回转窑(英文)

Coal Pyrolysis in a Rotary Kiln: Part II. Overall Model of the Furnace FABRICE PATISSON, E ´ TIENNE LEBAS, FRANC¸OIS HANROT, DENIS ABLITZER, and JEAN-LE ´ ON HOUZELOT In order to simulate coal pyrolysis in a rotary kiln in the steady-state regime, a mathematical model has been developed which calculates the temperature profiles in the charge, the gas, and the furnace walls, together with the gas composition and the degree of removal of volatile species. The model takes into account the principal physicochemical and thermal phenomena involved, including the complex movements of the charge; the gas flow; heat transfer between the charge, the gas phase, and the furnace walls; drying and pyrolysis of the coal; the cracking of tars; the combustion of volatile species; and the combustion and extinction of the coke. The data necessary for the model were obtained by specific experiments or from the literature. The model has been validated by comparing its predictions to measurements performed on an industrial rotary kiln. The model has been used to study the influence of operating parameters such as the furnace rotation speed, in order to optimize the process. It is shown how a modification to the extinction zone leads to an increase in coke yield of 0.75 pct. I. INTRODUCTION published in the literature,[1–5] none applies specifically to the pyrolysis of coal. Another point which distinguishes theROTARY kilns are widely employed in the cement, present study from previous work is that the description of metallurgical, and chemical industries due to their simplicity the conversion of an individual coal grain to coke led us to of use and their ability to continuously treat granular or develop a specific kinetic and thermal model, which haspowder solids with excellent heat and matter transfer been termed the grain model. The latter has been describedbetween the solid and gas phases. The present article consid- in detail in Part I of this article[6] and is incorporated in the ers the use of the rotary kiln for the pyrolysis of coal to overall model of the kiln presented in this Part II of theproduce high-reactivity coke for electrometallurgical appli- article. After considering the movement of the charge, heat- cations. The coal is introduced at one end of the furnace in transfer, and combustion phenomena, the article goes on tothe form of 1- to 2-cm-diameter grains and flows parallel describe the model itself and the results obtained.to the kiln axis, which is slightly inclined to the horizontal, due to the rotation of the tubular body. The coke produced is discharged at the other end. A counterflow of air is intro- II. MOVEMENT OF THE CHARGE duced at the coke end and becomes laden with volatile In a rotary coal pyrolysis kiln, the charge moves in thespecies evolved by the pyrolysis, while at the same time so-called rolling mode, according to Henein’s classifica-being deplenished of oxygen due to the various combustion tion.[7] In this mode, the bed of grains can be divided intoreactions, which provide the energy necessary for the two zones, corresponding to the surface of the bed, whereprocess. the grains roll due to the effect of gravity, often called theThe goal of the present article is to describe a complete active zone and composed of one or several layers of grains,mathematical model of the kiln intended for process optimi- and the lower part of the bed, where the solid undergoeszation. The model describes all the physicochemical and a circular movement imposed by the kiln wall. Figure 1thermal phenomena of importance for the process, including schematically represents the trajectory of a grain in a rotarydrying of the coal, the removal of volatiles, cracking of tars, kiln. When the grain is in the lower part of the bed, it doesthe combustion of the volatile species, the partial combustion not advance parallel to the kiln axis. Once it has come toand extinction of the coke, the solid and gas flows, and heat the surface, it follows the line of greatest slope, whichtransfer between the charge, the kiln walls, and the gas. depends on the dynamic rest angle of the charge (b) and onAlthough several rotary kiln simulation models have been the inclination of the bed with respect to the horizontal plane (u). The plane grain is then reincorporated into the bed in FABRICE PATISSON, Research Scientist, is with the Centre National a random manner. de la Recherche Scientifique (CNRS), Laboratoire de Science et Ge´nie des In the general case where the height of the bed (H ) varies Mate´riaux Me´talliques (LSG2M), E´ cole des Mines, 54042 Nancy, Cedex, along the z-axis (cf. Figure 2), the angle u varies. The chargeFrance. E´ TIENNE LEBAS, Research Scientist, formerly with LSG2M, profile (H(z)) generally depends on the solid flow rate andEcole des Mines, is with the Institut Franc¸ais du Pe´trole, 69390 Vernaison, on the heights of any diaphragms present at the entranceFrance. FRANC¸OIS HANROT, Research Scientist, formerly with the Cen- tre de Pyrolyse de Marienau, is with IRSID, Usinor Research Center, 57283 and exit of the rotary kiln. The furnace represented in Figure Maizie`res-le`s-Metz, Cedex, France. DENIS ABLITZER, Professor, is with 2 has an overflow feed at the entrance, while the exit is free, LSG2M, E ´ cole des Mines. JEAN-LE ´ ON HOUZELOT, Professor, is with as in the Carling kiln (France), which was used in the presentthe Laboratoire des Sciences du Ge´nie Chimique, E´ cole Nationale study as both a reference and a database.Supe´rieure des Industries Chimiques, 54001 Nancy, Cedex, France. Manuscript submitted January 18, 1999. We carried out specific investigations to gain a better METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 31B, APRIL 2000—391 Fig. 1—Trajectory of a grain in a rotary kiln. Fig. 3—Heat fluxes in a cross section. ts t (z) 5 8C« sin a(z) 2 a(z) 2 sin a(z) dgr dwi [5] Fig. 2—Kiln sections and nomenclature. where dgr is the average grain diameter and C« 5 !3 p6(1 2 «b) [6]understanding of the charge movements and to measure the charge flow parameters. Details of this work are given where «b is the intergranular porosity of the bed,elsewhere.[8] Only the results of direct interest for the present study will be described here. Thus, the mean residence time «b 5 1 2 ra rgr [7] was determined by using painted coal grains as tracers in a pilot furnace or grains of pozzolana in the industrial kiln. and is determined to be equal to 0.4, for both the coal andThe fraction of time spent by the grains on the bed surface the coke.was measured in the pilot furnace by filming the successive The use of Eqs. [1] through [7] has been validated byappearances and disappearances of the painted grains, while comparing their results to measurements made in a pilotthe axial profile of the charge was established with the aid furnace.[8]of an articulated feeler arm. This profile can be evaluated A final point concerning the charge flow is that the axialby integrating the following equation:[9] flow of the charge can be assimilated to a pure piston flow.[8] dH dz (z) 5 tan u1 cos b 2 12FV tan b vd 3wi 1 4H(z) dwi 2 12H(z)dwi 2 2 2 23/2 [1] III. HEAT TRANSFER Heat-transfer mechanisms play an essential role in thewhere u1 designates the inclination of the kiln with respect rotary kiln coal pyrolysis process. In terms of heat transfer,to the horizontal position, FV is the solid volume feed rate, the furnace can be considered to be a gas/solid counterflowv is the angular rotation velocity of the kiln, and dwi is its exchanger. In the main part of the kiln, the gas is the sourceinside diameter. The filling angle (a) is deduced from H by of energy, heating both the solid and the furnace wall. The latter acts as a regenerator, since, as it rotates, it is reheateda(z)5 2 cos21 11 2 2H(z)dwi 2 [2] in contact with the gas, then gives up part of this energy when it passes beneath the solid. Toward the end of the kiln, and the axial velocity of the charge (u) is then given by on the coke side, it is the charge which represents the source of energy and heats both the gas and the kiln wall. All along the kiln, part of the heat received by the wall is dissipatedu(z) 5 8Fm(a(z) 2 sin a(z)) d 2wira [3] to the external atmosphere. The heat transfer between the gas, the solid, and the wall where Fm is the mass flow rate of solid and ra is the apparent involves radiation, convection, and conduction (Figure 3).density of the solid bed. The mean residence time (t) of the Because of the high temperature of the gas, which is abovegrains in the kiln is obtained by integration: 1000 8C over a large part of the kiln, radiation from the gas to the solid (Frgs) represents a significant source of energy.t 5 eZ0 dzu(z) [4] Radiation from the gas to the wall (Frgw) and from the wall to the solid (Frws) must also be taken into account. Radiation from the wall to itself is negligible (Section III–D). Convec-Finally, the formula given by Hanrot[10] is used to calculate the average fraction of the residence time spent by the grains tive heat-transfer processes include the gas/wall (Fcgw) and gas/solid (Fcgs) exchanges. The regeneration processon the surface of the bed, 392—VOLUME 31B, APRIL 2000 METALLURGICAL AND MATERIALS TRANSACTIONS B involves conduction and radiation from the wall to the solid with which it is in contact (Fcws). The heat losses are due Frgw 5 Awu Ergw s (T 4g 2 T 4wu) with Awu 5 dwi1p 2 a22 [10]to conduction through the wall and dissipation in the external atmosphere by convection (Fcwa) and radiation (Frwa). Frws 5 AsErwss (T 4wu 2 T 4s) [11]This description of the heat exchanges is limited to those in a section perpendicular to the kiln axis. Apart from the where Fr is the radiative flux per unit length of the kiln; Aheat transported by the gas and the solid, longitudinal heat is the area per unit length; s is the Stefan’s constant; Tg , transfer can be neglected compared to that in the transverse Ts , and Twu are the temperatures of the gas, the charge surface,direction. The manner in which the different heat fluxes and the uncovered inside wall, respectively; and the coeffi- in Figure 3 are calculated is described subsequently. The cients Er are functions of the emissivities of the grains, objective is to consider a heat-transfer model which is suffi- the gas, and the wall. We calculated the values of these ciently complete to take into account all the phenomena coefficients using different correlations.[1,11,16–18] We com- mentioned previously, while remaining simple enough to be pared the quantity of heat radiated toward the charge calcu- readily used in the overall model of the rotary kiln. This lated from these values to that given by the more-rigorous part of the study makes use of the abundant results available model of Gorog et al.[12] The expressions given by in the literature on heat-transfer phenomena. Manitius,[1] Ergs 5 «g«s A. Radiation Phenomena Ergw 5 «g«w [12] Calculation of the heat fluxes exchanged by radiation Erws 5 «w«s(1 2 «g)inside the kiln requires a knowledge of the radiative proper- ties, particularly the emissivities, of the solid, the kiln wall, lead to the smallest difference (less than 5 pct) and were, and the gas, together with the choice of a radiation model. therefore, chosen in the present work.The emissivity of the coal grains («gr) during pyrolysis is The last radiation flux corresponds to the heat lost from thepoorly defined and difficult to determine, as discussed in Part external surface of the kiln to the surrounding atmosphere.I of this article.[6] An accurate knowledge of the emissivity of Considering a gray wall emitting into an infinite medium atthe refractory concrete inside the kiln wall («w) is also diffi- a temperature Ta givescult, since coal and coke tend to become attached and diffuse into the concrete. The values used were «gr 5 0.9 and Frwa 5 Awo«wos (T 4wo 2 T 4a) with Awo 5 pdwo [13] «w 5 0.9, and their influence on the results of the model is where the subscript wo designates the outer wall, which isshown in Section VI–A. The emissivity of the gas present generally heavily oxidized steel. The emissivity «wo is takenin the rotary kiln depends on its concentration in absorbant to be equal to 0.9.species such as H2O, CO2, CO, and CH4, together with the presence of dust. In the case of a H2O 1 CO2 mixture, the emissivity of the gas can be calculated from the formula[11] B. Convection «g 5 CCO2 «CO2 1 CH2O «H2O 2 D« [8] Convective heat exchanges occur between the gas and the surfaces over which it flows, i.e., the uncovered kiln wall and where the emissivities of each constituent («CO2 and «H2O) the bed surface, but, because of the over-riding importance of are calculated from their partial pressures, corrected by the radiative heat transfer at high temperatures, convection plays coefficients CCO2 and CH2O, which are functions of the total only a minor role in the present case. pressure, and where D« represents a correction for overlap- The Reynolds number for the gas flow is greater than ping of the spectral bands for the two gases. The contribution 40,000, indicating a turbulent regime. The heat-transfer coef- of gases other than CO2 and H2O to the emissivity of the ficient between the gas and the wall (hcgw) can, therefore,gas phase in the rotary kiln is small, due to their low concen- be calculated from the correlation[19] trations, except perhaps in the first few meters of the furnace. The contribution of dust is generally allowed for in the Nudwi 5 hcgwdwi lg 5 0.036 Re0.8dwi Pr 0.33 1dwiZ 2 0.055 [14]form of an additive emissivity term («dust), which is difficult to evaluate. which is valid in the turbulent regime when 10 , Z/dwi ,Among the models for radiation in rotary kilns proposed 400. Applied to the Carling kiln, this gives hcgw 'in the literature, the most comprehensive ones[12,13,14] give 10 W m22 K21. For the convective exchange between thea detailed description of the heat exchanges between the gas and the bed surface, experience shows that hcgs is proba-gas, the charge, and the walls by employing the so-called bly greater than hcgw. Barr[13,14] found a ratio of 1:2 betweenn-zone method.[11] They show that the majority of the radia- these two coefficients. The value taken in the present casetive heat transfer is localized in the cross section of the was hcgs 5 10 W m22 K21, and the influence of this parame-furnace, the heat flux between two surfaces separated by ter is shown in Section VI–A. The speed of rotation andan axial distance of more than one furnace diameter being the coal grain size do not seem to have a significant influencenegligible.[12] Different simplified approaches have been on convective heat transfer between the gas and thereported in the literature.[15] Most of the equations proposed solid.[20,21]are of the following form: Another convective heat exchange occurs at the outer kiln wall, which is cooled by natural convection in the sur-Frgs 5 AsErgs s (T 4g 2 T 4s) with As 5 dwi sin a 2 [9] rounding air, or by forced convection when there is a wind. METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 31B, APRIL 2000—393 The correlations proposed by Ozisik[22] gave a value of hcwa 5 0.26 W m22 K21 for natural convection and 3 # hcwa # 40 W m22 K21 for forced convection with a trans- verse wind speed between 1 and 28 m s21. A point value of hcwa was also determined experimentally from temperature measurements on the Carling kiln, which gave hcwa 5 17 W m22 K21, corresponding to a wind speed of 10.6 m s21. We considered hcwa to be a given, but uncertain, parame- ter. Its influence on the results is shown in Section VI–A. The heat fluxes corresponding to these three exchange mechanisms can be expressed by Fcgw 5 hcgwAwu(Tg 2 Twu) [15] Fcgs 5 hcgs As(Tg 2 Ts) [16] Fcwa 5 hcwaAwo(Two 2 Ta) [17] Fig. 4—Radial temperature profiles in the wall at various angular positions, at the coal end of the kiln.C. Contact Between the Covered Wall and the Solid In the lower part of the bed, the movement of the grains is imposed by the kiln wall. Heat transfer, therefore, occurs conducting material on the outside (2 cm of steel). Theessentially by conduction and radiation. Using an overall thermal resistance of the steel is about 300 times lower thanheat-transfer coefficient between the covered wall and the that of the insulating layer, so that, for the purposes of thesolid (hcws), the heat flux transferred per unit length is calculation, only the refractory concrete need be considered.given by The thermophysical properties of the concrete were deter- mined as follows: rw 5 2230 kg m23, cpw 5 1150 J kg21Fcws 5 hcwsAwc(Twc 2 Ts) with Awc 5 adwi 2 [18] K21, and lw 5 1.6 W m21 K21, by density measurements, differential calorimetry, and the laser-flash method, respec- Numerous models, based on the theory of heat penetration, tively.[24] These values vary little with temperature above lead to an equation of the type 200 8C. For a kiln operating in the steady-state regime, the temper- hcws 5 2leff !paefftcws [19] ature profile within the wall is constant for an observer on the ground. However, for a reference point in the wall, the profile can be considered to be periodic, since a given pointwhere leff and aeff are, respectively, the effective thermal in the wall returns to the same temperature every revolution.conductivity and diffusivity of the bed of grains, and tcws is It is the latter representation that is used for the presentthe contact time between the wall and the charge. This calculation. The temperature variation is considered in aformula gives values that are high compared to the measure- sector of wall of volume r Dr Dw Dz in the transient regime,ments and must be modified by introducing a contact resis- and the calculation is continued until the periodic behaviortance between the wall and the grains.[13] Calculation of this is reached.resistance requires a knowledge of numerous parameters, Figure 4 illustrates the shape of the temperature profileand the approach is difficult to apply in practice. Based on calculated in the wall at different angular positions for aa dimensionless form of Eq. [19] together with available charge temperature of 20 8C and a gas temperature of 910experimental results, Tsheng has established a new 8C, corresponding to the values at the kiln entrance on thecorrelation,[20] coal side. It is seen that only the first 15 mm of the wall are significantly affected by the temperature oscillations. Ithcws 5 11.6 leff Awc Go0.3 with Go 5 pvad 2wi 2aeff [20] should also be noted that the temperature variations at the surface of the uncovered wall do not exceed 75 8C and are In all cases, calculation of hcws requires a knowledge of small compared to Tg 2 Twu and Twu 2 Ts. That is why the the effective conductivity of the charge, which, in the present radiation flux from the wall to itself can be neglected in case, is determined from the correlation of Zehner and comparison to Frgw and Frws. Schlu¨nder.[23] IV. COMBUSTION PHENOMENA D. Heat Transfer in the Kiln Wall The combustion phenomena that occur in the rotary kiln coal pyrolysis process merit a detailed study, since theyHeat is transported through the wall by conduction. In order to obtain the fluxes exchanged on the inside with the represent the sole source of energy. They naturally include the combustion of the volatile species produced by pyrolysisgas and the solid and with the atmosphere on the outside, it is essential to calculate the temperature profile in the wall. of the coal, but also include the combustion of the coke formed in the bed, together with that of the dust generatedThe wall of a pyrolysis furnace is composed of a thick layer of insulating material on the inside (20 cm of refractory throughout the length of the furnace. The volatile species are essentially of two types, namely,concrete