Recent Developments in Carbon Baking Technology

J. Hurlen and T. Naterstad

As the aluminum industry strives to im­prove productivity and efficiency, carbon baking technology is one area receiving at­tention. With theanode pumtaccounting/or Q significant portion of total smelter invest­ment, completely new installations are often out of the question. But the alternative of retrofitting existing carbon bakingfurnaces can often provide incremental capacity, re­duce fuel consumption, and cut unwanted emissilms.

INTRODUCTION

Anode carbon manufacturing is usu­ally an integrated part of the primary aluminum industry. Thus,developments in carbon technology reflect both the specific and general requirements set by this industry.

•

b

Whereas quality deterioration of raw materials and increased fuel costs guided the attention in carbon anode develop­ment in the early 1970s,l .... subsequent trends of development have largely con­verged toward total cost-effectiveness, combined with the increasing emphasis on improved anode quality and manu­facturingenvironment. This general shift in priorities can be attributed to the fol­lowing factors: • The severe recession in the primary

aluminum marketduring 1982-1983 strongly sharpened the focus on investment and operation costs.

• Advanced computer systems for design, dynamic modeling, and on­line process control have created incentives for improved design and operation.

• Superiornewcelldesignsoperating at elevated performance levels ex­ceeding a current efficiency of 95% and energy requirementsofless than 13 kWh/kg aluminum have dearly identified anode shifts and irregu­lar anode perfonnance as primary causes of yield losses.

• Environmentalawarenessandsafe­ty consciousness, imposing in­creasingly stricter standards, have becomeprimaryqualityfactorswith which current and future expecta­tions must comply.

Carbon Economy Over the course of the last ten years,

anode consumption figures have been

PRODUCT

reduced 5-15% by several companies. The partial impact of anode cost on the total aluminum production costdepends on numerous factors and varies consid­erably between different plants. How­ever, the cost distributions shown in Figure 1 are typical. The distribution of the direct carbon anode costs (Figures 1 b and Ie) reflects the dominance of raw materials and baking costs.

Dependingondesignfactors,economy of scale, infrastructure, and location, the cost of a complete anode manufacturing plant amounts to about 12-15% of the total smelter investment.

ANODE QUALITY REQUIREMENTS

The differences between high-perfor­mance and low-performance anodes have been brought out by the combined demand ofincreasing operation load and improving perfonnance regularity.S.)) Typical indicators of inferior anode quality are: • Irregularanodewearrelated to lack

of overall consistency of anodes. • Excessive anode consumption rate

related to excessive oxidation. • Deterioration of reduction celI per­

fonnanceduetocarbondustcaused by inhomogenous anode oxidation.

• Anode cracking during the initial heat-up period because of poor thennal-shock resistance.

To date, routine anode-quality mea­surements for the above-mentioned properties have been based on different

• ConIlnUOUI Of blrtch

, Figure 1. (a) The direct cost distribution for primary aluminum production. (b) The direct cost distribution 01 green anode production. (e) The direct coslclistribution of bakedanodes.

20

ANODE PASTES

Figure 2. A flow sheet for anode carbon processing.

JOM • November 1991

company standards, due to very few adequate international standards.

In practical terms, the product consis­tency may be quantified by twice the standard deviation (2s) of any property recorded, assuming random sampling of ten anodes out of a minimum lot of 100 forstalisticai reasons. Depending on raw-material supplies and the anode properties considered, Iong-Ienn con­sistencies (20 ) better than 5-20% may be d ifficult to fulfill .

Although the basic anode perform'lncc level, measured as metal produced per tOMe 01 anodes, is largely defined by the cell design and operation strategy em­ployed, the achievable cell control preci­sion depends significantly on a consis­tent anode quality.

The sensitivity of the anode quality to target precisionsdiffers from step to step in the semicontinuous process of anode manufacturing (Figure 2). Predominant differences of anode perfonnances are, however, most frequently attributable tooffsets on a few steps, including anode baking.

C."BON BAKING

Anode baking is a complex physico­chemical process combining coking and calcining in one operation.IU ) The basic challenges in quality control are • Maintaining an inert abnosphere

around the anode blocks to avoid oxidation.

• Controlling the heating rate to pro­vide optimal conditions for maxi­mum binder coke yield and strength with minimum thermomechanical tensions in the block matrix.

• Controlling the maximum tem­perature and soaking time to balance the kinetic crystal growth of all carbon phases present and to pro­vide acceptable iso-reactivity at the lowest possible level.

The acceptable heating rale depends mainly on block size, anode properties set by the preceding processing steps, and the actual temperature range con­sidered. General property changes with temperature and sensitivity with regard to block defonnation, cracking. and ho­mogeneity are summarized in Table I.

Baking Furnace Designs

Current large-scale anode baking fur­nace designs are all of the so-called ring furnace type, which essentially are high­temperature heat exchangers applying indirect heating by combustion gases from fired fossil fuel . The formed anode blocks are located in four to eight pits in each heat-exchanger section, with granular coke protecting against oxida­tion by flue gases and providing heat transfer from the flue walls.

Typically, a complete baking furnace includes 28 to 48 sections in series lo­cated in two parallel rows with cross-

1991 November - JOM

Table I. Physk:ochemlcal Effects 01 Heat Treating Anode carbon Blocks

1S<>-350

350-450

600-900

900-1,200

Physkochtminl ChIns" Thermal expansion of pitch. Release of tensions caused by forming /cooling.

Redistribution of pitch into voids by pitch expansion. Post-imp~nation of the aggregate.

Release of light binder volatiles.

Coking • Transi tion from plastic to

solid matrix. • Release of the major nol"lCOking

volatiles.

Post-Coking Release of heavy cracked volatiles.

• Annealing of tensions.

Crystalline reorientation and growth of the binder coke as well as of the lawest calcined aggregate coke.

M~jor Practical Aspects Reduced densi ty. Slight release of aggregate interlocking.

Risk of stud hole slumping/ deformations. Permeability, mechanical strength, resistivity affected.

Slight reduction in density of the aggregate packing.

Dilalometric tensions by thermal gradients causing expansion and contraction within the same block.

No particular effect within ordinary heating rates.

Dilatometrk tensions by contraction. Macroscopic cracks may arise if the previous avaagt! t""<llci ning level is largely exceeded (> l 000c) for the coke.

Table II. Baking Fumace Technology Evaluation Criteria

Process Character Investment Analysis

Control Puamdtr Annual capacity Scrubber requirement (flue gas and scrubber inlet / outlet) Specific investment cost

Product Quality Heat treatment level and homogeneity (equivalent temp. avg. and std. dev.l

Heating rate 3OO-6OO"C « Il"C/ h) Operation Costs

Thennal Energy Fuel injected (gas/oil) Coke consumption Tar volatiles burned

Maintenance ,.,bo, Refractory replacement (min. six years avg.) Operators and bricklayers

over connections at each end. Fans lo­cated at the outlet scrubber end provide a d raft through a joint exhaust pipe sur­rounding the furnace. Ambient air is thus pulled through sections on cooling, to recover heat before entering the peak­fired and preheating sections. Exhaust manifolds connect each firing zone to the joint exhaust pipe.

On completion of the heat treatment, one section is disconnected at the cool­ingend whilea new section is connected at the preheat end by moving the ex­haust manifold to the nexl section in front.In this way, the firing zone is moved around the furnace ring.

Although the principle of all ring fur­naces is the same, the detailed design and operation featu res may be quite different.1 .... 1· Two distinctly d ifferent ring furnace concepts have dominated the aluminum industry for more than 30 years-the so-called horizontal-flue (HF), or open-top, furnace, and the ver­tical-flue (VF), or dosed-top, furnace. The essential difference is related to double (HF) or single (VF) flue wall ron­structions, flue gas passages and firing practices. In Figures 3 and 4, schematic drawings of the two different designs are compared. Mathematical and physi-

cal modeling have been applied over the last ten years to make the various designs more efficient.11-lS

BAKING FURNACE EVALUATION CRITERIA

The available documenrtation of the technology is of primary importance in distinguishing between poor operation routines and the shortcomings of the technology.

Feasibility Analysis

Although precise cost estimates do require detailed knowledge of local con­ditions, first-hand comparative evalua­tion of baking technologies may be car­ried outon a general basisITableIl}. The cost d istribution of a typical turn-key greenfield installation is shown in fig­ure 5. Investment cost differencesamong current technologies are essentially at­tributed to furnace cost, plant building size, and flue-gas demand.

Since the major baking furnace cost is determined by the refractory demand and thedosely related bricklaying time, the investment cost is, to a first ap­proximation, inversely proportional to the baking furnace capacity ratio (CR),

(Continued on page 24.)

21

(Omtinued from page 21.) which is the ratio of annual baking ca­pacity (in tonnes) to the tonnes of re­fractory installed. The plant building size required to cover the furnace length is also fairly closely related to the CR

Rue-gas piping, fans, and scrubber costs are, for a defined scrubber tech­nology, roughly proportional to the specific flue-gas demand. The relatively large specific flue-gas demand of be­tween 4,000-5.000 Nml/t baked anode, typical of the HF furnace, has in recent designs been successfully reduced to a level approaching 2,500-3,000 Nml/ t baked anode, which is typical of the most efficient and modem VF furnaces.

Figure 6 depicts improvements in closed baking furnaces during the last decade. The latest improvements indi­cated (HA) were based on the Hydro Aluminium design by replacing refrac­tory materials within existing concrete shells. Although published, up-to-date

infonnation on the CR of open HF fur­naces is scarce,15.19 indications are that CR values close to the current levels shown by the Hydro Aluminium baking furnace technology have been achieved by the most efficient designs.

Depending somewhat on the technol­ogy applied, economies of scale, and local conditions, the current inveshnent cost of baking plants amounts typically to $450-600 per tonne of annual baked capacity, or 7-10% of a greenfield smelter inveshnent.

Product Qualtty

In the multistep, continuous anode manufacturing process depicted in Fig­ure 2, the need for selective monitoring methods to screen the product quality impacts by each step. is apparent. Using techniques described in References 30 and 31, single-anode monitoring can be correlated to the calcining level (equiva­lent temperature) distribution, as well as

Figure 3. A schematic drawing of lhe VF-type furnace.

ElliE GAS Rt!!G \lAIN

Figure 4. A schematic drawing of the HF-type furnace.

24

to its influence on net anode consump­tion and anode dusting.

From anode perfonnance data gener­ally available, it is evident that modem baking furnaces are capable of produc­ing adequate anodes that comply with the requirements of modem cell tech­nology. However, inferior baking fur­nace designs and inadequate operation targets and control systems may still account for an excessive specific gross and net anode consumption of at least 5% at several plants.

Operating Costs

In addition to the significant impact on operating costs, environmental awareness has increasingly focused on the energy efficiency of baking furnaces. The total fuel saving of more than 20% over the last five to ten years of the most efficient designs is likely to be followed by a further reduction of 10% or more over the next five years. Current energy consumption levels for efficient baking fumacesrange from 1,315-1,585 kWh/ t baked anodes.

Majormaintenanceofbakingfumaces is related to required refractory re­placement due to thermal strains and chemical wear by aging. Hence, the spe­cific maintenance costs may be well esti­mated on the basis of refractory replace­ment per tonne of baked anodes pro­duced. A more precise comparison would include attributed labor as well as any production losses through the service life of the furnace.

Since short-term flue-wall replacement usually accounts for more than 50% of the total maintenance, the flue-wall life is frequently used as a maintenance cost

FlQure 5. The distribution of investmentcosts in a greenfield baking plant with a capacity of 100,000 tonnes per year . ..

H

H

~± .. =-----~, .. cc------," .. o-------,~ .. Figure 6. Baking furnace capacity ratio vs. year 01 furnace installation (HA refers to Hy­dro Aluminium baking furnace technology).

JOM • November 1991

Ruuelolll R~PlAeUIUI

kIIliFR.1T INOOn

.. ,

..

..

.. "

" .. 10 ,01 no .44

IllIllG' fl Ut Will CI Cl"

Figure 7. Average specific refractory replace· ment lIS. flue wall cycles.

indicator. However, as seen in Figure 7, the specific flue-wall refractory replace­ment versus flue-wall life may differ very much with the design, because of highly different weights of fl ue-wall constructions.

For the best current designs, the aver­age total lifetime (20-30 years) mainte­nance requirement is likely to approach 11.5 kg of refractory per tonne of baked anodes produced. Substantial reduced routine maintenance requirements over the last eight to ten years are mainly att ributable to modified fu rnacedesigns, h ighly upgraded re fractory quality standards, and improved firing control systems.

Overall labor requirements, including operation and short-term furnace main­tenance, at updated baking plants are typically in the range of 0.3-{).5 hours per tonne of anodes produced . Differ­ences due to the baking furnace tech­nology applied are minor.

Information related to the in-plant environment at p lants applying the d if­ferent technologies has not been fully evaluated. In preliminary comparisons, the combustion effi ciency of tar compo­nents released from the green anode binder has typically been 96-98% in VF­furnaces and significantly higher in HF­furnaces. With scrubber systems in place, the difference in external emis.c;iom; ap­pears to be minor.

APPLYING NEW BAKING TECHNOLOGIES

A comparison of the performance characteristicsof updated baking furnace technologies ind icates that the most significant d ifferences are attributed to investment, fuel, and maintenance costs, whiled ifferences in labor costs and envi­ronmental aspects appear to be minor.

The comparatively high baking plant investment (typically 7-10% of a total greenfield smelter investment), com­bined with the fact that anode baking accounts for some 6-8% of aluminum production costs, has increased the in­terest in new baking technologies in connect ion w ith futu re greenfield smelter p rojects. The application of new technologies, combining requirements for increased capacity with reduced fuel and maintenance costs, represcntsa very cost-effective retrofit alternative fo r several existing baking plants. Examples of retrofit concepts realized have dem­onstrated benefits such as 40-50% ca­pacity increases as well as a reduction in the fuel consumption of some 40-45% in existing closed baking furnaces.

Thus, a retrofit of old baking furnaces may be an alternative to new baking­plant investments in connection with smelter expansions, as the capacity in­crease may be obtained with an invest­ment of only 20-30%.

References

1. P.S. Rht<Icy.nd o. DuT,...,."bl.y, -Refinery Fwd.tod<s Cob S!ru<w", . nd Alumini um Cen Anod .. ," Ughl M ... I, 1971 (Wam:ndalp, PA:TMS, 1977), pp. XlI -316. 2.5.5. Jon .. , R.O. Hild"bT.ndl,.nd M.e. H~lund. -V. ,.,.· bon 0/ Anod" P~rforman'" willt C"k" Qualily.- TMS P'f'P' ... 1"",.", A71·'17I W • .....-nd.I • . I'A; TMS. 1977). 3. S.5. lo_. R.D.l lild.t>,.ndl.nd M.e.llodIUlld, - Inn""""" of High SulphurCok .. on Anod. Porlonn.n"",- Lighl M,la!, 1979(W.m:-nd,I •. PA: TMS. 1979). pp 553-574. 4. 1'.1. Rhtdcy , nd S.K. N.dkorn;. -Cakiner FHd.tock Chora<l<-ri<tics. nd C.ki...-dCokeQuolity.·· Ugh! M,!." 1984 (W.m.-nda~, PA: TMS. 1984). PI'. 85'4j68. S. E. Bamllon. "M..m.ni'm$ 0/ Corbon Ou,t Form.ti"n in Aluminium l'roduction Pot$." ISCUBA (1'171), P!'. 87-94. /). J.F. Roy 800'ro,Grrbo". 20(19621. pp. ~53ti. 7. R.E. Cut&hall .nd V.L Bullough. - lnAu.n", 01 Bakin\! T~mpo..",,,urp.nd " node HI<"CtS upon C' ,bon Sl<>ugloing. U;;hl M ... ', 1m (W • .......,d.le. I'A; TMS. 1985). PI'. 1039--1076 8. W.K. ~;oberand R. 1'""",houd. -o.t ... mining P~~~ "nod. Prup<-rtiM 10' "luminum Production:' J. M ... " (Nov~mbl'r 19671, pp. 43-<45. 9. W.K. Fi.)u,r<"l.L · In~rd~pend"""" s..~\V(..-n Anod.Net Con",mplion and C"II D.$ign.. 1'01 Oper.ting P.r. meter. • nd Anode I'tup.", ..... " U~h! M ... I, 1991 (W ..... nd.I., I'A: TMS.lll'J1J.pp.661-698. 10. N. Bird, 8 . . \kF.n • ...,y, .nd 8." . Sodler, ·'Some I'ucticat Con""l~~n~ 01 u ... Carboxy . nd Airburn He'Clion, 01 Anod~C •• bon •. - U!{hl M, loI. 1990 (Warrendo!o:. PA: TMS. 1990). pp. 467--47l. II. O.N. Itold"", e! . 1., -Evalu.hon 01 I"" Uniformity ot Baking in Horiwnt.lond V~rti<al R~~ Ring Furn" ... " lithl

REGISTER BY PHONE 1-800-966-4TMS

MR.! , 198J (Warrendale. PA: ThtS, 1\I8J), pp. 797-804 . 12. J.·C Tho".... , "Anode R ... ctivity Innum~o/ lilt, Baking 1'"",",,--Third Aust,.I •• ian Aluminium TKhnologyC<>u15O (1'11';9), Pl'. ,4(1-2,1 . 13. T. F""""", .nd T. N.le .. ~, "C. rbon: 8as-ics and Pn,.·

t· I .. ," U"d .... !.nd;"~ '''' It.U·H"""U p"",tsS lot Produclion

Alumi",um, ed. 1(. Grptt...im . nd II. Kvand~ (Du,; ... ldorl. rmony: " luminium.V .... I.g.. 198M. PI'. 62-103

14. K.H. Fi.heT, -Oiff"'e~ in II ... Baking ~.nd It. Economy " I Opel"'~I.>n .. Rel.,~ II.>' CI.- and .n Op<n Ring Pit $v.,,,,,," (paper p~lod 01 the loond TM5--AIME Annu. 1 M..-l ing.Ch,caSo. Fo-b. 2S- M~",h I, 1973). 15. H. Onder e! 01.. "Horiumlal Flue Versus V.,,-ti.,.1 Flu~ Baking Fuma~- (popel" pt"eS<"I\l<:d at the 105lh TMS-A IME Annual Meeting. 1.;00 Voga • • NV. k-b. 22-26. 1976) . 16. O. I.""b&:n "'at" A New Ring Fuma",Con<ept: Dosign .nd Ope-t.tion,- Ughl ~I." 1987 (W.r~n<t.I •. PA: TMS. 19/17). pp. 497-503 . 17. A. Furm.n . nd H. M."i"",","A M.lht"",ti",1 MOOf l Simu loti ng on ArlO<k 110"" Fuma~: Ugh! Mrl.ls 1980 (W • .......-.d.I". PA:TMS. 19801. pp. 753-71>3. 18. J. Hu'!.,n ~t a!. "'Oper.tion Ch.,.ct.ri.tics of a V.rtic,,1 FI.,. Baking Furrw:o: Ugh! MtI.l, 1981 (W .. ",nd.I. , PAc TMS, 19S11. pp. 569-58l. 19. F. Kelk-r . nd J.H.M. Di • ..,lhorst, "Modern Anod, Bak. Fum_ [)o:vclopmf nt.- ibid., pp. 611-621 20. A. 50-" •. A. Murgil. and S. Chen:hi, -I mprove"",nt. in , he Op<ratlon 01 a C1o>ed Tyl'" Furnace lor Baking "nod",,: T..,hnkal Anatysi. oflhe ResuH.Ob1a iMll." Uth! M .... !' 1981 (W.rrendale, 1''': TMS, 1962). pp. 7105-778. 21. e M. ~t"" e! al.. "!mpfO"om..,.1 in Carbon Flaking T..,ltnology." ibid., 1'1'. 753-71>3 22. M. E. d. Ffrnandez. I. Ma,letto .• ~d H. Marti",,,.. "Combin<d M.,hrnoa tical Simul.1i<m . nd E, per;nwnl.1 Swdio-sonaCI_od BakingFumace.- i~ R~I. 11.pp.~17. B. C.s. s.cn..n.nd LJ. P"II.rd. ·~ •• l ~in~8 01. V~rli<. 1 Hu. Ring Fumace." i~ R,",. 7. pp. 1071-IOS7. K C Dre)"t'T, "H"".ontal F1u~ RinS Fur ... ", u.. .. ign .nd P~Conl~l: I)ghl M ... I, 1988IW."'-""<l.I~. PA: TMS. 1911S). pp. 303-323. 25. D.T. Stev..,.son, "M.t~ticol Mod~1 10" Ring Ty f't' Anode IIoking Fu",."".- ibid., 1'1'. 315-322. 26. U. M. nnweiler. S. Oderbol" and P. SuI,bo:'II~r, -New C""""pI for Firing Anod. 8 •• • Furn""" wit" !i .... y Fu.1 Oil.- in Rcl9. pp. 1>35-049. 27. L. LJem. ns • . "Anode ~Cont""l; ",-"""" IOOu $trial Implen>mtati<m$and Devclop"""'~." in Ref. 9. pp. 657-665. 28. D. M.nn .... eil.,-, P. Sukto"rg~r. and S. Oderbol • . "I'rore;;s Control in.n Anode Bak. Furno", Fin<! wi'h I teavyOiI; in R.,j. 9, pp. 667-671. 29. P. Hu,·.rnM. P. Fa)'<"l. and ].1_ I ........ r<" • .a. "R"","", Imp""vcm~n"olth. Anode l'roductiOll Process. Indu<lrial Application.and R"....It. Obtain<!d.- in Rri. 11. pp.821-M2 XI. I· Hurlen .nd T. N.te",t.>d. -New A,p<"C1s of C.rb .. " Flaking" (pa p'r "",,,,,nlcd al lho, ABALI A8M Cunl",,",,,,,. Soo Paulo. Bra,il, Nov. 9-11. 19t171. 31. S.R. 8 •• ndt .... -g. O. Ud,.nd J A,,,,,,n. -Anod~ Quality R. tat<!d loOp"rational I'".md""'br Baking. - in ~ef. lb. pp. 5'17-602

ABOUT THE AUTHORS ____ _

J . Hurlen earned his M.Sc. in electrochemis· try at the University of Trondheim. Norway, in 1970. He is currently direclor al Hydro Alumi­nium-Technology Services. Mr. Hur1en is also a member 01 TMS.

T. Naterstad earned his Ph.D. in high-tem­perature chemistry at the University 01 Trondheim. Norway, in 1973. He is currenUy senior manager at Hydro Aluminium-Tech­nology Services. Dr. Naterstad is also a member 01 TMS.

" you want more information on this subject, plaase circle re.ctet servlee e. rd number 58 .

Q Q Q

You can register for any TMS-sponsored meeting by simply picking up your telephone and dialing the above toll-free number. It's fast, easy, and convenient!

1991 November . JOM 25