The Exact Art and Subtle Science of DC Smelting: PracticalPerspectives on the Hot Zone

ISABEL J. GELDENHUYS1,2

1.—Mintek: Pyrometallurgy Division, 200 Malibongwe Drive, Randburg 2194, South Africa.2.—e-mail: isabelg@mintek.co.za

Increasingly, sustainable smelting requires technology that can process met-allurgically complex, low-grade, ultra-fine and waste materials. It is likelythat more applications for direct current (DC) technology will inevitably followin the future as DC open-arc furnaces have some wonderful features thatfacilitate processing of a variety of materials in an open-arc open-bath con-figuration. A DC open-arc furnace allows for optimization and choice ofchemistry to benefit the process, rather than being constrained by the elec-trical or physical properties of the material. In a DC configuration, the poweris typically supplied by an open arc, providing relative independence and thusan extra degree of freedom. However, if the inherent features of the technologyare misunderstood, much of the potential may never be realised. It is thusimportant to take cognisance of the freedom an operator will have as a resultof the open arc and ensure that operating strategies are implemented. Thisextra degree of freedom hands an operator a very flexible tool, namely virtu-ally unlimited power. Successful open-arc smelting is about properly manag-ing the balance between power and feed, and practical perspectives on theimportance of power and feed balance are presented to highlight this aspect asthe foundation of proper open-arc furnace control.

INTRODUCTION

Direct current (DC) open-arc furnaces have somewonderful features. These furnaces are good atprocessing fine feed materials (because of the openbath) and are also very good at treating feedmaterials with complex compositions. The power isgenerally supplied by an open arc, providing rela-tive independence in respect of power input andthus an extra degree of freedom. Open-arc operationallows a choice of chemistry to benefit the process,rather than being constrained by the electrical orphysical properties of the materials (e.g., resistanceheating). The freedoms and features of an open arc,however, mean that operational strategies are crit-ical to success. Mintek’s involvement in develop-ment and testing of a variety of DC smeltingtechnologies is well established with more than30 years of experience operating pilot DC fur-naces.1–6 Through this work Mintek continues theevaluation of DC furnace technology for a variety ofdiverse metallurgical applications and has devel-oped an in-depth understanding of the features as

well as the subtleties of operating these furnaces. Inlight of this experience, some aspects of theseinsights are presented in the form of a practicaloverview, primarily reflecting on the intention ofthe open arc while highlighting the basic operatingprinciple that is required to adequately leverage thepower of lightning.

Enthusiastic pyrometallurgists are interested inthe thermodynamics and engineering of high-tem-perature processes; however, many fellow pyromet-allurgists will agree that some aspects of smeltingfall within the realm of the ‘‘subtle arts’’; some mayeven say aspects of smelting belong in the realm ofmagic. Quoting Rowling, the author of Harry Potterand the Philosopher’s Stone may seem whimsical,but the following extract reflects the essence andthe passion of pyrometallurgy and bath smelting,and the intention of this paper, quite well. In thequote, Professor Snape introduces young magiciansto potion-making, which in the opinion of theauthor may as well have been written as anintroduction to operating an open-arc, open-bathsmelter.

JOM

DOI: 10.1007/s11837-016-2171-z� 2016 The Minerals, Metals & Materials Society

‘‘You are here to learn the subtle science andexact art of potion-making. As there is little foolishwand-waving here, many of you will hardly believethis is magic. I don’t expect you will really under-stand the beauty of the softly simmering cauldronwith its shimmering fumes, the delicate power ofliquids that creep through human veins, bewitchingthe mind, ensnaring the senses…’’.7

For those of us ensnared by the beauty of thesimmering open-arc, open-bath cauldron, this paperaims to facilitate reflection on the nuances andprinciples of operating an open-arc DC smelter. Thepurpose is not to negate the need for science andengineering or good furnace design but to reflect onthe practical challenges often misunderstood orunderestimated. Unfortunately, no furnace design,whether it consists of the most advanced coolingsystems, or fancy instrumentation and control sys-tems, can overcome poor operating strategies. It isunfortunate that many operations do not under-stand the true nature and intention of DC open-arcsmelting and hopefully this paper will provide someinsight into the art and science of DC smelting.

THE ROAD LESS TRAVELLED

DC furnaces have been successfully implementedfor a variety of commercial applications. However,the technology is often still thought of as the newkid on the block, but perhaps the true story can bebest described as the road less travelled. Theoriginal DC furnace concept predates alternatingcurrent (AC) furnace technology evidenced by thework of Sir William Siemens.8 Siemens first used aDC arc furnace with a vertical graphite cathode tomelt material in contact with a water-cooled bottomanode in 1878. The first AC electric arc furnace(invented by Paul Heroult) was patented and firstoperated in 1900.6 Thus, technically, DC furnacetechnology is the older brother of AC furnaces.Electric furnace technology became almost entirelyAC-based because of the use of AC for efficientpower transmission from large central power sta-tions. DC implementation only really became viableonce low-cost high-power solid-state semiconductorrectifiers became available. Since the mid-1980s,the technology has been widely implemented forsteel-scrap melting and, in addition, metallurgicalprocesses like ferrochromium and ilmenitesmelting.

Descriptors like ‘‘unproven’’ or ‘‘high-risk’’ arestill used to describe DC technology. Contributingtowards this reputation is the fact that the some DCfurnace installations were marred by difficultiesduring start-up and experienced design challenges.Barnes et al.9 described a variety of factors thatcontributed towards the Chambishi DC furnace’soperational challenges. The Barnes paper refer-ences the well-known McNulty curves10 to evaluatethe Chambishi project in the context of the classi-fications developed by McNulty. The paper provides

some insight into managing risks, and, although weknow that no start-up will ever be perfect, thesuccess or failure of a project is all about mitigatingrisks. The four types of projects identified byMcNulty are briefly summarized as follows tohighlight the types of risks:

� Type 1: Mature technology, used elsewhere,scale similar to or smaller than prior applica-tions of the technology, and thorough pilottesting completed.

� Type 2: Prototype technology (early or firstimplementation), incomplete pilot testing, andsevere operating conditions (e.g., high tempera-tures), innovative parts of technology work, butauxiliary and support systems not tested ordesigned to suit.

� Type 3: As for Type 2. In addition, limitedpiloting was done and/or feed variability iscommon. Design flaws in simple systems, e.g.,feed systems may contribute, and often engi-neering was ‘‘fast-tracked’’.

� Type 4: As Types 2 and 3, but with more complexflowsheets. A lack of understanding of chem-istry, product quality or raw material character-istics is often an issue. Inadequate training ofstaff also adds to the difficulties and delays.

Flowsheet complexity, variability, design flaws insimple systems and fast-tracked projects are clearlysignificant contributors to delays, and all of thesecan contribute towards project failures. These char-acteristic mistakes are not unique to DC projects.New or novel applications are often selected specif-ically to address properties of the raw materials(ores are increasingly complex, lower in grade andnon-standard). Despite rational views and facts,new technology is also measured against a higherstandard, and failures reflect poorly on the reputa-tion of the technology, even if the root cause is notdirectly related to the technology. In order tominimize risk, it is critical that operators under-stand the technology they are implementing. Train-ing of new plant teams is therefore critical, yet, evenwith the best of intentions, it remains a tremendouschallenge to get new teams up to speed. Although itis of course extremely important to fully appreciateand understand new technology (through piloting),even if the technology is not really that new, it isusually new to the plant teams responsible forcommissioning and eventually operating thefurnaces.

Mintek’s test facilities have indeed been usedquite successfully to demonstrate DC smelting formany applications. The smelting step is often testedthoroughly but usually in isolation and quite earlyin the project development phase. Upstream anddownstream integration is seldom demonstrated atpilot scale. This can result in poor or inadequatedesigns for simple auxiliary systems like feed or off-gas systems. However, smelting is more than just

Geldenhuys

the technology and the equipment, and furnaces aredesigned successfully all the time without inte-grated testing. Equipment is, however, operated bypeople, and operating an open-arc furnace requiresa different operating approach often not intuitive tooperators. Unfortunately, while Mintek’s metallur-gists and operators continue to gain practical expe-rience in the exact art of operating DC furnaces,knowledge transfer to new plant operators has beenmuch harder to achieve. Some of the most successfulimplementations of DC technology committed earlyfunding to send their new plant teams to Mintek tooperate a DC furnace for an extended period, priorto commissioning the industrial furnace. It is not,however, always practical or affordable, but it is agood example of addressing a high-risk item headon.

A BRIEF HISTORY OF DC SMELTINGIMPLEMENTATION

Ferrochromium alloy smelting in a DC open-arcfurnace for direct processing of ore fines has come ofage over the past 30 odd years.11 South Africa’schromite resources are well known for their friablenature, and during the early 1980s, Mintek andMiddelburg Steel and Alloys (now incorporated inSamancor) jointly developed the DC open-arc pro-cess for the production of ferrochromium with theobjective of exclusively smelting chromite ore fines.DC application to ilmenite smelting followed shortlyafter, via the first installation of DC smelting forAnglo American at Namakwa Sands (now incorpo-rated in Tronox).6,12 Both DC operations initiallyendured the pain of ‘‘being first’’, but are stillsuccessfully operating their furnaces.

Early adopters of technology often feel the needfor secrecy, wishing to maximize the benefit fromlessons learnt for themselves. This is also true forDC smelting. A glaring lack of data and publicationsin the public domain and on DC operations speaksto this point, further complicating the matter ofknowledge transfer among users. The art of DCsmelting is rarely captured in publications. The DC‘‘user group’’ is still a relatively small community iftruth be told, and the tendency to protect know-howand experience has thus resulted in the dissemina-tion of information more in the form of operationallore than through scientific publications. As DCtechnology matures, more data will hopefully bepublished and perhaps DC-specific sessions at majorconferences may become a reality in the near future.

Thus far, the story of DC is a story of paradox.Some operations are extremely successful and con-tinue to expand and thrive, while others havestruggled and even failed. Although proper furnacedesign and addressing the risks associated withproject complexity contribute to successful imple-mentation, there is something else at play, namelyhow quickly an operation learns to understand theconstraints and the power of the open arc.

THE MORE WE CHANGE THE MORE THINGSSTAY THE SAME

The intention of a reductive smelting process is ofcourse to separate the valuable metals from thegangue by selective reduction. If the desired balanceis maintained and managed, the open-bath open-arcDC operation achieves phenomenal results (recoveryand throughput). DC smelters are generally intendedto be high-intensity units. Just because one canachieve high throughput, this does not mean that thisis a free ride. An interesting feature of the earlyadopters of DC technology is that the intendedproducts and metallurgy were actually quite wellknown. From a metallurgical perspective, there is nocause to believe that smelting in a different type offurnace should significantly alter the process. Fer-rochromium production in a DC furnace focusedprimarily on treating fines directly, and the subse-quent metallurgical benefits were not a direct objec-tive. The benefits associated with DC ferrochromiumsmelting are as a result of the freedom to adjust theslag composition to achieve metallurgical objectives,and in this case the DC furnace is a metallurgicalenabler, allowing yield or grade optimisation unpar-alleled in the ferrochromium industry. Despite theperceived differences, the basic metallurgy is notdifferent (the properties of the well-mixed open-bathenables the operations to operate closer to equilib-rium). Ilmenite smelting is another good examplewhere one could argue that DC furnaces are not thatmuch different from the original technology (as usedby Richards Bay Minerals in South Africa, forexample). Richard Bay Minerals uses the processtechnology originally developed by Quebec Iron &Titanium (QIT), namely rectangular six-in-line gra-phite-electrode furnaces in open-bath mode with ACopen-arc operation.6 High-titania slag, the primaryproduct from ilmenite smelting, is highly conductive,and there is no alternative but to operate with anopen arc, regardless of the type of furnace technology.

The two examples intend to illustrate that,despite application of a new furnace technology,the principal chemistry and process parameterswere actually not that new. However, at implemen-tation, the operations needed to learn how to runthese furnaces to achieve the benefits associatedwith this technology or, with a more negative slant,to avoid destroying their furnaces.

Initially, DC technology was strongly associatedwith feeding via hollow electrodes as it was thebelief that the magic of the hot arc is only accessibleif the feed is fed directly into the plasma arc.Generally, feed arrangements for DC furnaces haveshifted towards maximizing throughput and reduc-ing the cost of electrodes, and Mintek rarely oper-ates or recommends a hollow electrode feedarrangement. Feed is still, however, fed into thefurnace engine room, the hot spot or arc attachmentzone where the arc supplies the power. In order tomaximize the benefits of the open arc, feed should

The Exact Art and Subtle Science of DC Smelting

be presented to the hot zone, although some feedcan be accommodated towards the sidewalls toassist with shielding the roof from radiation fromthe bath (if operating with an open-bath, themajority of radiation is from the hot surface of theslag). Arc stability can be negatively impacted iffeed rates are not well controlled, however, as slug(intermittent, variable feeding) feeding into an arcor even the hot zone is obviously undesirable. Theaim should always be to provide feed to the hot zonein a controlled manner, as an unstable arc canquickly result in poor outcomes.

SO YOU HAVE AN OPEN ARC?

The flipside of the flexible independence providedby the open arc requires that tight control of themass feed rate is required in order to balance outthe very stable, high-intensity power input attain-able via the DC open arc. Best practice conversa-tions for operating a DC furnace starts with oneword, namely balance, which is easier said thandone. Electrical input (arc stability and power) isusually very accurate and controlled, while feedinga furnace as accurately is not as easy nor simple.The arc attachment zone and the surrounding hotzone is where most of the magic happens in an open-arc furnace. If one considers the sheer quantity ofenergy generated by the arc (in a single hot spot,centrally located in the furnace), the importance ofproviding the hot zone with stable, continuous freshfeed seems obvious. The hot zone can be both ablessing and a curse. The key to sustainable smelt-ing is to manage the power-to-feed balance in theengine room (the hot zone). This key unlocks thebenefits of a DC furnace, yet it is often ignored orunderestimated. Unfortunately, the lesson is thenlearnt the hard way, either through damage toequipment as a result of a furnace failure or poorproduction outputs. Optimization and refinementsare part of any process, but it is difficult to improvea process if the basic principle is not in place. Thepower-feed balance should always be at the top ofthe agenda, regardless of the maturity of an oper-ation. Operating a DC furnace without a soundstrategy is ‘‘furnace suicide’’. With great powercomes great responsibility. The phenomenal abilityof a DC open-arc furnace to provide power to the hotzone comes with a great responsibility to feed thefurnace properly, in order to match or balance thepower. If the power-to-feed balance of this high-intensity machine is neglected, trouble will follow.

‘‘IT’S COMPLICATED’’

All operations have, of course, a range of vari-ability, either naturally (from the earth) or due toengineering or control limitations. ‘‘Tight control’’ isrelative, of course. In order to get the best controlpossible, it is thus important to understand theprocess variability, equipment limitations and themetallurgical boundaries very well and this should

be incorporated into the control strategy. Once theseare quantified, the impact of equipment constraintsand variability on the longevity of the furnaceshould be quantified. Variability always comes witha price tag. It could be compromised temperaturecontrol or poor product quality, or low availability orrestricted throughput, or all of the above. If, due tofeed composition variability, the furnace needs to beoperated at a higher temperature than suited, itmay, for example, require more frequent tap-holereplacements. Due to the nature of high-tempera-ture processes, the cost of process input variabilityis often throughput or integrity issues. Many riskscan be mitigated, especially if the operators fullyappreciate the power of the open arc and if propercontrols are put in place to manage the impact of thevariability. Many high-temperature processes areoperated within fairly narrow metallurgical bound-aries, primarily determined by the nature of thematerial being processed, or the product gradebeing targeted, or even simple economics. A DCsmelter can be a powerful tool that can overcomesome typical constraints (like direct processing offines), but it is obviously not really a magicalcreature that can be allowed to roam free. DC arcfurnaces are by no means a solution for all metal-lurgical problems, and the technology is not able toovercome poor control, unreliable feed systems orcomplete lack of feed composition control.

High-temperature processes are by their verynature unforgiving, but while a DC furnace is a verystrong, well-defined muscle, we (the operator/plantmetallurgist/furnace designer) need to provide thebrains to curtail the brawn.

THE DARK SIDE OF THE FORCE

The multi-phase multi-component systemsinvolved in smelting processes have so many vari-ables that it is often difficult to pick out the mostimportant relationships, or even to identify thecause or causes of a ‘‘disturbance in the Force’’ asthe Jedi Master Obi-Wan Kenobi eloquently andfamously said in Star Wars.13 In keeping withGeorge Lucas’ Star Wars analogy, the Force isstrong with DC furnaces. However, the Force needsto be in balance (the Dark and the Light). It is thejob of the operator to bring balance and to achievethe desired outcomes. Imbalance will result inunwanted consequences, and either too much poweror too much feed is of course undesirable. In order toachieve balance, both sides of the Force arerequired, neither can exist without the other. Themythology of the Star Wars story thus providessome sage and practical guidance as well as acautionary tale, namely to heed the temptation ofthe Dark Side (power). Operators of furnaces areunder tremendous pressure to meet throughputtargets, and, in the short term, it is easy to succumbto the Dark Side (operating with excess power). It isusually easier to operate a smelting operation at a

Geldenhuys

slightly higher temperature than design. Tappingdifficulties are generally overcome by increasing thetemperature; ‘‘the show must go on’’ principle.Seldom will you find a plant where the operatorsrun the furnace slightly cold by choice. Systematicexcess energy input will, of course, have long-termnegative impacts on refractory and tap-hole life, yetin the short term the temptation to use this powerremains within the operator’s reach.

A DC furnace is just a very (very) large weldingmachine. If you don’t add the flux and move thewelding arc accordingly, you will burn a hole andruin your work. This balancing principle (feedingthe hot zone) is true no matter what scale of openarc you have. Regardless of whether it is an arcwelder or a mega-scale DC furnace, you mustbalance the hot zone with the ‘‘cold’’ feed. The hotzone is where most of the balancing should happenor you are going to find yourself in a lot of trouble.

POWER-TO-FEED BALANCE

The specific energy requirement (SER) of a smelt-ing process can be simply expressed as MWh permetric ton of total feed or power (MW)/feed rate(ton/h). SER is the energy required to transform thefeed materials at 25 �C into the product streams atthe desired temperatures at which they leave thefurnace. SER is thus inherently the power-to-feedratio we strive to achieve, and regardless of how aplant opts to express this ratio, it remains thefundamental starting point. Operators sometimesforget that the theoretical SER changes quitesignificantly if the chemical composition or temper-ature of the raw materials deviate from the theo-retical baseline. A controlled feed and energy ratiois critical for reductive smelting processes, as theproduct quality relies on achieving the desireddegree of reduction and separation at the targettemperature. Variances in the feed recipe impactproduct quality and the energy balance (often both).Acknowledging the impact of feed composition vari-ability is important, yet we cannot compensate forall variances, some of which are due to natural feedvariability, while others may be due to poor feedcontrol or equipment deficiencies. Again, as long aswe understand the impact of compositional vari-ances and provide the operators with guidelines todeal with these, the power-to-feed balance can bemanaged.

DIFFERENT STROKES FOR DIFFERENTFOLKS

Ilmenite smelting is a good example of a smeltingprocess where poor carbon control can lead to bothproduct quality and operability issues. The carbonbalance is fundamentally built into the SER.Ilmenite concentrates are generally fairly homoge-nous (less natural variance in the feed to deal withthan other ores). However, as a result of poorreduction (too little carbon), not only is the product

quality compromised but excess energy (not con-sumed to reduce iron oxide) can increase the slagtemperature. The slag, with increased concentra-tion of iron oxide, is naturally more fluid, and anexcursion with poor carbon control can easily resultin hotter, more aggressive slag. Frequent poorcarbon control can compromise the slag freeze-lining. On the other hand, excess carbon and/orpoor temperature control can cause slag foaming.Although intentional foaming practices are usedquite widely in industry, an uncontrolled foamingevent is obviously undesirable. Slag cleaning pro-cesses are generally not as sensitive to variability inthe carbon addition, mainly because the primaryproduct, the metal phase, is a small proportion ofthe total feed. The degree of reduction is then notquite the ‘‘the knife’s edge’’ balancing act of ilmenitesmelting. Slag-intensive furnaces are basically aslag rejection mechanism, and the bulk of theenergy input is used to increase the feed to thedesired operating temperature to separate the valu-able metals from the gangue. A relatively smallportion of the total energy input is related to thereduction reaction. Slag-intensive processes arethus more like balancing on a gymnastic high-beam.Note, however, that it remains a balance betweenpower and feed, and the sensitivity to variabilitydepends on the process.

NAVIGATION AND FURNACE CONTROL

Carbon control and fluxing variability obviouslyimpact slag composition, slag properties and theenergy balance and cannot be ignored when manag-ing the power and feed balance. Assuming, however,that the process is well understood and the impactof feed composition is taken into account in thepower-to-feed balance, managing the feed-to-powerinput ratio can be quite well described throughnavigational principles. An operator requires infor-mation to manage a furnace, much like a pilot needsflight instruments to navigate a plane from onepoint to the other. The sophistication of flightinstruments vary greatly, with some advancedaeroplanes virtually flying without the need for apilot. Yet, generally, planes have some instrumen-tation to indicate direction and altitude. In the sameway, mass fed and power information is needed tooperate a furnace, and this is especially true for anopen-arc furnace. A basic mass balance is, of course,also important—what goes in must come out—butin order to at least have a chance of success, it isimportant to start with the basics, namely input.Successful operations invest resources to improvemeasurement and/or estimation of input and outputdata. If the quality of the information improves, sodoes the control efficacy.

Furnace navigation starts with attempting toaccurately balance raw material feed rates withfurnace power input, or at least continuously aimingto improve or correct this ratio. The feed into the

The Exact Art and Subtle Science of DC Smelting

furnace is generally controlled from hoppers orstorage bins integrated with a mass loss system tomonitor and adjust the feed rate and feed ratios tothe required levels. Mintek uses continuous, accu-mulative calculation and manages the power-to-feedbalance ratio throughout any given feed period(expressed as how closely the actual power-to-feedratio is relative to the desired process ratio, i.e. netenergy in/mass fed). While the aim is to be perfect atall times, this is obviously never true. This practicecan be described via navigational principles appliedby pilots. The basic control philosophy described isapplied by pilots and furnace operators alike:implement corrective action by evaluating progressagainst target (Figs. 1 and 2).

Figure 3 graphically depicts navigational out-comes via four simplistic scenarios. The pilot ineach case aims for the mountain, while the dia-grams illustrate the outcomes of four scenarios. Inscenarios (a) and (b) no action is taken, while (c) and(d) illustrate outcomes due to intervention.

Scenario (a) shows a ‘‘perfect’’ outcome. It is awonderful day with no wind, and the pilot can justpoint the aeroplane in the direction required. This isthe only situation in which heading and course staythe same throughout the flight. Our furnace oper-ator’s job is to aim the furnace (target feed rates,power input), and all works out perfectly and everymeasurement is accurate and no interventions arerequired, but this is not a likely scenario.

Scenario (b) shows what would happen if aconstant crosswind impacts the aeroplane after thepilot set the heading and took no further action(until the aeroplane runs out of fuel). The aeroplanevery definitely will never arrive at the destination.If left unchecked, the furnace will not achieve thedesired outcomes either and there is always acrosswind when operating a furnace.

Scenario (c) illustrates a case where the pilotadjusts the direction regularly (comparing theactual progress relative to the destination) and overtime a change in heading occurs. The aeroplane isnot actually flying a ‘‘straight course’’, but rather afunny curved path. At least the aeroplane willarrive at the desired destination, even if not com-pletely efficiently. If drift occurs, the furnace

heading may need adjustment; however, as withthe flight path illustrated in the diagram, theultimate efficiency may not be ideal; this is quite agood approximation of the majority of furnacecontrol. A furnace has a memory, however, as itremembers what you did to it (the good, the bad andthe ugly). If the influence of a non-ideal parameteris not proactively managed, the path travelled mayhave some consequences that needs managing. Touse the flight analogy, the aeroplane will need morefuel to reach the destination if the response to driftis not managed regularly or timeously. Although notperfect, this type of corrective control is common,especially if the information (instrumentation) isless sophisticated or delayed. Commonly, we receiveslag analysis and temperatures retrospectively, i.e.we’ve already drifted due to a crosswind and thusrequires a heading adjustment as per this example.

Scenario (d) is an outcome of an elegant compro-mise between heading and course. It requiresaccurate data for the direction and speed of thecrosswind to achieve. Although more ideal, it is notthe most likely manner of furnace control as it

Fig. 1. High-speed photographs of DC arcs: the power of lightning(photographs by Reynolds, Mintek, 24 May 2012).

Fig. 2. Tapping the cauldron (pilot furnace) at Mintek (photograph byGeldenhuys, Mintek, 16 April 2004).

Fig. 3. Navigation (a) with no cross wind, (b, c) with crosswind.Reprinted with permission from Ref. 14.

Geldenhuys

requires accurate, up-to-date data not readily avail-able. We should, however, always strive to improveour ‘‘flight’’ instruments towards this ideal.

Navigating a furnace with advanced instrumen-tation and accurate, timely information is of coursethe ideal. At the rate technology is changing,perhaps an ‘‘autopilot’’ furnace may eventuallybecome a reality. However, until science catchesup with fiction, we have to rely on the best controlsystem available to us, namely experiencedoperators.

DOES IT MATTER HOW YOU GOT THERE?

Generally, an operator should only implementminor tweaks (periodically) to manage drift to ensurethat, at the end of a discrete period, the desireddestination is reached. These discrete periods and toa degree the objective destination are optimized foreach plant based on the information available to theoperator. The degree of manual intervention isdetermined by the sophistication and accuracy ofinstrumentation, data, equipment and control sys-tems. However, no control system can compensate forpoor information or inadequate equipment. An expe-rienced operator can intervene manually to overcomemajor deviations, as long as the principle of the powerand feed balance is well understood. Too often, anoperator realises shortly before a tap that the carbonor power input ratio for the past few hours has driftedout of range and then tries to correct the error in ashort, intense burst just before the slag tap. Underthe right conditions, this type of intervention cancause severe process instability and will not ade-quately address the imbalance caused by the drift. Ifevents beyond the operator’s control cause a signif-icant deviation from the plan (i.e. major down time ormajor failures), adjustments or interventions mayneed to be more radical (e.g., either slower feed ortotal feed stoppages). It may be prudent to land,refuel and re-calculate a new course if completely offtrack. A start-up strategy and procedure should beprovided to operators to catch up if the furnaceexperiences a significant down time; consistency iskey.

When feed control is fairly accurate (and compo-sition and temperature are relatively consistent),frequent adjustments may not be required and thepath may be close to the ideal. However, in the realworld, input parameters vary all the time, andcontrolling a furnace may become quite a challengeif the frequency (and complexity) of adjustmentsincrease.

It is hopefully clear that an aero plane and afurnace both require some basic instrumentation.Although our objective is to reach a destination asefficiently as possible, this is not likely (ever).Operations obviously strive for perfection (or atleast systematic improvement), but all furnaces areoperated in the real world, with real-world limita-tions. Furnaces also have excellent memories; it is

not just about the destination. The impact ofexcursions tend to add up if left unattended. Minorvariances in tapping temperatures and productcomposition are expected and are usually due tothe natural uncertainties in our system. Frequent,large adjustments to major operating targets, espe-cially reactively, can easily lead to furtherunwanted complications. A hot slag tap should betaken seriously, of course, but dumping feed into thefurnace after the slag was tapped will not addressthe problem and most likely introduce more insta-bility (much like trying to fix the carbon balance justbefore a tap; it is too late to make a simple masscorrection). A significant portion of the memory inthe example (the slag), has ‘‘left the building’’. Howwe react to an outlier should be informed by the‘‘what went wrong’’ question, as this determines theintervention required. Compensating accurately forhistory is not trivial, and operators unfortunatelyrespond to some variances with a hammer whenperhaps a scalpel is more appropriate.

In order to arrive at the desired destination,accurate information from the feed system andpower supply is required. Although this may appearobvious, it is alarming how often this information isnot available, or poor quality data are used. In theabsence of accurate information, the operator isdriving a car at high speed with no brakes and onlymanages to stay on the road through very dramaticinterventions, rather than controlling the vehicle atthe appropriate speed limit (SER) for the roadconditions. Power input to a furnace is normally oneof the most accurate inputs to the mass and energybalance. Rate of energy loss is probably the nextmost reliable number, especially for well-instru-mented, water-cooled furnaces, and probably onlymarginally less accurate than the power input if theinstrumentation is suitable for use. The trick isusually to determine an accurate mass fed (and feedcomposition), accurate power input and an accurateenergy balance to allow for optimization.

CONCLUSION

Optimising a smelting operation is a complex,multi-faceted adventure. If the very stable, high-intensity power input that can be attained in a DCfurnace is matched by equally stable and controlledfeed to the hot zone, the furnace can achievephenomenal efficiencies. Unfortunately, feed sys-tems are often not able to match the accuracy ofpower input, often leading to compromises and evenfailures. Sadly, a Rolls Royce furnace is oftenmatched with a feed system resembling a used VWbug from the 1960s. It is important to address feedinput as one of the primary risks when smelting inopen-arc mode.

A summary of good practices or principles for DCsmelting is a daunting task, and while topics likethermal efficiency, arc length management, massbalances, metallurgical control and many others

The Exact Art and Subtle Science of DC Smelting

should ideally be addressed as sub-topics of goodpractices, optimizing should start with ensuringthat the operator of an open-arc furnace under-stands how to manage the powerful hot zone. Thehope is that this paper will provide food for thoughtand remind furnace builders and operators of thefundamental principle entrenched in the open-arcoperation. A DC furnace often allows for processoptimization beyond the norm, but when things arenot going to plan make sure the power and feed isbalanced or else nothing else will matter.

Mentors would often start a young pyrometallur-gists’ induction into the world of smelting with thefollowing sage advice: ‘‘Take care of the slag and themetal will take care of itself’’. Although valid advice,for a DC furnace one might perhaps rather shapethe new generation through the following modifiedversion: ‘‘If you take care of the hot spot, the slagand metal will take care of themselves.’’

ACKNOWLEDGEMENTS

This paper is published by permission of Mintek.The author wishes to thank the many enthusiasticpyrometallurgist in the Pyrometallurgy Division atMintek for their influence and mentorship over thepast two decades and may the DC Force be with you.

REFERENCES

1. N.A. Barcza, T.R. Curr, and R.T. Jones, Pure Appl. Chem.62, 1761 (1990).

2. I.J. Geldenhuys and R.T. Jones, in Proceedings of COMPyrometallurgy of Nickel and Cobalt 2009, ed. by J. Liu(CIM, Montreal, 2009), p. 415.

3. I.J. Geldenhuys and R.T. Jones, in Proceedings of the SixthSouthern African Base Metals Conference, ed. by P.A.P.Fouche (SAIMM, Johannesburg, 2011), p. 477.

4. R.T. Jones, Q.G. Reynolds, T.R. Curr, and D. Sager, in Pro-ceedingsofSouthernAfricanPyrometallurgy2011, ed. byR.T.Jones and P. den Hoed (SAIMM, Johannesburg, 2011), p. 15.

5. R.T. Jones, in Celebrating the Megascale: Proceedings of theExtraction and Processing Division Symposium onPyrometallurgy in Honor of David G.C. Robertson, ed. byP.J. Mackey, E.J. Grimsey, R.T. Jones and G.A. Brooks(Wiley, San Diego, 2014), p. 129.

6. R.T. Jones and T.R. Curr, in Proceedings of SouthernAfrican Pyrometallurgy 2006, ed. by R.T. Jones (SAIMM,Johannesburg, 2006), p. 127.

7. J.K. Rowling, Harry Potter and the Philosopher’s Stone(London: Bloomsbury Pub, 1997), p. 102.

8. W. Siemens, English patents, No. 4208 of 1878 and No.2110 of 1879.

9. A.R. Barnes and R.T. Jones, in Proceedings of COM NewTechnology Implementation in Metallurgical ProcessesSymposium, ed. by CIM (CIM, Montreal, 2011), p. 111.

10. T.P. McNulty, Min. Eng. 50, 50 (1998).11. I.J. Geldenhuys, in Proceedings of the Thirteenth Interna-

tional Ferroalloys Congress, ed. by M. Tolymbekov (P.Dipner, Karaganda, 2013), p. 31.

12. M. Gous, in Proceedings of Southern African Pyrometal-lurgy 2006, ed. by R.T. Jones (SAIMM, Johannesburg,2006), p. 189.

13. Star Wars: The Complete Saga (Episodes I–VI), Dir. GeorgeLucas, Blu-Ray, 2011.

14. Jorg Emmerich, Flight Gear Flight Simulator (FGFS)Online Handbook, http://www.emmerich-j.de/Handbuch/EN/B7_RNAV.html. Accessed 22 July 2016.

Geldenhuys