DYNAMIC MODEL AND SIMULATION OF EAF STEELMAKING PROCESS

ACTA METALLURGICA SINICA {ENGLISH LETTERS)Vol. 16 No. 3 pp 197 203.June 2003DYNAMIC MODEL AND SIMULATION OF EAF STEELMAKINGPROCESSQ. Li and X. HongShanghai Enhanced Laboratory of Ferronetallurgry, Shanghai University, Shanghai 200072, ChinaManuscript received 27 May 2002; in revised form 23 September 2002Based on mass and energy balances and theories of therrmodynamics and kirne:tics, adynamic model in conjunction with relative operation parameters was derined. Withthis model, the EAF process in later stage of scarp melting was dynamically described,including time dependent uariations of bath temperature, slag quantity, compositionsof steel, slag and ofF-gas, etc. Steelmakting process in a domestic 100 ton UHP ACEAF was chosen as a compustational r:carnple, (nd the simulated variations of tem-perature of hot metal and element contents was shown consistent with the actualprocessmeasurements within average error scopes of |OT|≤15K, |[O%C]|≤0.036,i[△%Si]| <0.00027, |[A%Mn]|≤0.002, |[A%P]| ≤0.00089 and |[O%S]| <0.001 re-spectively at the test points. The model will be extended to predict the whole produc-tion process including scrap melting procedure and evaluate its energy and materialconsumption, the produsction efficiency, elc., twhich will provide assistance for the im-provement of EA F operation and conlrul.KEY WORDSEAF, dynamic rode, sirnuslation1. IntroductionIn electric arc furnacc (EAF) steelmaking process, variable charging and different op-eration conditions cause continuous change of the furnace state and system parameters, itsbehavior cannot be accilrately described with a static model. In order to realize simula-tion and optimization of EAF process, transient process modelling is necessaryl1,2]. In thispaper, an advanced ruodel in the form of coupling diferential equations2- 4J was derivedbased on rass and energy balances, taking into account the metallurgical reaction kinetics,gaseous reactions, dynamics of materials and energy streams and exlensive thermody namiccalculations. With this model, dynamic prcdictions and simulations would be expected inconjunction with relative operation parameters.2. AssumptionsTo allow simplification and facilitate t.he model following assumptions were introduced.(1) The simulation was conductcd during the later stage of scarp melting process whenthe heat transfer from are is supposed directly to the liquid and gaseous phases, and thenfrom the liquid phases to the solid phases.(2) The formulation of scrap melting process was merely based on heat transfer. Theheat absorbed by the solid may be utilized to heat or mclt itself. The quantity of each onedepended on a temperature ratio between the solid中国煤化工(3) The scrap lump was dealt with as roundness 8:YHCNMHGnedtobemeltsymmetrically.198.(4) The solid phases consisting of the scrap and the solid slag were at the same temper-ature, while the liquid metal, molten slag and gas phase were all at the same temperature.(5) The injected oxygen reacted merely with [Fe], and then the resultant of (FeO) wouldreact with carbon powder injected and other elements in the hot metal.(6) The oxidations of [C], [Si] were ranked in apparently first order reactions and theoxidation rate of carbon powder was proportional to its injection rate and content of ferrousoxide in the liquid slag.(7) All reactions took places at the same temperature.(8) The two-film reaction theory was used to describe the dephosphorization reaction5whose rate was limited by the mass transfer rates of phosphor in both the hot metal andthe liquid slag. The changes of [S] and [Mn] were dealt with in the same way.(9) The heat and mass transfcr cofficients were constant.(10) A relationship of the leaked air flow rate proportional to the square root of thevalue of the relative pressure in EAF was deduced by dimensional analysis.3. Variables and Model DescriptionThe variables and related parameters were listed in Appendix I. Accordingly, the dy-namic model was mainly constructed with following nonlinear coupling differential equa~tions.Emsr-MsrQ2(Ts/T)(1)OtAHsr+ Cp,sr(T-T)0mar1= _ me - S+(i=Fe， C, Si, Mn, P, S)(2)8tHthere,(i = Fc)i:=ka;(Xi- xi)(i=C, Si)Aen 1%i1-(%)/Lps.(i=Mn, P, S)17Pm Eq)+17(Psik(1)Lp.i)irpo= -(2MreO)wc+ (MreO)rgr+MReO)rc + (2MneO)is + (MmeO)riMnMO2f(McMcMsi” MMnkcmpeOWgrms,1a[%i]amar100=-(ri +(i%)}s);(i=C, Si, Mn, P, S)(3)otTesr, 1Omsl, 1=-MsI, sQ3(Ts/Ti) .(4)△Hs1+ Cp,中国煤化工0msI,s_ 8mgl,TYHCNMHG“sue(5)0t1998mFeO= -TFeO(6)Htomim-)gas - r + fri9leak .(i=CO, CO2; N2, O2)(7)tmgashere,Gleak ={ -kpVPRT(Pr> 0)( kp√|Pe| (Pr<0)i= fi(W)(irc + fgr)(i= CO, CO2, N2, O2)J fco=-1, fco,=0, fo2=0, fN2=0(Xo2 < xq2)f:{fco=0, fco,=-1, fo2=1/2, JN2=0 (Xo2 > X8q)oTPg + QRr- kx(Ti - Tair)(8)msr, ICp,sr,1/Msr,l + 7s,JCp,sL,/Mst,IATs(Q2 + Q3)(1 - Ts/T)(9msCp.sr/Msr + ThsL,Cp,Is/ Ms,sOPrR，.(5 )M; Ot(i=CO, CO2, O2, N2) (10)DtM;'HtvlThe correlative terms of heat income were listed as follows.(1) Chemical rcaction heat:Fe+ 1/202 + FeOCO + 1/2O2→CO2C+(FeO)→Fe+CO[C] + (FcO)→Fe + CO[Si] + 2(FeO)→2Fe + (SiO2)[Mn] + (FcO)→Fe + (MnO)[P] + 5/2(reO)→5/2Fe + 1/2(P2Os)[S]+ (CaO) + Fe→FeO + (CaS)Q1= ZIC(NHsc)react - E(0H,;)resut]Herc,瓷refers to the molar reaction rate of cach term above.(2) Physical heat中国煤化工Heating and melting of scrap Q2 = Asrm&(ms/m.MHCNMHGIleating and melting of slag Q3 = Agmi,s (ms/msl,s,. ns1-1(41- 1s)200Heating of injected carbon powder Q4 = wgrCpger(Ts - Tair)/Mc .Hcating of injected oxygen Q5 = wo2Cp,o2(T - Tair)/Mo2Heating of leaked air Q6 = kpVTPrTCp,air(T- Tair) (PRr <0)Qr=Q1-2Qi(11)4. SimulationIn order to test and verify the model, the process of a specific domestic 100t UHP ACEAF in commissioning was chosen for simulation, so that more test points available thanusual process could be used. Meanwhilc, some unusual operations would appear but wouldhave no effect on the verification of the model. The changes of the operation paraneterswere shown in Fig.1. The initial temperature of hot metal was 1811K, while that of scrapwas 800K. Other initial conditions were listed in Table 1. The model was solved using theclassical 4th-order R-K algorithm, and the results were shown in Fig.2 to Fig.7.00 r208-- PE, 10MW.--. w kg/s10-一Scrap.... Hot metal80七02030”1300t， mIn1, minFig1 Change of operation parameters.Fig.2 Change of mass of scrap and hot metal.2000 f.04)4..... 1 CI，(Mn]X 1500, [SI02021■T, Measured1000T. Calculated--- T ,Calculated-。0.000.0203040t,mir1, mln中国煤化工Fig.3 Temperature change of scrap and hotMHCN M H Gntent in hot metal,netal.The symbols denote the measured values.20112Liquid slagCaO ..4t多49u kg/s一Pqr Pa2-Solid slagFeO/SiO, MnO PO5a0203010340I, minFig.5 Change of slag composition.Fig6 Change of furnace pressure and off-gas fanfow rate.Table 1 Simulation conditions and related paramneters of kineticsm&rmgr,1 mg, mgh.。 m&ao m;ozmieo mMno mis)mp2os[%C}°20000 80000 4000 () 3000500200168.70.395[%Sj' [%Mn]" [%P]° [%S]° (C%)er (Si%)r (Mn%)er (P%)er (S%)ermoomo20.0074 0.175 0.02 0.03 0.2% 0.007%0.2%0.02%0.032%7.8mx, m02kd,c ked,Ssikehpk(p)ki(s)k(s)KjMn]k(Mn)10.3233.1x10-4 4.1x10-5 4.5x10- 42.0x10-4 8.9x10-4 2.5x10- 45. Results and DiscussionThe quantities of the scrap and the hot2co,metal were shown in Fig.2.The temper-ature variation of the gaseous phase, solidphases and liquid phases in EAF were shownin Fig.3. It was obvious that the simulatedcurve of the liquid steel temperature coin-cidcd with the measured values fairly wellwithin the error scope of士15K. The liquidsteel temperature reduced slowly during the624324time of switch of when it was eleven min-t, minutes aftcr the simulation started, which couldbe attributed to the injection of the carbon-Fig.7 Change of off gas composition.oxygen lance through energy balance analy-sis, th18 being certain that the operation of carboil-oxygen lance has a great effct on theliquid phase temperalure. The temperature of the scrap rose ceaselessly.Fig.4 showed the variations of each element in the hot mctal. The quantities of carbon,silicon and phosphor in the hot metal were. all dec中国煤化工: process. Thedecreasing rate of carbon content in the hot metal|Y片CNMHGvertheoxygenlance operation starts. It was known, when the oxygen mnjection startea, the content offerro1s oxide in the slag would increase (Fig.5). As a result, the balanced content of carbonin the hot metal near the slag metal interface was decreased, which increased the powerof mass transfer of carbon from metal to slag. Comparatively, the balanced. content ofsilicon in the hot metal near the slag-metal interface was much lower than its contcnt inthe bulk, so the variation of the oxidization rate of [Si] was infuenced by the change ofthe oxygen lance operation faintly. As to phosphor, manganese and sulphur, the directionand rate of mass transfer are determined by both the variation of distribution coefficientbetween slag and metal and the content of element in hot metal. As it was shown inFig.4, dephosphorization rate was decreased generally, which was mainly caused by thecontinuous decrease of the content of [P]， while the effect of the decrease of balancedcontent of [P] near the slag mctal interface caused by the increase of the content of (FeO)was comparatively small. On the other hand, the rise of temperature of hot metal couldrestrain the reaction considerably. The sulfur content in the hot metal represented a trendof monotonous increase, while the manganese content in the hot metal would go up anddown. The content of ferrous oxide in the slag was high, so as for desulfurization of thehot metal, which is sensitive to ferrous oxide content in slag, would be dificult, and sulfurin the slag will go back to the hot metal if the content of (FeO) is high. The fuctuationof manganese content was mainly caused by the fuctiation of ferrous oxide content in theslag. .It was shown that the simulated variations of elemcnt contents in the hot metal hadthe same tendencies as the measured values within average error scopes of |[O%C]| <0.036,|[A%Si]| <0.00027, |[A%Mn]| <0.002, |[△%P]| <0.00089 and |[O%S]| <0.001, respectivelyat the test points. The adjusted parameters of kinetics were listed in Table 1.As one of the operation parameters, off-gas fow rate should be controlled accordingto EAF inner pressure. The off-gas fan fow rate in the simulation can be adjusted to theEAF pressure dynamically in order to control the pressure in a favorable scope. Referringto Fig.6, the off-gas fow rate went up rapidly whenever the car bon-oxygen lance injectionstarts, and would decrease gradually when the lance operation stopped. As a result, therelative pressure could be controlled to a valuc of - 3Pa with only a minute deviation. Thecontents of all gases (especially CO gas) in the EAF (Fig.7) fuctuated greatly, which wouldbe thought reasonable according to the characteristic of modern EAF operation.6. Conclusions(1) A dynamic model of EAF steelmaking process in a form of nonlinear differentialequations was derived, and a simulation for a 100t UHP AC EAF process was carried out.(2) The simulated variations of hot metal temperature and element contents in hotmetal had the ability to reflect the change tendency of the actual process. In this sim-ulation，the average error scopes are |OT| <15K, |[O%C]| ≤0.036, |[△%Si]| <0.00027,|[O%Mn]| <0.002, |[△%P]|≤0.00089 and |[△%S]|≤0.001, respectively at the test points.It could be concluded that the rcaction mechanism of the model was primarily proved.(3) The model will be improved to predict the whole process and evaluate its energyand material consumption, the production efficiency , str，which, will he helpful to the中国煤化工improvement of EAF operation and control. .TYHCNMHG203REFERENCES1 M. Hofer, P.L. Steger, J. Lehner and W. Gebert, Steel Times 225(3) (1997) 108.2 A.A. Linninger, M.A. Hofer and A.A. Potuzzi, lron and Steel Engineer 35(3) (1995) 43.3 J.G. Bekker, I.K. Craig and P.C. Pistorius, ISIJ Int. 39(1) (1999) 23.4 R.D. Morals, H.H. Rodriguez, A.S. Romero, R. Lule and J. Camacho, 1995 Electric Purnace ConferenceProceedings, p.237.5 Y. Qu, Sleelmaking Principles (Beijing: Metallnrgy Indlustry Press, 1994) (in Chinese).Appendix I Variables and related parametershimass of material i, kgFeOFeO in liquid slagm{initial mass of material i, kgCaO .CaO in liquid slag[%i]_mass percentage of element i in hot InetalSiO2 in liquid slag[%i]° .initial mass percentage of element i in hot, netalMnOMnO in liquid slag(%i)mass percentage of element i in liquid slagP2OsP2Os in liquid slag(%i)°initial mass percentage of element i in liquid slag(S)S in liquid slagTliquid phase temperature, KCOCO in gaseous phaseTssolid phase temperature, KCO2CO2 in gaseous phasePrrelative pressure value, PaNzN2 in gaseous phaseOO2 in gaseous phasesr, Ihot metalgasgas in EAFsl,s .solid slaggcarbon powdersl,1liquidt slagAsr, Aslproportional coefficient between the surface area and initial mass of scrapand solid slag, m /kgAgI-1reaction area of metal-slag interface, m2Cp,;thermal capacity of mrlaterial i, J/(molK)fr,imass fraction of gas i in gas phaseOHsr, OHsl melting enthalpy of scrap and slag, J/kgAHr，standard enthalpy of formation of material i, J/kg(i%)srmass percentage of element i in scrapkodimensionless coefficient for reaction of carbon powder with (FeO)ksapparent rate constant of reaction between element [国and (FeO), kg/sk问, k(i)mass transfer ceficients of elment i hot metal and in slag, m/sproportional coefficient between leak-air entry rate and relative prossurc, kg/Pa1/2ker-l, kg1-1 coefficients of heat transfer from liquid phase to scrap and solid slag, J/(Km2s)k:integrated heat transfer coefficient of EAF, W/sLpnamely (%i)/[%i], cefficient of element i distribution between slag and hot metalMimolecular weight of material iPereal power, WQnheat income of physical and chermical reactions, J/s9gasflow rate of gas from furnace to duct, kg/sleakleaked air fow rate, kg/sgas constant, 8.314J/(molK)reaction rate of material i, kg/stemperature of atmosphere, Kvolurne of gas in furnace, m'additiou rate of material i, kg/sXi, X9qmole fraction and balanced mole fraction of materia1Pm, Paldensities of liquid metal and slag, kg/m3中国煤化工HYHiCNMHG