Numerical Simulation for Coal Carbonization Analysis of a Coke Oven Charge Using PHOENICS Huiqing Tang1 Zhancheng Guo1 Xinming Guo2 ABSTRACT

Transcription

1 Numerical Simulation for Coal Carbonization Analysis of a Coke Oven Chare Usin PHOENICS Huiqin Tan 1 Zhanchen Guo 1 Xinmin Guo 2 1. Institute of Process Enineerin, Chinese Academy of Sciences, Beijin, , P.R.China 2. School of Metallury & Ecoloy, Uni. Sci. & Tech., Beijin, Beijin, , P.R.China ABSTRACT A Computational Fluid Dynamic model is analyzed for cokin process in a coke oven chare usin PHOENICS CFD packae (Version 3.2). The model reported in this paper consists of a combination of a two- dimensional model for coke oven as phase flowin in porous media and a one-dimensional model for semi-coke phase toether with chemical reactions, heat and mass transfer between the two phases durin coal carbonization in the coke oven chare. The model simultaneously calculates the transient state composition, temperature of all two phases, velocity of as phase and porosity, density of semi-coke phase. Numerical simulation is illustrated in the predictions of evolution of volatile ases, as flow paths, profiles of density, porosity of the coke oven chare, profiles of temperatures of the coke oven as and the semi-coke bed. The predictions aree well with the published data on cokin process. KEY WORDS: coal carbonization; porous media; numerical simulation; PHOENICS 1 Introduction Cokin process has remained basically unchaned for over 100 years. Coal, crushed so that some 80% of the particles are <3mm diam., is chared from the top into slot-type ovens. The ovens are heated indirectly throuh the side walls, which are usually made of silica refractory brick. There have been many models developed for it. D. Merrick [1~5] performed distinuished study on cokin process. Besides havin proposed a series of mathematical models, he meanwhile has iven a ood summary on the study of cokin process. Yet, all these foreoin studies could not ive a comprehensive understandin of this process. To date, no model has ever considered the as-solid coupled problems in cokin process. As B. Atkinson and D. Merrick [1] pointed in their paper., cokin process still needed a more advanced study for cost-effective purposes. With the advent of modern hih-speed computers and CFD technoloy, it has become possible to construct a comprehensive mathematical model of this complicated multidimensional and multiphase process in the coke oven. The purpose of this paper is to extend a CFD model of cokin process in coke oven to ive some detailed information durin cokin process. 2 Model Formulation Two phases are considered in the model, they are coke oven as (COG) and lump coal chare. A two dimensional mathematical model considers mass, momentum, enthalpy and chemical species conservation for as phase at transient state, and a one dimensional mathematical model enthalpy conservation for the lump coal phase. The eneral conservation equation for COG as phase is iven in Eq. (1)

2 Ψi Ψi ( ερψ i) + ( ερuxψ i) + ( ερuyψ i) = ( εγ Ψ ) + ( ε ) i Γ Ψ + S i Ψi t x y x x y y Eq. (1) is used to represent all conservation equations of the as phase by settin Γ and to appropriate values accordin to the dependent variable, as listed in Table 1. Note thatε is the volume fraction of the as phase. Ψi S Ψ i (1) Tab. 1 Terms used in Eq. (1). The symbol - indicates this term is not applicable Equation Name Ψ i Gas continuity 1 Γ Ψi S Ψ i _ R volatile + R moisture Gas Momentum Gas Momentum u x u y µ F x µ F y Gas enthalpy H λ CH 4 mass fraction C 2 H 6 mass fraction m CH4 m C2H6 D D CO mass fraction m CO D CO 2 mass fraction H 2 O mass fraction m CO2 m H2O D D H 2 mass fraction m H2 D E s CH 4 R R CH 2 6 R CO R CO 2 R H2O R H 2 NH 3 mass fraction H 2 S mass fraction m NH3 m H2S D D R NH 3 R H2S

3 Heat transfer is considered to be one-dimensional for the lump coal chare for this assumption is most accurate when simulatin commercial ovens where the lenth/width ratio is up to 30 and heiht/width ratio is up to 10. [5] The chare is considered to be uniform within vertical planes. Thus the enthalpy conservation equation for the lump coal phase is iven by Tc Tc ρ cc pc = ( kc( )) + S( t, x) (2) t x x Treatment of Momentum Transfer Gas-solid dra term uses Erun s expression in an isotropic form to account for the resistances of lump coal to coke oven as flow in x direction and y direction are shown in Eq. (3) and Eq.(4) [7] C C F = µ G G G (3) x 1 2 x x x ρ ρ C C F = µ G G G (4) y 1 2 y y y ρ ρ Treatment of Heats of Formation in Enthalpy Transfer The enthalpy of content of the as phase was calculated by H = H + H (5) 298 T where H T T = Cp 298K dt The convective as-solid heat transfer is neliible for velocity of the as phase is relatively small. However the volatile-release and moisture evaporation reaction enthalpy transfer sources should be included. The source is in Eq. (6) E = Cp T R i=ch 4, C 2 H 6, CO,CO 2, H 2 (6) s c i i Release of COG Release of volatile matter from coal chare follows the sequential scheme of first-order reactions as Eq. (7). Coal Metaplast Metaplast Semi coke + primary volatile matter (7) Semi coke coke + second volatile matter For any part of bulk coal chare, Composition of the volatile matter in it is defined in terms of the

4 followin 9 species:ch 4,CO,CO 2,C 2 H 6,TAR,H 2,H 2 O,NH 3,H 2 S. The kinetics of volatile matter release are described in the model by a system of parallel first-order reactions for which the rate constant varies with temperature accordin to an Arrhenius relationship [2],which is iven by E [ exp( )]( ) i Rti, = ci m0, i mt, i RTc m (8) = R dt (9) ti, 0 ti, The heat effect of volatile matter release on the solid lump chare is nelected in this paper for this effect is somewhat included in the calculation of the specific heat of coal chare which will be discussed in the followin sections. t Boilin of moisture Coal chare usually contains moisture. As the chare heated, the moisture near the oven walls boils. The reaction is as follows HOl () = HO ( ) (10) 2 2 It s assumed that at 373K in the nearest plane to the oven wall. The followin equation defines the position of the boilin plane. T = 373K c Tc ( ) xb+ = 0 x ( w ) = moisture xb 0 (11) where x b is the distance of the boilin plane from the coke oven wall and the suffices +,- refer to a approach from the low and hih temperature sides respectively. Boilin is represented in the model as a point sink (one dimensional) of moisture and its associated latent heat. At any instant, mass and heat balance at the boilin plane therefore ive respectively: R H 2O dtct, = ( kct, ) xb /3. 2e 6 (12) dx dtct, Stx (, ) = H= ( kct, ) xb (13) dx Gas phase properties Tar bein excluded,density of coke oven as is calculated from the ideal as equation, thus it is a function of temperature and pressure. Kinetic viscosity and specific heat of the as are assumed to be constant. They are 0.8E-5m 2 /s and 1500J/k [8] respectively.in order to evaluate the thermal conductivity and mass diffusivity of the as, for simplicity, The dimensionless numbers, Pr and

5 Sc, are all considered to be 1.0, which is correct for most of ases under hih temperature. Solid phase properties and modifications For cokin coal, density, thermal conductivity and voidae (also the volume fraction of COG as phase) are the functions of temperature and composition of coal in cokin process. Effective thermal conductivity The equation for the overall effective thermal conductivity of cokin coal is combined with equations for the COG as conduction and radiation components of heat transfer in two models of the chare thermal conductivity, one model describes particulate coal and the other semi-coke [1]. For particulate chare, the model is iven by Eq. (14). The model combines the thermal conductivity of coal chare (k 0 ), the conductivity of the as phase (k 1 ), the effective thermal conductivity of radiation (k 2 ) and the thermal conductivity of moisture (k 3 ). where kc wmoisturek3 wmoisture e k1 k2 e k 0 = + (1 )/( '/( + ) + (1 ')/ ) (14) k = ( d / ) T k k s (10 ) Tc = 2.28(10 ) T e' = 1 (1 ε ) k = = c 1/ c After resolidfication at the critical temperature, the effective thermal conductivity of coke chare becomes = ε + ε k (15) kc (1 ) k0 4 where 5 k4 = 4.96(10 ) T c Specific heat Published experimental measurements related the specific heats durin thermal decomposition are based on the total heat required for carbonization to various temperatures. Heats of reaction can be assumed neliible. In this paper, the model of specific heat of coal and coke is iven by Eq.(16), which is proposed by D.Merrick [3] Cp = ( R / a)[ (380 / T ) + 2 (1800 / T )] (Jk -1 K -1 ) (daf basis) (16) daf 1 c 1 c 5 Where 1/ a = y i / ui (ui=12,1,16,14,32), y i is the mass fraction of C,H,O,N,S of the coal 1 chare on daf basis. The meanin of function 1 could be referred to literature [8]. If ash and moisture in coal are included, the specific heat is iven by

6 Cp c 3 = wicpi 1 (i=ash, daf coal, moisture) (17) Voidae of coke-oven chare An important factor affectin flow resistances is the chare voidae fraction available for as flow. The approach adopted in the present model was to assume that, in terms of the resistance to the as flow predicted by the Erun equation, all of the chare can be rearded as a bed of particles of constant diameter m. The differences in flow resistance assumed to exist are reflected by variations in the effective porosity with position in the chare. The variation in porosity was chosen arbitrarily to correspond to a minimum value in the plastic layer. The values are polyfitted usin the data based on literature [5]. ε = c (18) max(0.1, Tc 5.20E 5Tc 4.52E 8Tc 1.36E 11 T ) In the above equations from (6) to (18), some are functions of the daf-basis composition of cokin coal. The daf composition depends on the cokin time for the occurrence of chemical reactions. And thus these compositions are evaluated by mass balance with cokin process oin on. Boundary Conditions Gas flow boundary conditions Boundary conditions on the as flow are expressed in terms of the flow rate and can be of two types: 1. At a surface open to the atmosphere, the pressure is fixed to zero. 2. The boundary condition of the centre line of the oven (the line x=0 in this paper) is set as symmetry boundary conditions. 3. For the lump solid phase, the temperature of the coke-oven wall is fixed to 1473K and for the as phase, the coke-oven wall is adiabatic. 3 Method of Solution The model is solved numerically usin PHOENICS. All calculation is performed usin a Cartesian rid of 215 y-directional and 22 x-directional divisions. The size was found to be acceptable by rid dependence test. Grids layout is illustrated in Fi Results and Discussions The calculation uses eometrical and operational data from a coke oven of inner volume 25.2m 3 operatin at a chare of some 17 ton. Operational data and bulk solid properties are described in Table 2.

7 Tab. 2 Data for calculation Coal Specification (wt%, daf basis) C H O N S (diff) Chare conditions Moisture content Volatile matter Ash yield Dry bulk density Temperature (%) (%) (%) (k/m3) (k) Oven Specification Oven Width(m) Oven heiht(m) Oven lenth(m) The cokin process lasts some 13~14 hrs. With the model proposed above, it s possible to describe the whole process. Initial condition of simulation is that at the start instant, the temperature of every rid in the eometry is 373K, the as composition is 100% water vapor with velocity manitude bein 0.0m/s. And initial composition of the coal is listed in table 2. These assumptions are nearly 2h after chare. The activation enery of each species in COG is listed in table 3. Also in table 3 is the final yield of each species in COG which is evaluated with the method in literature [2]. The frequency factor in Eq. (8) is for all species in COG. Tab. 3 Parameters values used in Eq. (8) CH 4 C 2 H 6 CO CO 2 TAR H 2 H 2 O NH 3 H 2 S E/(kJ/mol) M 0 /(k/k coal) Heat Transfer and Temperature profiles in chare The calculated temperature history profiles of the chare are shown in Fi. 2. In this and subsequent fiures, only half of a vertical cross-section of the coke oven is shown. The model reproduces the main features of the measured profiles and shows reasonable areement with data on coke-oven chare temperature histories. However the source term in Eq. (2) includes only the heat source of water boilin. The inorance of reaction heat of volatile matter releases leads to temperature at the chare centerline increases more rapidly than the measured and the data reported [1]. Fi.3 shows the history of the chane of voidae of the chare. it has a close relation with the COG as flow pattern which will be discussed in the followin section. As shown in Fi. 3, the

8 voidae fraction chanes much more rapidly near the coke oven wall than in the center. Evolution of COG The predictions are illustrated as COG paths and y- direction velocity profiles in Fiure 4 for the different staes from 2h to 10h after charin. At the beinnin of cokin (2 hours after charin), COG escapes the coke oven from top surface of chare around the central line and little as escape alon the wall. At time from 3h to 4h after charin, the as flow is clearly divided into two parts: One flow of as cross the cool side of the chare (near the center line). Most of COG escapes from this reion. And the other, a small part of COG escapes from the channel at the wall. From 5h after charin on, the flow pattern of as still chanes. The main COG-crossin reion is slowly moves from the reion near the center line to the channel at the wall. Before the stae at 6h after charin, the manitude of y-directional velocity increases quickly and after that instant, the manitude of y-directional velocity decreases slowly. After the cokin time reaches 10h., there is scarcely any COG evolvin from the chare. The main chanes then are resolidfication, contract of semi coke and chare temperature increasin. The reason of the COG flow pattern can be explained in terms of the voidae distribution. COG produced mainly in plastic layer will tend to move towards the nearest open surface the top of the oven and the channel at the wall. The area of low voidae reion and hence hih resistance between the COG and water vapor producin reion and the surface impedes this natural movement of the as. However, points in the plastic layer are near two open surfaces, viz. the top and the wall, and it s reasonable to expect a lare portion of COG to cross the plastic layer here than elsewhere. Fi. 5 shows the pressure drop in the coke oven at different cokin stae. Pressure drop of the coke oven somewhat reflects the COG producin rate. From Fi. 5, it clearly shows that at time form 5h to 6h after charin, the pressure drop reach its maximum and hence, in this period, the chare release a lare amount of COG. Investiation the composition of COG shows that at initial stae, the main component of COG is water, which occupies 90 %( mass fractions). The two other main compositions (H 2 and CH 4 ) are small. The as flow pattern at 5h after charin is illustrated in Fi. 4. In this stae, COG mainly escapes from the channel at the wall while there is only a very small COG flow near the center line. In this stae, the components of COG are mainly H 2 and CH 4, Fi. 6 illustrated the as phase mass fractions of methane at different cokin staes, and other as components could also be investiated in a similar way. Indicated from Fi. 7, at time from 2h to 3h after charin, moisture in the coal chare evaporates quickly and thus the main composition in the as phase is water vapor. While at 5h~6h after charin, other COG species releases vehemently. CH 4 mass fraction in some reions in the oven at 3h is hiher than at 4h and 5h. This is due to the producin rate of COG. As illustrated in Fi. 2, in time from 3h the coal temperature is low except at reion near the wall.

9 AT last, the temperature of COG is investiated. The temperature of COG has a close relation to the temperature of coal chare. The as temperature profiles at different staes of cokin process are illustrated in Fi. 8. Gas temperature in the coke oven at 3h is much hiher than those at the followin staes such as (4h, 5h and 6h). This is due to the evolvin rates of COG at correspondin cokin staes and the inorance of convective heat transfer between as and coal. 5 Conclusions A CFD model to describe COG flow pattern and coal carbonization behavior in the coke oven has been developed usin PHOENICS CFD packae. For typical operatin conditions, the model predicts that the major chemical reactions and physical structures modification within the chare. It s expected that this model is an appropriate base from which to include full multiphase interaction as well as chemistry. Acknowledement The authors ratefully acknowlede National Natural Science Foundation of China (NNSFC) for the financial support (UNDER PROJECT NUMBER ) Nomenclature Cp: specific heat (J/k.K) D: diffusivity (1/m 2.s) d s :mean particle size of coal chare (m) C 1, C 2 : constant E : heat source (W/m 3 ) E: activation enery (kj/mol) F : dra force (N/m 3 ) G: mass velocity of as phase (k/s) H: enthalpy for as (J/k) k: thermal conductivity of coal chare (Wm -1 k -1 ) m 0,i : final yield of species I of volatile matter (k/m 3 ) m t,i : yield of species I of volatile matter at instant t (k/m 3 ) P: pressure (pa) R : rate of release of COG component i R: as constant T: temperature of as or coal (k) t: time (s) u: x directional velocity component of as phase (m/s) v: y directional velocity component of as phase(m/s) w: component mass fraction in coal chare x: x directional coordinate (m)

10 y: x directional coordinate (m) Greek Symbols ψ:eneral dependent variable in Eq. (1) Γψ:diffusion coefficient forψ in Eq.(1) μ:viscosity (Pa/s) λ: thermal conductivity of as ρ: density (k/m3) ε:volume fraction of as, demensionless Subscripts : as c: cokin coal i: COG as speciest t: instant x: x direction y: y direction References 1. Atkinson, B., Merrick, D. Fuel, : 553~ Merric,D. Fuel, 1983,62: 534~ Merric,D. Fuel,1983,62: 540~ Merric, D. Fuel, 1983, 62: 547~ Voller, V.R., Cross, M., Merrick, D. Fuel,1983, 62: 562~ Smith, J. M. Chemical Enineerin Kinetics McGraw-Hill, New York, Szekely, J. Evans, J. W. Sohn, H.Y. Gas-solid Reactions, Academic Press New York, Reid, R. C. Prausnitz, J. M. Polin, B. E. The Properties of Gases and Liquids, 4 th Ed. McGraw-Hill, New York, (1988)

11 Fi rids used for the calculation

12 Fi. 2 History of chare temperature profiles Fi. 3 History of chare porosity profiles

13 2h 3h 4h 5h 6h 7h 8h 9h 10h LEGEND (m/s) Fi. 4 Y-directional velocity contours and streamlines patterns at different cokin stae

14 Fi. 5 Pressure drop from the bottom of coke oven to the chare surface at different staes

15 2h 3h 4h 5h 6h 7h 8h 9h 10h leend Fi. 6 CH 4 as phase mass fraction at different staes of cokin process

16 2h 5h Fi. 7 water-evaporatin rate at different cokin staes

17 2h 3h 4h 5h 6h 7h 8h 9h 10h leend Fi. 8 Gas temperature profiles at different staes of cokin process