutionsin North America. I am concentrating on the CISRresearch program, because I have more personal knowl-edge of that program.

Ironmaking: Two recent research programs in iron-making designed to make incremental improvements inthe processes are controlling the cohesive zone in theblast furnace and understanding the final stages ofreduction by gases.

The cohesive zone begins when the ferrous materialsbegin to soften and ends with final meltdown. Thecohesive zone impedes proper gas flow through thefurnace and decreases indirect reduction by CO and H2.The cohesive zone should be deep in the furnace and asshort as possible. The cohesive zone can be improved bycustom designing the ferrous burden and the chargingpattern. Pellets, sinter, and ore are often subjected to anarbitrary softening and melting test, which gives someguidance as to how they behave in the blast furnace butdoes not reflect the fundamental behavior that is theobjective of the present research.

The recent work[1] included detailed characterizationof the ferrous materials, melting on-set experimentsusing a laser confocal microscope, softening and melt-ing with X-ray observations, quenched samples, and

thermochemical modeling of phase diagrams to predictliquid slag formation. As seen in Figure 1, it is possibleto see slag formation in this acid pellet. However, theinitial amount of liquid may be small and due to a minorconstituent such as Na2O. More importantly is thetemperature when large amounts of liquid form and theviscosity of the liquid. This will be discussed later indetail. The lower its viscosity, the greater the impact theliquid slag has on the behavior of the pellets.Experiments were run in which six prereduced pellets

were heated under load with the displacement measuredalong with the temperature (Figure 2). Typical displace-ment vs temperature curves are shown in Figure 3. Inaddition, X-ray pictures of the pellets were taken;examples are show in Figure 4. As can be seen, the acidpellet at 1201 �C has deformed significantly with adisplacement of about 63 pct. The basic pellet reached1310 �C for a similar displacement, and the meltdowntemperature range was only about 40 �C. In separateexperiments, the samples were quenched and the phasespresent determined. Calculations using FactSage (EcolePolytechnique, Montreal, CA) were conducted to deter-mine the phases present. For example, the phases for 60prereduction of a basic pellet are shown in Figure 5. Asmall amount of liquid forms at about 1230 �C and asignificant amount of liquid forms at 1300 �C, incorrespondence with the observations.Another interesting observation is that, with the acid

pellets, slag exuded from thepellet interactingwith adjacentpellets; this did not occur with basic pellets. It wasconcluded that this is because the viscosity of the slag in

Fig. 1—SEM pictures from quenched samples from a laser confocal microscope.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—125

theacidpelletswasmuch lower thanthat in thebasicpellets.The reason the viscosity was much lower is the higher FeOcontent of the slag, about 60 pct vs less than 30 pct.

The fundamental behavior determined in this studywill aid in pellet composition design and the chargingpattern when using mixed burdens. Details of this workcan be found in a series of recent publications byNogueira and Fruehan.[1]

The other major method of producing iron is solid-state gaseous reduction or direct reduction. This iron isprimarily used in an EAF. Therefore, a high degree ofreduction or metallization is desirable since the reduc-tion of the remaining FeO by carbon is endothermic,resulting in higher electrical energy consumption and

Fig. 2—Experimental setup for displacement vs temperature mea-surements on X-ray picture.

Fig. 3—Displacement vs temperature curves for acid and basicpellets.

Time (s) 0 23098 24357 25237 25479 30837 T (oC) 279 1141 1201 1236 1254 1502 D (%) 0 45.3 63.4 75.3 80.1 100

Time (s) 0 22526 25666 26328 26792 27374 T (oC) 276 1162 1310 1341 1358 1392 D (%) 0 20.9 60.3 69.3 82.7 100

Fig. 4—X-ray pictures of acid and basic pellets during displacement vs heating experiments.

126—VOLUME 40B, APRIL 2009 METALLURGICAL AND MATERIALS TRANSACTIONS B

lower productivity in the EAF. In the direct reductionprocesses, in particular fluid bed processes such asFinmet or Circored, it is often difficult to obtain highdegrees of reduction, greater than about 92 pct. Thisphenomenon depends on the ore type and process—inFigure 6, the reduction of Mount Newman concentrate(MNC) and a hematite single crystal (HSC). The MNCreaches a fairly high degree of reduction, where HSClevels off at about 75 pct reduction.[2] This phenomenadepends on the ore type and on the overall reductionprocess.

The ore particles were examined by SEM analysis,and it was found that the rate slowed when a dense layerof iron formed around the unreacted FeO, as shown inFigure 7. When this occurs, gaseous diffusion is notpossible and the only mechanism for further reduction isatomic diffusion of oxygen through the iron layer. Thisprocess is exceptionally slow because the oxygen solu-bility in solid iron is only a few parts per million, andhence, the driving force for diffusion is low.

The reason why the dense iron formed in some cases isnot completely known. One significant observation was

that the rate of reduction slowed significantly, and thewustite grains were relatively large, 1 to 2 lm vs 3 to5 lm, when the rate slowed.The rate mechanism is shown schematically in Figure 8.

If there is no iron layer, or it is very thin, the rate or fluxor oxygen removal of removal is given by

JO ¼ �WFeOSk PH2� PHe

2

� �½1�

where

JO=flux of oxygen,WFeO=weight of FeO,k=chemical rate constant, andPH2

;PHe2=hydrogenpressure and equilibriumpressure.

When the iron layer becomes sufficiently thick, therate becomes controlled by diffusion of oxygen throughthe solid iron. If spherical FeO grains are assumed for

Fig. 5—Calculation of phases present as a function of temperaturefor a 60 pct reduced basic pellet.

Fig. 6—Rate of reduction of MNC and hematite single crystals at973 K in 60 pct H2-40 pct N2.

Fig. 7—SEM picture of HSC partially reduced showing dense ironlayer around wustite grains.

FeO

FeO

Fe

Fe H2 → 2H

2H + O → H2O

H2

O

H2 → 2H

H2

H

H2 + FeO → H2O + Fe

PH2+ PH2O >1

Fig. 8—Schematic diagram showing rate mechanism for solid-statediffusion through an iron shell.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—127

simplicity, the diffusive flux through the iron layer isgiven by

JO ¼DO

1r � 1

r0

½2�

where

DO = diffusivity of oxygen in Fe,r = radius of reaction front, andr0 = initial radius of the particle.

The rate of reduction is controlled by the slower of thetwo mechanisms or the one whose rate equation givesthe slowest rate.

Whereas not all of the parameters required in Eqs. [1]and [2] are known, some comparisons can be made.Using published values for the parameters indicates thatsolid-state diffusion becomes important at about 1 to2 lm.

Obviously, the grain size alone is an oversimplifica-tion for the slowing of the rate for some ores and notothers. However, the influence of grain size can beillustrated by the following example. Consider two ores,A and B, for which the rates of gaseous reduction aresimilar, i.e., they have similar values for the rateconstant k. For ore A, the wustite grain size is about2 lm and the rate slows at about 60 pct reduction fromwustite to iron (74 pct from hematite). The value of r atthis point, where solid-state diffusion governs the rate, is1.2 lm. If for ore B the average grain size is 1 lm, thevalue of r at which the critical rate is achieved is0.15 lm, which represents over 99 pct reduction beforethe rate slows. Observations indicate there is a distribu-tion of grain sizes and not all grains have a dense ironlayer. Nevertheless, this simple example illustrates theimportance of the wustite grain size on when the ratemechanism changes from chemical to diffusion controland the critical degree of reduction that can be expected.

Another interesting finding is that, when the equilib-rium pressure of (H2+H2O) at the Fe-FeO interfaceexceeds one atmosphere, the iron shell can fractureallowing for gaseous diffusion. In this case, hydrogen

atoms can diffuse through the fractured iron layer andreact with the FeO. If the total pressure is less than oneatmosphere, the gas cannot escape and the reaction doesnot proceed. As shown in Figure 9, with a hydrogenpressure of 0.6 atm, reduction virtually stopped at70 pct. When the hydrogen pressure was increased to1.0 atm, the equilibrium pressure at the Fe-FeO inter-face exceeded 1 atm and presumably cracked the ironshell and allowed reduction to proceed. Also, holding orannealing of the samples at the magnetite (Fe3O4) for aperiod of time reduction was lowered for some ores, asshown in Figure 10. Presumably, this may cause thegrains to coarsen, resulting in larger more difficult toreduce wustite grains. The results of this research haveled to improved processing to achieve higher degrees ofreduction.

A. Oxygen Steelmaking

Research has been conducted on the recycling of wasteoxides and postcombustion (PC), which will be discussedin this lecture. Postcombustion is the combustion of theCO produced by decarburization to CO2 above theliquid metal bath, releasing energy. In the basic oxygenfurnace (BOF), the energy can be used to melt morescrap, by about 10 pct of the charge, or to melt skulls.However, if PC is not done properly, the CO2 can reactwith the Fe-C melt (depostcombustion) or the heat maybe released too high in the furnace and not transferred tothe bath, just heating the gas or refractories.Computational fluid dynamics (CFD) modeling was

conducted to optimize the process.[3] The modelincluded an accurate combustion reaction, all relevantforms of heat transfer, and depostcombustion reactions.The BOF lance had 16 PC nozzles. The problem couldbe simplified since there are axial systems, so the modelwas a 3-D model of 1/16 of the furnace. Examples of thecalculated results are shown in Figures 11 and 12, whichshow the CO2 concentration in the gas and the temper-

ature, respectively. This was for 750Nm3

s of O2 with 18 pct

Fig. 9—Experimental results indicating the effect of increasinghydrogen pressure on the rate of reduction of HSC at 973 �C.

Fig. 10—The effect of holding reduction at Fe3O4 for 120 min(degree of reaction is from Fe3O4).

128—VOLUME 40B, APRIL 2009 METALLURGICAL AND MATERIALS TRANSACTIONS B

of the oxygen used in 10 nozzles 1.5 m above the bath.This study demonstrated that CFD could be a usefultool in optimizing PC and using the energy associatedwith the CO effectively. Different cases with respect tooxygen distribution between the primary and PC nozzlesand height of the PC nozzles were investigated to findthe optimum condition.

B. Electrical Furnace Steelmaking

The steel industry can reduce energy consumption andCO2 generation by expanding EAF production to high-quality steels replacing integrated production. This willbe limited due to the limited quantities of high-qualityscrap. Processes are being developed and improved toproduce scrap substitutes such as DRI, HBI, and pigiron. However, these processes consume as much energyas the blast furnace.For the existing EAF process, two examples of

research aimed at reducing energy consumption areimproved, slag foaming and PC.Slag foaming is widely used in the production of

carbon steels and extensive research on the fundamen-tals has been conducted. Slag foaming allows for longarc operation during flat bath operating conditions byusing the foam to shield the sidewall refractories fromthe electrical arc. However, for stainless steels, slagfoaming is much more difficult to achieve. This could bedue to one of two reasons: the slag has poor foamingcharacteristics or sufficient gas is not generated to createthe foam. The ability of a slag to foam is given by thefoam index which is only a function of the slagproperties and defined by Eq. [3].

Hf ¼XQ

A½3�

Hf = foam height,Q = gas flow rate,A = area of foam region, andR = foam index.

The foam index in laboratory experiments was foundto be similar to that for carbon steelmaking slags as longas there was not an excessive amount of solid undis-solved slag particles. Typical results are shown inFigure 13 in which the foam height is plotted vs thesuperficial gas velocity or (Q/A); the slope is the foamindex (R).The foam index for stainless steelmaking slags is

similar to that for carbon steels. Therefore, the foama-bility or foam index is not the reason for poor foaming.Another possibility for the poor foaming is the lack ofgas generation. For foamy slag practices, CO gas isnormally generated by injection of carbon into the slagproducing CO by the reaction

FeOð Þ þ C ¼ Feþ CO ½4�

However, for stainless steelmaking slag, there is littleFeO and the most reducible oxide is chromic oxide(‘‘CrO’’). The rate of reaction of carbon with slagscontaining FeO and CrO, in which the amount of COwas generated, was measured in laboratory experiments.Typical results for FeO containing slags are shown inFigure 14. For CrO containing slags, the rate wasextremely slow, less than 10 pct of the rate for similaramounts of FeO.It therefore can be concluded that the lack of foaming

in stainless steelmaking is due to poor gas generation.

Fig. 11—CO2 concentration with PC in a BOF.

Fig. 12—Temperature profile in a BOF with PC.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—129

Consequently, foaming should be able to be improvedby making additions that generate gas. A number ofadditions were examined,[5] such as waste oxide bri-quettes, nickel oxide, calcium nitrate, and limestone(CaCO3). For example, when CaCO3 is added to steel,which promoted foaming, it dissociates at about 900 �C:

CaCO3 ¼ CaOþ CO2 ½5�

Typical experimental results are shown in Figure 15.It was found that the rate was controlled by heattransfer to the CaCO3 particle. Based on these results,the amount and rate of CaCO3 addition were calculatedto give the desired foam level.

Postcombustion in the EAF was examined using asimilar CFD model that was used for the BOF. Themodel was more complex due to the lack of symmetryand the presence of scrap. The model was used toexamine a system with gas injections in the sidewall ofthe furnace. Typical results are shown in Figures 16 and17 in terms of the CO-CO2-O2. As seen in Figure 16, theO2 is consumed, producing CO2 at the metal surface.

CO2 re-reacts with carbon in the metal depostcombus-tion. In Figure 17, the gas profiles and velocities areshown. In this case, CO2 is reacting with the electrodes.The model also computes the energy transferred to the

Fig. 13—Determination of the foam index for stainless steelmaking slags by plotting the foam height vs the superficial gas velocity.

Fig. 14—Rate of reduction of FeO in EAF slags by carbon.Fig. 15—Rate of gas evolution from CaCO3 in slag.

Fig. 16—Gas composition profiles for PC in an EAF breakthroughtechnology.

130—VOLUME 40B, APRIL 2009 METALLURGICAL AND MATERIALS TRANSACTIONS B

scrap and bath and the electrical energy saved. The CFDmodel can be used to optimize the process with regard tothe PC injector’s position and angle and the O2 flowrate.

The international steel industry has been searchingfor an alternative to the blast furnace for over 50 years.The blast furnace continues to be the dominant iron-making process, because its efficiency has continuouslyimproved and no alternative, with the exception of thedirect gaseous reduction process, in unique situations,has proven to be economically competitive. However,due to the current need for a lower capital cost processand the pressure for CO2 abatement, new processes arestill required. It has become evident that no singleprocess will fit all needs, but different processes will beused for different requirements.

Any new process should be in ironmaking, where75 pct of the energy and most of the CO2 is emitted. Theprocess should use fines and eliminate cokemaking. Ifpossible, the process should use renewable energy. Inaddition, the process should require less capital and beeconomical on a small scale.

New processes to produce liquid hot metal, which arebeing commercialized, include COREX, HIsmelt, andprocesses using ore and coal pellets or mixtures in rotaryand multihearth furnaces followed by electric furnacemelting. Examples of the latter processes include IronDynamics, Redsmelt, and Sidcomet. Whereas theseprocesses appear attractive for several applications, theydo not reduce CO2 emissions. Obviously, natural gas–based processes such as Midrex, HyL, and FINMETreduce CO2, but natural gas availability is limited.

Research is being carried out on two ironmakingprocess concepts that could reduce CO2 emissions. Thefirst is a renewable energy process[6] using ore-woodcharcoal pellets in a rotary hearth furnace, which arereduced to about 70 to 80 pct. The prereduced mixtureis then added to a coal-based smelter, such as an AISI orHIsmelt, for final reduction and gangue separation bymelting. The smelter is run at a conservative 30 pct PCwith the off-gas used in place of natural gas in the rotaryhearth. A schematic diagram of the process is shown inFigure 18. By using wood charcoal, the net CO2

emissions are reduced by 70 to 95 pct due to the fixationof CO2 when regrowing the wood. By combining the

RHF with the smelter, productivity can be significantlyincreased (30 to 50 pct) and energy and be decreased.Extensive research has been carried out on the

reduction of charcoal pellets. Reduction takes place bymeans of two reactions: reduction of iron oxide (‘‘FeO’’)by CO followed by oxidation of carbon by CO2.

‘‘FeO’’ þ CO ¼ Feþ CO2 ½6�

CO2 þ C ¼ 2CO ½7�

The fundamental rate constants for these reactionswere determined using small samples when heat andmass transfer can be neglected. Typical results areshown in Figure 19. Using these constants, and inde-pendently determined heat-transfer coefficients, a reduc-tion model for pellets was developed and verified, asshown for typical results in Figure 20. The pellet modelwas then incorporated into a three-stage model for theRHF. From this model, the production rate, as afunction of desired reduction and pellet depth, wasdetermined; typical results are shown in Figure 21.Finally, the RHF model was linked to an energy andmaterials balance model for the smelter to determine theoverall performance of the process. The overall processhas reasonable productivity and energy consumptionand greatly reduces net CO2 emissions.Another method to reduce CO2 emissions is based on

using hydrogen. However, a critical question is where toget the hydrogen. Natural gas derives about half itsenergy value from hydrogen, but its availability islimited. Many coals have high hydrogen or volatilecontents, which normally cannot be used for cokemak-ing. A new process concept based on using high volatilecoals, which are charred at a lower temperature, iscurrently being investigated. A schematic of the processis shown in Figure 22. The char is used in a HIsmelt unitand the hydrogen-rich off-gas can be used for directreduction or other uses, such as fuel cells. Smelters workextremely well using char in that there is much higherPC and productivity. For example, HIsmelt hasachieved over 70 pct PC using char.

III. CONCLUDING REMARKS

The steel industry is a major consumer of energy andproducer of CO2. Work must continue to incrementally

Fig. 17—Gas profile and velocities for PC in an EAF.

Hot metal

OxygenHigh energyoff gas

75-80% Pre-reduced hot iron pellets

Composite pellets

Fig. 18—Schematic diagram of renewal energy steelmaking processusing wood charcoal with a rotary hearth furnace and a smelter.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—131

lower energy and CO2 in existing processes. In aparallel and equally important effort, a new break-through ironmaking process should be explored andprocess models developed. The use of renewable energyand high hydrogen containing coals should be encour-aged. Steel companies alone do not have the resources

to meet these goals. We need international collabora-tion and government assistance. Finally, there shouldbe incentives for lowering CO2, not just penalties, suchas carbon taxes.

ACKNOWLEDGEMENTS

The author thanks the many students and research-ers who conducted this research and the member com-panies of the Center for Iron and SteelmakingResearch, the American Iron and Steel Institute, andthe United States Department of Energy for theirsupport.

REFERENCES1. P. Nogueira and R.J. Fruehan: Metall. Mater. Trans. B, 2004,

vol. 35B, pp. 829–38.2. R.J. Fruehan, Y. Li, L. Brabie, and E.J. Kim: Scand. J. Met., 2004,

vol. 34, pp. 1–7.

0

0.02

0.04

0.06

0.08

Time (min)

Mas

s lo

ss (

g)

Kc = 5.0x10-4 mol/mol.s.atm

KFeO = 4.0x10-3 mol/mol.s.atm

Mixed reaction model

Experiment

(a)

0.8081g sample

0

0.02

0.04

0.06

0.08

0.1

0 100 200 300 400 500

0 100 200 300 400

Time (min)

Mas

s lo

ss (

g)

Kc = 5.0x10-4 mol/mol.s.atm

KFeO = 4.0x10-3 mol/mol.s.atmMixed reaction model

Experiment

(b)

0.4745g sample

Fig. 19—(a) and (b) Determination of the rate constants forreduction of FeO by CO and oxidation of carbon by CO2 in FeO-charcoal mixtures.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0Time (min)

Mas

s lo

ss (

g)

Temperature: 1280°C

Large Pellet

Mass: 3.8599gInitial Diameter: 16.7mmFinal Diameter: 12.4mmWood charcoal/hematite: 0.1766

Simulation with final radius

Simulation with initial radius

Experiment

108642

Fig. 20—Experimental and model results for the reduction of ore-wood charcoal pellets.

Fig. 21—Production rate for varying pellet layer as a function ofdegree of reduction.

Fig. 22—Schematic diagram of a conceptual process to use highvolatile coals to produce hydrogen and liquid iron.

132—VOLUME 40B, APRIL 2009 METALLURGICAL AND MATERIALS TRANSACTIONS B

3. Y. Li, R. Matway, and R.J. Fruehan: DOE Report, United StatesDepartment of Energy, Washington, DC, 2004.

4. J.J. Kerr and R.J. Fruehan:Metall. Mater. Trans. B, 2004, vol. 35B,pp. 643–50.

5. J.J. Kerr and R.J. Fruehan: 2000 Process Technology Conf. Proc.,ISS, Warrendale, PA, pp. 1049–63.

6. O. Fortini and R.J. Fruehan: Metall. Mater. Trans. B, 2005, vol.36B, pp. 865–72.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 40B, APRIL 2009—133