CFD Simulation of a Hydrogen/Argon Plasma Jet Reactor for Coal Pyrolysis

CHEM. RES. CHINESE U. 2004, 20(4),446-451CFD Simulation of a Hydrogen/ Argon PlasmaJet Reactor for Coal PyrolysisCHEN H. G.1.2,and XIE K. C 11. Taiyuan University of Technology, Taiyuan 030024, P. R. china;2. City University, London, ECIV OHB, UKReceived June 3, 2004A Computational Fluid Dynamics(CFD) model was formulated for DC are hydrogen/argon plasma jet re-actors used in the process of the thermal H,/Ar plasma pyrolysis of coal to acetylene. In this model, fluidflow, convective heat transfer and conjugate heat conductivity are considered simultaneously. The errorcaused by estimating the inner-wall temperature of a reactor is avoided. The thermodynamic and transportproperties of the hydrogen/argon mixture plasma system, which are usually expressed by a set of discrete data, are fitted into expressions that can be easily implemented in the program. The effects of the turbulenceare modeled by two standard k-E equations. The temperature field and velocity field in the plasma jet reactorrere calculated by employing SIMPLEST algorithm. The knowledge and insight obtained are useful for thedesign improvement and scale-up of plasma reactorsKeywords Plasma, Reactor, Computational fluid dynamics, Coal pyrolysisArtcle|D1005-9040(2004)-04-446-05IntroductionSzekely 4. first tried to use standard k-e model in aThe thermal plasma pyrolysis process of coal turbulent flow plasma reactor. In the simulation ofacetylene has been studied to some extent. Fur- an air plasma reactor for inorganic ultrafine powderthermore, the reactors operated at power levels as synthesis, the enthalpy of the inner wall of the rehigh as 8-9 MW were built in Germany and USA. actor was artificially assumed 6)But this process has not been run commercially be-Computational Fluid Dynamics(CFD)is a dis-cause of high energy consumption and investment cipline that encompasses the numerical solution ofcost. So it needs to be improved further and to in- the equations of motion(mass, momentum and en-crease the efficiency. So far the researches have ergy)in a flow geometry of interest, together withbeen focused on improving the apparatus and subsidiary sets of equations reflecting the problemchanging the technological parameters. Coal hy- at hand. In this study, the CFD model was estabphere isvery complex phenomenon, the reaction mechanism CFD package PHOENICS was utilized to solve it.of which remains unknown. First of all, the plas- The CFD simulation results contribute to the im-a jet pyrolysis reactor needs to be developed, but proved reactor designaccurately understanding fluid flow and heat trans- Configuration and Geometry of Reactorsfer within reactor volume is of paramount imporTwo types of reactors including a cylindrictanceside slit feeder will be calculated. The reactors area plasma jet reactor, a kind of gas-solid ultra- tubular with an abrupt enlargement. The enlargshort contact downer, is usually treated as a black- ment ratio is 2.3. They are denoted as type a andbox. Nontransferred arc plasma generators, widely type b by the cold gas injection position before orused in spraying, heating, and surface treatment, after the enlargement, respectively. The outer wallhave been studied in detail. But little researches of the reactor is made of 5 mm-thick graphitehave been done on the numerical simulation of plas-cooled and kept at 298 K. Thema reactors. Chen a] numerically calculated the中国煤化工 rpe b reactor is madeplasma reactor for ultrafine particle preparation, ofCNMHGperature of which iswhich is laminar flow involved. Dilawari and also kept at zyo ncirculating cooling water.To whom correspondence should be addressed. Email h g. chen @city, ac ukNo. 4CHEN H. G. et al.44High-temperature plasma gas(H2/ Ar mixture )en- actors are schematically shown in Fig. 1.ters along the axis, whereas cold gas stream is injected from the 2 mm wide slit The reactors are soconnected after the dC plasma generator that nocurrent flows through the reactors nMathematical Model1 AssumptionsThe CFD numerical simulation is based on thefollowing assumptions (1)The plasma is in localthermodynamic equilibrium at atmospheric pres-sure:(2)the plasma is optically thin,(3)the electromagnetic forces are neglected (4) the fluid flows axisymmetric and steady-state;(5)the fluidflow is turbulentGraphite Capper2 Governing EquationsFIg 1 Schematic diagram of calculation domain forBecause of the axisymmetric assumption, thereactor type A(A)and reactor type B(B).governing equations can be expressed in terms ofThe conditions on the boundaries of the do-cylindrical coordinates(z, r)where z and r repre- main are specified as followssent the axial and radial direction, respectively. (1) Inlet(z=0).The transport equations for the numerical simulaw=wo,v=0, k=Awtion of a turbulent plasma jet reactor can be writtenin the general formr, and T=to(P)+13(rm4)=where A=0.0045,L0=0.5478()+1(r+a(1)(2)Exit(z=L1):where f is the general variable: I, the correspond0,中=w,U,k,E,Ting diffusion coefficient; S, the source term; w and (3)Axis of symmetry(r=0, 06.5 mm)5 Numerical CalculationIt is found from Fig. 2(A)that within the coreThe numerical calculation was performed byvelocity along the line of axis is very fastemploying an iterative, finite difference algorithm with the maximum velocity being 330 m/s but onknown as SIMPLEST.a. This algorithm has been the interface between the core zone and surroundapplied successfully to plasma reactors(3-6.ng zone (r=6.5 mm), the velocity drops toa finer grid was used in recirculation and nearm/s. The velocity field is very uneven in thethe wall region where large gradients in velocity whole reactor. The surrounding zone caand temperature are expected. The nonuniform sidered as the stagnant flow zone because the veloc-taggered grid system is formed by 70 divisions in ity is very small. The radial velocity gradient at r-direction and 55 divisions in z-direction; such 6.5 mm is -1.4x105s-I, whereas the axial velocgrid size has been checked to ensure the results in- ity gradient is very 'smalldependent of the grid resolutionThe contours of the axial and radial velocity5820m/s0.0150.0090060.0030.009A0.520.009中国煤化工CNMHGFIg 2 Computed veloclty field (A), axial velocity disttemperature field (D)using k-E model in reactor type A(case No. 1)No 4CHEN H. G. et aldistributions are presented in Fig. 2(B) and distribution, radial velocity distribution and temperature field in reactor type b for case No. 1 areshown in Fig 3. As for type B reactor, the veloci-which means there is a recirculation flow. Com- ty field near the inlet zone changes obviously. Coldpared with the axial flow, the radial flow is much gas stream is injected before the abrupt enlargeslower, but the radial flow pattern is very complex ment, which facilitates the injection of reactantsdue to the injection of cold gas stream. In addi- into the core zone. On the whole, the velocity fieldtion, the recirculation zone is found. The axial ve- after the abrupt enlargement does not change funlocity component w at z=0. 4 m unusually decreas- damentally, but the recirculation flowes because of the occurrence of slightly strong re- stronger. From Fig. 3(B), the axial velocity comcirculation zone at 2=0.4-0.5 mponent w at the inlet and that at z=0. 35-0.5 mThe temperature field, shown in Fig. 2(D), change greatly, forming a striking contrast to thatshows the temperature within the core zone is very of type a reactor. The distribution of radial velocihigh and very close to the inlet temperature ty component v also shows a big difference. There3900K. The radial temperature gradient is about is a strong radial flow near 'the inlet. From8X10K/m on the interface between the core Fig 3(D), it can be seen that the temperature as azone and surrounding zone. The closer to down- whole decreases greatly starting next to the inletstream, the less the radial temperature gradient is. for the sake of the injection of cold gas stream.The variation of temperature within the reactor There exists a local "thermal island "in the zonesvolume is not as great as that of the velocity. The of z=0. 21-0. 41 m and r<0. 005 m. The changeouter-wall temperature of the reactor is very low. of temperature distribution is very complicated, es-Owing to the strong thermophoresis of the thermal pecially in the zones of x=0. 3-0. 4 m and rplasma, radial cold gas stream is not injected into 0.008-0 012 m. In summary, for the same velocthe high-temperature core zone. So in this case ity of cold gas stream(vini=4.637 m/s), the veloci(vn=4. 637 m/s)the feedstock such as coal parti- ty and temperature fields within the reactor volumecle is unlikely carried into the high temperature change fundamentally just by changing the injec-core zone. The conversion of the coal and yield of tion position. The structure of type B reactor isacetylene are lowbetter than that of type A reactor to ensure theype B Reactfeedstock more easily to enter the high temperatureThe computed velocity field, axial velocity5820m/s0.0150.0150.0120.01210570.0090.0060.5O.40.300.50.40.30.20.10.0150.0120.01215850.0o928710003中国煤化工050.40.30.2CNMHGFig, 3 Computed velocity field (A), axlal velocity distrlbution(B), radial velocity distribution(C) andtemperature field (D)usIng k-f model In reactor type B(case No. 1)CHEM. RES, CHINESE U.Vol 20core zone, which is in favor of pyrolysis reactions. acetylene reaches 74% at the power level of52k2.1 Type A ReactorConclusionIn case No. 2, i.e., Uini=46 37 m/s, the re-A CFD modemulated to simulate dc arcsults differ a little. There are the recirculation zone thermal plasma jet reactors. Standard k-E turbuand the clear interface between the core zone and lent model and SIMPLEST algorithm are utilizedsurrounding zone. In the core zone of the reactor, in solution process. The velocity and temperaturethe axial velocity component w changes continu- fields within the reactor volume are obtained. Byously, which does not decrease at z=0. 4 m as in comparison, it is found that type B reactor is adcase No. 1. From the radial velocity distribution, vantage in structure, which ensures that the reacthe penetration depth is greatly enhanced and very tants are injected into the high-temperature coreclosed to the core zone. Also, the flow pattern is zone The simulation results agree indirectsimpler than that of case No. I. The temperature qualitatively well with those of technologicalfield of the surrounding zone is influenced obvious- perimental.ly due to the increase of the cold gas velocity. The Appendixtransient region between the core zone and sur- Nomenclatureounding zone becomes largerTurbulence factor2.2 Type B reactCConstantIn case No. 2, the difference of velocity field Eotexists at the inlet and in the recirculation zone FGas flow rateThe axial velocity component w at z0.012 m. The cold gas stream seems not to af- L,, L y Reactor length, position of cold gas injecfect the temperature field, which is almost identition, mcal to that of case No. 1. Compared with case NeGenerator power, kw1, the radial velocity component is very small and R,, R, Radius of reactor inlet, radius of reactorits distribution is much more complex. The axialouter wall, mflow is dominant It is noticeable that there is aRadial distance, mslight strong flow along the positive r-direction at s Source termthe exit of r=0, 008-0 012 m.Temperature, KFrom the above simulation results, the structure of type b reactor and large radial injectionTvvRadial velocity component, m.slocity of cold gas stream provide better velocity and wtemperature fields for pyrolysis reactions. So farAxial distance,mno experimental investigation into the velocity and sKinematic rate of dissipation, m.stemperature field within the H:/Ar plasma jet reac- General variabletor for coal pyrolysis has been conducted, which is r Diffusion coefficientvery difficult at least, if not impossible for the time A, 4, All Molecular viscosity, turbulent viscosibeing, for an experimental measurement because ofty and effective viscosity, kg.mtechnical difficulty and high expenditure. But thesimulation results can still be indirectly qualitative- P Density, kgm-3ly confirmed by technological experimental results o Prandtl/Schmidt of turbulence-kinetic-ener-for liquefied petroleum gas(LPG and pulverizedSubscriptsoal pyrolysis 7, 10. 1. Based on the above experi- 0 Axial inletmental conditions, the conversion of Shenfu coal injreaches 39. 88% and the yield of acetylene is中国煤化工36. 60%at the power level of 48. 1 kwt, the ReferCNMHGconversion of LPG achieves 76% and yield of [1] Chen H G, Zhang X, B. Zhang Y. F, et al.,Coal CowCHEN H. G. et al.451vrio,1996,19(2),19[7] Chen H. G,Ph. D. Dissertation, Taiyuan University of[2] Qiu J. S, Wang Q, Ma T. C.,J. Chen, Ind. and EngTechnolog, Taiyuan, 1999[8] Boulos M. L, Fauchais P, Pfender E, Thermal Plasma[3] Chen X, Acta Mechanica Sinica, 1987.19(1),52undamental and Applications, Vol. 1. Plenum Press, New[4] Dilawari A. H, Szekely J, Int. J. Heat Mass TransferYork, 19941987,3011),2357[9] Patantar S. V, Numerical Heat Transfer and Fluid Flow[5] Dilawari A. H, Szekely J, Metallurgical Transactions BMcGraw-Hill Book Company, New York, 198089,20B(2),243[10] Chen H. G, Xie K. C, Petroleum Science and Technology[6] Qu J. F, Zheng G. L, Engineering Chemistry&Metallur-2003,21(586),709gy,1990,(4),28[11] Chen H. G, Xie K. C, Energy Sources, 2002,24(6),575中国煤化工CNMHG