Experiments and simulations of gas-solid flow in an airlift loop reactor

Particuology 9(2011)130-138Contents lists available at ScienceDirectParticuologyELSEVIERjournalhomepagewww.elsevier.com/locate/particExperiments and simulations of gas-solid flow in an airlift loop reactorChaoyu Yan, Chunxi Lu*, Yiping Fan, Rui Cao, Yansheng liuState Key laboratory of Heavy oil Processing China University of petroleum, Beijing 102249, chinaARTICLE INFOABSTRACTThe hydrodynamics in a gas-solid airlift loop reactor was investigated systematically usiReceived 3 March 2009measurements and CFD simulation, In the experiments, the time averaged parameters, such as solid frac--ceived in revised formtion and particle velocity, were measured tfiber probe. In the simulation, the modified gidaspowptember 2009Accepted 100ctober 2009drag model accounting for the interparticle clustering was incorporated into the Eulerian-Eulerian CFDodel with particulate-phase kinetic theory Predicted values of solid fraction and particle velocity werewith experimental results, validating the drag model and the simulation. the results showAirlift loop reactorthat the profiles of particle velocity and solid fraction are uniform in annulus. However, the core-annulusstructure appears in other three regions (draft tube region, bottom region and particle diffluence region).as-solid two phases flowhich presents the similar heterogeneous feature of aggregative fluidization usually occurred in nomal fluidized beds. Simulated profiles of particle residence time distribution indicate that the airlift loopDrag force modelreactor should be characterized by near perfect mixing.o 2011 Chinese Society of Particuology and Institute of Process Engineering Chinese Academy ofSciences. Published by Elsevier B V. All rights reservedcirculating system of the fluidized bed mainly consists of a denseALR section and a dilute riser section, as shown in fig. 1. The alrAirlift loop reactor(ALR)is an important reaction device, and section is a key part of the coupled gas-solid fluidized bed For thehas been widely used in many process industries with multiphase design and optimization of the coupled gas-solid fluidized bed, aflow, such as gas-liquid( Kemblowski, Przywarski, Diab, 1993), better understanding of its hydrodynamics is requiredliquid-solid( Wang, Xu, Gao, Lin, 2003)or gas-liquid-solid sys-With the increase of computational power, the numerical simu-tem( Hwang Cheng, 1997). However literature about ALR of lation has become an additional tool to predict some microcosmicgas-solid system is few. In recent years, only several researchers fluid dynamics and transport phenomena in multiphase flowshave developed a type of gas-solid ALR system on the basis of the which are difficult to be revealed completely by current experi-ories of gas- liquid ALR and conducted a series of experiments on mental measuring technique. Hence, computational fluid dynamicsgas-solid flow( Liu Lu, 2001: Liu, Lu,& Shi, 2002, 2004: Zhang, Lu,(CFD) can offer a new approach to understanding the complex& Shi, 2004). This gas-solid aLr has been successfully used as strip- phenomena that occur between the gas and particulate phases. Inper in fluid catalytic cracking unit( FCCU)to remove oil gas from recent years, many researchers( Cooper Coronella, 2005: Helland,deactivated catalyst or non-condensable flue gas from regenerated Occelli, Tadrist, 2002: Johansson, van Wachem, Almstedtcatalyst(Lu, Xu, Lu, &Shi, 2002). The gas-solid ALR system has some 2006: Lu, Zhao, Ding, Gidaspow, Li, 2007: Pain, Mansoorzadehsimilar hydrodynamic characteristics of gas-liquid ALR, as well as Gomes, de oliveira, 2002: Peirano, Delloume, Leckner, 2001that of usual gas-solid fluidized beds. Therefore, the gas-solid ALR Taghipour Ellis, Wong, 2005; van Wachem, Schouten, Krishna,system has gradually become a research branch and evolved into a van den bleek, 1999: Witt, Perry, Schwarz, 1998: Zimmermannbroadened application of conventional ALR.Taghipour, 2005)have carried out simulation studies on theIn the study, the application of gas-solid ALR system is mainly hydrodynamics of gas-solid bubbling bed or turbulent bed bymed at the background of the petroleum coke combustion pro- Eulerian-Eulerian two-fluid model. Brandani and Zhang(2006)cess. Based on the combustion characteristics of petroleum coke, a have proposed a new model to predict the behavior of fluidizedcoupled gas-solid fluidized bed combustor is proposed in the pre- beds based on the normal Eulerian-Eulerian two-fluid model. Thevious study of this work(Yan, Lu, Liu, Cao, Shi, 2009). The overall inter-phase drag force is a dominant force in gas-solid fluidizedLawr to *he successful simulation ofthe hyd中国煤化工 re found that the CfD sim∵m邮盐11089733803;fax+861085lateCNMHGaversal drag laws, such asGidasplay overestimate the bed2011 Chinese Society of Particuology and lnstitute of Process Engineering Chinese Academy of Sciences. Published by Elsevier B V. All rights reserved.C Yan et aL/ Particuology 9(2011)130-138Nomenclatureconcentration of tracer or bed material in solid mix-ture phaseqt) cross-sectionally averaged tracer concentrationdrag coefficientdiameter of particle(um)dpe effective mean diameter of particle cluster(um)Ddiffusion coefficient of tracer or bed material in solid1. Air blowermixture phase( m2/s)2. Surge tarrestitution coefficient3. Air rotameterE(r) density function of tracer residence time distribu4. Draft tubetion(saxial height(m)5. Airlift loop sectiontracer sampling time(s)6. Riser sectiontracer mean retention time(s)7. Horizontal separatorg acceleration of gravity(m/s2)8, 10. Cyclone separatorradial distribution functionparticle circulating flux(kg/(m2s)9, 13. Butterfly valvek kinetics energy(m2/s2)Il. Storage tank12. Particle returning piperadial position(m)inner radius of the alr column(m)Rep Reynolds number of particlessource itemugd superficial gas velocity in draft tube( m/s)superficial gas velocity in annulus(m/s)up particle velocity(m/s)Vg gas phase velocity(m/s)solid phase velocity(m/sij, k direction coordinatebed voidage1. Schematic diagram of the experimental apparasolid fractionnter-phase momentum exchange coefficientdissipation of energyon(kg/(ms))the concurrent-up gas-solid flow in a riser Lettieri, Newton, anddiffusion coefficient for the energy fluctuation Yates(2002)found that, for FCC catalyst in a fluidized bed,the(kg/(ms))experimental terminal velocity(u)was much higher than thesolid bulk viscosity(Pas)culated one(ut), and the ratio of u /ut increased with increasinge granular temperature(m2/s2temperature. If a homogeneous bed was assumed to be characsolid phase shear viscosity(Pas)terized by the presence of particle clusters, the effective diameterPg density of gas phase(kg/m)of clusters, dp. can be back-calculated from the experimentalvalues. They also figured out the effective mean particle clusterPp particle density (kg/m)stress tensor(Pa)diameters in the range of 200 to 474 um for FCC catalysts withSauter mean diameters from 49 to 71 um. Gao, Chang, Xu, Lan,and Yang(2008)thought that the cohesive inter-particle force duesolid phaseto the van der waals attraction was responsible for aggleon of particles, which led to a reduced drag force and kept thesteady fluidization of bubbling bed with FCC particles in the realgas phaseoperation Based on the experimental terminal velocity of FCC cat-alyst, Gao, Chang, Xu, et al. (2008 )and Zhang(2004) suggestedthat the effective mean particle cluster diameter of 265 um shouldexpansion in dense bubbling beds or turbulent beds of Geldart a be used to replace the actual 58 um for FCC particles to modifyparticles. McKeen and Pugsley(2003)reported that the generally the Gidaspow drag force model in the CFD simulation of gas solidpoor simulation results for Geldart A particles could be attributed to flow in bubbling strippers. Cao(2006)also proposed a modifiedthe existence of a significant inter-particle force that was neglected Gidaspow drag force model, which used a piecewise function toin most simulations. The existence of particle clusters results in calculafirient There were four intervals based onlarger effective particle sizes and hence reduces fluid-particle drag differe中国煤化工 del. i.e, dense phase, subforces, which in turn lead to reduced bed expansion. On the basis of denseasethe EMMS approach(Ge & 4, 2002: Liet al, 1999)and the concept and tiC NMH Gticle cluster diameter wasof particle cluster, Yang Wang, Ge, and Li(2003a, 2003b)proposed used to replace the mean particle diameter, but in the dilute anda drag model of structure-dependent drag coefficients, which was dispersed phases where particle clustering is not important, theincorporated into the two-fluid model to simulate the behavior of mean particle diameter was usedTable 1cess Engineering, Chinese Academy of Sciences, Beijing, China. ThePhysical property of sand particles.optical fiber probe was placed at different axial and radial positionsin the four regions. Fig. 2 also shows schematically the optical fiberParticle density(kg/m)probe and the axial measurement points in the ALr section. TheAerated bulk density(kg/m)radial measurement positions of the optical fiber probe are listedRepose anglin Table 2In this work, a modified Gidaspow drag model was proposedon the basis of an effective mean diameter of particle clusters. By Eulerian-Eulerian two fluid model( Fluent 6.2.16, 2005), consist-model with particulate phase kinetic theory was established. The ing of a set of momentum and continuity equations for each phasegas-solid flow behavior in the alr section of the coupled fluidized which are linked through pressure and inter-phase exchange coet-bed was investigated systematically by combining the CFD simu- the kinetic theory of granular flows(Gidaspow, 1994: Sinclairlation with experimental validation. The flow parameters, particlevelocity and solid fraction, can help to give insight into the particle Jackson, 1989).trinsic flow characteristics in the alr section.3. 1. Governing equationsZ, ExperimentalThe governing equations of the system include the conservationFig 1 shows the cold model experimental apparatus of the cou-of mass and momentum. due to no chemical reactions and masspled coke combustor made of plexiglass. The overall circulatingtransfer between each phase, the accumulation of mass in eachsystem of the fluidized bed mainly consisted of a dense alr secphase is balanced by convective mass fluxes. The continuity equa-tion and a dilute riser section. the ALR section had a 0.476m inner tion for gas phase, with the relevant variables denoted by subscriptdiameter( L.D. )and 2.5 m high. A draft tube of lD 0.33 m and height1.3 installed in the center of the ALR Section. The riser section aegpg)a2+(a24)=0All experiments were carried out at the atmospheric pressureand the ambient temperature of approximately 25C. The solid par- The continuity equation for particulate phase, with the releyticles employed in this study were sand. The physical properties of variables denoted by subscript p or s, is written asthe sand particles are listed in Table 1. The superficial gas velocityranged from 0.772 to 1.674 m/s in the draft tube, and from 0. 223 to EsPp+ ax, (Es Ppupi)=00.519 m/s in the annulus. The corresponding superficial gas veloc-ity in the riser section ranged from 3. 156 to 5.989 m/ s. The overalsolids circulating flux with respect to the cross-sectional area of theEach computational cell is shared by the interpenetratingriser section varied between 40.8 and 229.4 kg/(m2 s). The gas-solidphases, so that the sum over all volume fractions is unity.flow characteristics were bubbling or turbulent fluidization in the eg+Es=1ALR section depending on the superficial gas velocity, but fast flu-idization in the riser section.The conservation of momentum for the gas phase is describedAccording to theexperimental results of our previous study (Yan byet al., 2009). the whole space of the ALR section can be divided intodiffluence region, as shown in Fig. 2.改凹g-up)+pngp8The particle velocity and solid faction in the aLR section wereynchronously measured by using optical fiber probe(E. Lu, Xu,Gao, Shi, 2003; Yan et al. 2009)produced by the Institute of Pro-The conservation of momentum for the particulate phase cane expressed as(ep风u-y)+psThe conservation of the kinetic energy of the moving particlesis described as follows by the grarature, e, which isderived from the kinetic theory of granular flow:中国煤化工CNMHGFig. 2. Schematic diagram of optical fiber probe, axial measurement points and +updivision of four flow regions in the ALR: (i)draft tube, I)annulus, ( lll)bottom and(M) particle diffluenceC Yan et aL/Particuology 9(2011)130-138Radial measurement positions of the optical fiber probe (inner radius of the airlift loop column, R- 238 mm).0609065107270811Particle diffluence region0.126025206300.7560924where e is granular temperature, defined by3.3, Drag lawsIn this work, a modified Gidaspow drag force model Cao, 2006Gao, Chang, Lu, Xu, 2008 )is adapted to simulate the hydrody-32. Constitutive equationsnamics in the alr the model is shown as follows:Constitutive relations required to close the governing equations B==CD3. EsegPgVs-hen0991000The solid pressure, which describes the change in the totalmomentum transport of the motion of particles and their inter- ResEgPgdplVg-vskactions, is used for the pressure gradient term and is expressedand dp is the single Sauter mean diameter of 78 um for the sandPp=Espp[1+2(1 +e)Esgo]eability of particle collisions when the particulate phase becomes B-10. doell -gs ).93, when 81000Particulate phase shear viscosity is expressed as+81++」+32s+eN12)b≈-团Particulate phase dilute viscosity is expressed asand dpe is the effective mean particle cluster diameter:13)B=1041-a3)k+171-, when eg≤0.8(21)The solid bulk viscosity, which accounts for the resistance of thesolid phase to compression and expansion, is expressed byIn this study, an effective particle cluster diameter of 125 umwas adopted in the drag force model, which was determined by5=hp81(14) comparison between simulated and experimental results. The val-idation of the drag force model of Eqs. (18)-(21)and the influenceThe collision dissipation of energy.esenting the rate of of effective particle cluster diameter on simulation results are dis-energy dissipation within the particulate phase due to inelastic cussed in Section 5.1particle collisions, is calculated fromy=3(1-2)2|4V4.1. Simulation code and numerical algorithmThe diffusion coefficient for the particulate phase energy flu*io fferential equations weretuation is shown assolved中国煤化工 -ese equations were dis2Tedl1+5(1+elgoes+2ePpdpBo(1+eVT(16)usedcreteC N MH Ge over the finite volumeCFD software code Flu-ent 6. 2. 16. The Phase-Coupled SIMPLE algorithm was applied(17for the pressure-velocity coupling. Second-order upwindcretization schemes for the convection terms were used theC Yan et al. /Particuology 9(2011)130-138under-relaxation factors for pressure, momentum, volume fration and granular temperature used in the iterations of modelOutlet of gas-solid two phasesresolution were 0.3, 0.7, 0.2 and 0.2, respectively. The restitu-tion coefficients between particle-particle and particle-wall wereboth 0.9. A small time step(0.001 s)with around 20 iterationsper time step was chosen, until convergence was reached. Aconvergence criterion of 10- for each scaled residual compo-nent was specified for the relative error between two successiveiterations In order to ensure that the simulation duration waslong enough to establish the desired operation conditions,atotal simulation time much greater than the mean gas residencetime was selected as 30s. The time-averaged distributions ofparticle flow variables. the solid fraction and the particle velocity, were computed covering a period of the last 20s in thiswork4.2. Simulation systemsSolid phase inletThe detailed structure of the simulated ALR is shown inFigs. 1 and 2. In the study, the two-dimensional grids werecreated in a CAD program called GAMBIT 2.2.30 and exportednto Fluent 6. 16. The grid size of 5mm x 5mm is equidistantin both horizontal and vertical directions in the alr, as shownin Fig. 3.Gas phase inlet in annulus4.3. Boundary conditionsGas phase inlet in draft tubeFig 3 shows the boundaries of the simulation domain The inletsFig 3. Simulation domain and gridof the gas phase and the particulate phase are designated as velocity Inlets in Fluent 6.2.16, where the direction of gas or particleflow is normal to the boundary surface. Depending on the inlet 5. Results and discussionconfiguration, the gas inlet velocity equals the superficial gas velocity in the draft tube or the annulus. The particle inlet velocity can 5.1. Validation of drag model and determination of effecivee determined from the particle flux in the coupled fluidized bed particle cluster diametersystem. The top outlet of the alr was set as an outflow boundary condition for the gas and the particulate phases. Wherever theFig 4 displays the comparison of solid fraction contours for dif-boundary was specified as wall, no-slip boundary condition was ferent effective particle cluster diameters under the same operationappliedconditions and at the same time of 7 0s. Fig 4(a)is the simulationThe gas and the particulate phases in the simulation domain result by using real particle diameter of 78 um and Gidaspow dragwere the same as in the experiments. In order to validate the sim- force model, showing that the bed expansion is overestimated andulation results, the simulation conditions were consistent with the the fluidization behaves like a fast fluidized bed, which is inconsis487e04626014146013.89e013.6501341。021e01195073002中国煤化工243·02CNMHG(a)4-78 um (b)p =100 um (e)do =125 ur (a)ahe - Isu um (e)ape-ZU0 umFig 4. Comparison of solid fraction contours for different effective particle diameters at uxd-0901 m/s, uy-0223 m/s, C, -45.7 kg (m2s)and t-7.0s for(a)Gidaspow dragforce model, (bHe) Drag force model of Eqs. (18)-(21).C Yan et aL /Particuology 9(2011)130-138are consistent, but a certain deviation exists. fig. 6(a)shows thatin the draft tube, the local solid fraction in the center region is0.6()d-100 um, Drag force model EAs (18H21)ree model Eas( IsH2I)lower than that in the region close to the wall. The radial non-)415μ, Dns force model(18a21uniform distribution of local solid fraction can be viewed as a core-d=200 um, Dn force modl Eqs(I&annulus distribution similar to that in normal fluidized beds thecore-annulus distribution of local solid fraction is related to thenonuniform radial distribution of the particulate phase velocity, asshown in Fig. 6(b)In ann0.1Fig 7 represents the exntal and simulation results of the091solid fraction and particle velocity in the annulus at a given condi--0.6040.2000.2040.6tion Fig. 7(a)shows that the radial distribution of the solid fractionin the annulus is relatively uniform. But in the regions close to theexternal wall of the draft tube and the internal wall of the alr coFig. 5. Comparison of solid fractions for different effective particle diameters at umn, due to the wall effects, simulation results show that the valueupd-0901 m/s, wu-0223 m/s, G, 45.7kg/(ms)of the solid fraction is higher than that in the center region of theannulus. Simulation result shown in Fig. 7(b) demonstrates that theparticles in the annulus have a downward velocity, and the radialThe influence ofeffective particle cluster diameter on numerical distribution of the particle velocity is uniform.results is shown in Fig. 4(b)-(e) when employing the drag forcemodel of Eqs. (18)-(21)in this work. The whole flow field basically 5.2.3. in bottom regionaccords with the experimental phenomenon. there exists a denseFig 8 shows the experimental and simulation results of the solidphase zone in the lower part of the reactor and a dilute phase zone fraction and particle velocity in the bottom region at a given con-in the upper part. However, the dense phase bed level increases dition, indicating that the radial distributions of the solid fractiongradually with increasing the effective particle cluster diameter. and the particle velocity in the region(r/R s0.75), i. e just belowis figure demonstrates that it is the reasonable effective particle the draft tube are relatively uniform, but change abruptly in thecluster diameter that can be used to describe the true flow field region(0. 75< r/R)<1.0), ie below the annulus or close to the wallcorrectly.of the alr columnFig. 5 gives out the quantitative comparison of solid fractionIn the region(r/rI<0.75)below the draft tube, theprofiles for different effective particle cluster diameters with the bles arising from the gas distributor are cut continually by theexperimental data. The simulation result using real particle diam- particles moving laterally from the annulus bottom. The sizeeter of 78 um deviates from the experimental data. Among the of gas bubbles in this region is thus decreased and the mixingsimulation results using effective particle cluster diameters, the between the gas bubbles and the particles is largely enhanced.effective particle cluster diameter 125 um can simulate the true resulting in more uniform radial distributions of the solid frac-flow field of the reactor most reasonably. Thus, the effective particle tion and the particle velocity in this region. In the region ofcluster diameter 125 ]m was chosen in the overall CFD simulation 0.75< r/R]<1.0. however, the gas flow rate arising from the dis-of this workributor is relatively lower than that from the distributor justbelow the draft tube, causing that the solid fraction increases5. 2. Profiles of solid fraction and particle velocityand the particle velocity decreases. Besides, due to the waleffect the particle aggregative tendency in this region is intensified,5.2.1. In draft tubealso leading to the increase in solid fraction.Fig 6 shows the experimental and simulation results of the solidfraction and particle velocity in the draft tube at a given condition. 5.2.4. In particle diffiuence regionindicating that the experimental data and the simulation results ofFig 9 represents the experimental and simulation results ofthe solid fraction fit well, while for particle velocity, the tendencies the solid fraction and particle velocity in the particle diffluencea0.60.9m040.30.1中国煤化工0.0CNMH-0.604-0.20.00.20.40.60.4T/RFig6 Profiles of solid fraction(a)and particle velocity(b)in draft tube at uja 0.901 m/s, uu".223 m/s and G,-457kg/(ms).C Yan et aL Particuology 9(2011)130-138a0.6b-0.1u0.3hs:0.2 m040.700.750800.850900951.000.700.750.800.850.900951.00r/RT/RFlg.7. Profiles of solid fraction(a)and particle velocity(b)in annulus at und-0901 m/s, ug-0223 m/s and Gs45.7kg/(ms)0.5h0.116m0-1below the draft tubebelow the draft tube0.51.0-0.505r/RFig &. Profiles of solid fraction (a)and partice velocity(b)in bottom region at usd0.901 m/s, uu=0.223 m/s and G, 45.7kg/(m s).region at a given condition, showing that the radial distributions produced by the gas bubbles breaking up. the particlesof solid fraction and particle velocity in the region above the draft outward also impose a shear action on the gas bubbles intube(r/R s0.75)are relatively uniform, but change abruptly in fluence region, leading further to decrease in gas bubblethe region above the annulus or close the wall of the alr column enhancement of the mixing between the gas bubbles and the par0.75<|r/R|<10)ticles. hence the radial distributions of the solid fraction and theGas bubbles moving upward from the draft tube gradually coa- particle velocity in the region above the draft tube (Ir/R]s0.75)lesce, and then ceaselessly break up into small bubbles in the are uniform macroscopically In the region above the annulusparticle diffluence region. Particles in the particle diffluence region (0.75< r/R] <1.0), the heterogenic radial distributions of the solidmove outward into the annulus under the combined action of the fraction and the particle velocity can also be attributed to the wallinertial force and the gravity itself, as well as the ejection force effect.a0.6h2.0mb20m theu0.3above the draft tube中国煤化工1.00.50.00.51.0CNMHG05 1.0idle velocity(b) in particie diffluence region at sd-0901 m/s, u0. m/s and G45.7kgM(m?s)C Yan et aL Particuology 9(2011)130-138a0.020b0.020u,(m/s)G, (kg/(ms))u,(m/s)G(kg/(m0016141614090016:0.223m/su.=0223ms20.012001200080.00820406080100100Fig. 10. Effect of operation conditions on particle residence time distribution in the Aur (a) superficial gas velocity;(b)solid particle circulating flux.5.3. Particle residence time distributionFig. 10 shows that variation of the superficial gas velocity and orthe particle circulating flux does not obviously change the shape of5.3.1. Simulation methodthe particle residence time distribution curve. However the peakParticle residence time distribution in the alr was simulated height increases and appears earlier with increasing the superficialusing the numerical impulse tracer method. After the steady flow gas velocity or the particle circulating flux.fields of the gas-solid two phases had been established based onthe Eulerian-Eulerian simulation method, the tracer particles with 6. Conclusionshe same physical properties as the bed material were injectedinto the flow field at the particulate phase inlet boundary as an (1) The hydrodynamics in the gas-solid ALR was investigatedimpulse. The tracer particles together with the bed material formsystematically using experimental measurements and CFDa solid particulate mixture phase in the simulation. the numesimulation incorporating the modified Gidaspow drag modelical tracer evolution complies with the impulse tracer function.to take into account the formation of particle clusters. Pre-The tracer transient concentration in the flow field was obtaineddicted values of solid fraction and particle velocity are basicallyby solving the following composition governing equation( Fluentin agreement with experimental results, validating the drag6216,2005)+(0x)=(a()(2)The profiles of particle velocity and solid fraction are uniform22)in the annulus. However the core-annulusin other three regions(draft tube region, bottom region andA time series of the cross-sectionally averaged tracer concenparticle diffluence region), which presents the similar hetero.ation, C(t). at the outlet boundary was then acquired by using thegeneous feature of aggregative fluidization usually occurring inFluent software. Furthermore, the density function ofthe tracerres-normal fluidized bedsidence time distribution, E(c)(- ), can be determined as follows (3)Simulated particle residence time distributions indicate that(Fogler, 1999):the alr can be characterized by near perfect mixingE(e=c(o)(23)Acknowledgementserefore the tracer mean retention time m, can be obtained The authors acknowledge the supports by the National Naturalthe following equation( Fogler, 1999):Science Foundation of China(Grant Nos. 20806090, 20976190 and20776155)tE(rdtE(tdtReferencesFig. 10 displays the particle residence time distributions in the Cao, B(2006 Study on the fow behow l6 80o-gr the prediction of the behaviour of53. 2. Simulation resultsBrandani, S& Zhang, K(2006)AALR at different operation conditions, The curve of the particle res- ference mixing particles. Unpublished doctoral dissertation. China University ofidence time distribution is composed of an early broad peak and a cooper S. Coronella. c._(2005) CFD simulations of particle mixing in a binarylong wave trailing, indicating that the ALR can be characterized bya near perfectly mixed reactor. The particle residence time distri- EC, Lu, C. Xu, C, Gao, J,& shi, M( 2003). A new method for measurement of localbution is related to the structure of the ALR and the gas-solid flowolid flux in gas-solid two-phase flow. Chinese Journal of Chemical Engineering.behavior in the reactor. The broad peak of the particle residence Fluent.(2005). Fluent 6.2.16, user's guide Fluent Inctime distribution curve demonstrates that a part of particles enter- Foglergineering(3rd ed. ) New jersey:ing in the reactor leave the reactor directly in a short time due to中国煤化工the entrainment of the gas phase. However, the long wave trailing Gary-al and computational studies onof the curve indicates that the other particles have relatively longCN MHGparately sized binary particles.retention time due to the particle loop flow and back-mixing in the Gao, J- Chang. J- Xu,C Lan, x. Yang. Y(2008) CFD simulation of gas solid flotngineering Science, 63, 1827particularly structured reactor Eq (24)gave a particle mean reten- Ge, w. Li,j( 2002) Physical mapping of fluidization regimes-The EMMStion time of about 45 s in the alr under the simulation conditionsapproach. Chemical Engineering Science, 57, 3993-4004138C Yan et aL Particuology 9(2011)130-138Gidaspow,D(1994). Multiphase flow and fluidization: Continuum and kinetic theory Pain, C C, Mansoorzadeh, S, Gomes, J LM. & de oliveira, CRE(2002).Anumericalestigation of bubbling gas-solid fluidized bed dynamics in 2-D geometries.lli. R& Tadrist, L(2002)Computational study of fluctuatingPowder Technology, 128, 56-77.and cluster structures in gas-particle flows. Intemational Jourmal of Peirano, E Delloume. V, Leckner, B (2001) Two-or three-dimensional simulaons of turbulent gas-solid flows applied to fluidization Chemical EngineeringHwang. S.h,& Cheng. Y. L(1997) Gas holdup and liquid velocity in threephase internal-loop airlift reactors. Chemical Engineering Science, 52(21/22). Sinclair. J. L,& Jackson, R(1989) Gas-particle flow in a vertical pipe with3949-3960Johansson, K. van Wachem, B G M. Almstedt, ATaghipour, F. Ellis, N, Wong. C(2005). Experimental and computational studytion of CFD models for fluidized beds: Influence of particle stress models, gasid fluidized bed hydrodynamics. Chemical Engineering Science, 60,gineering Science, 61.van Sachem, B Gten. J. C. Krishna, R van den Bleek. C M. (1999).Kemblowski, Z, Przywarski, J, Diab, A. 1993). An average gas hold-up and liquidValidation of the eulerian simulatedbehaviour of gas-solid fluidisedcirculation velocity in airlift reactors with external loop. Chemical EngineeringWang, Y, Xu, C, Gao, L, Lin, S(2003)on of solid flow rate and loading inLettieri. P, Newton, D& Yates, ]. G (2002). Homogeneous bed expansion of FCCts, influence of temperature on the parameters ofLi, J, Cheng. C Zhang Z, Yuan, ), Nemet. A, Fett F (1999). The EMMS model-Itsbubble formation in a 3D fluidized bed. Applied Mathemoodel for predictinwitt, P. J. Perry, J H, Schwarz. M. P (1998)A numerical1071-1080.Yan, C, Lu, C Liu, Y. Cao, R Shi, M (2009). HydrodynamLiu, M,&Lu, C(2001). Study of a new kind of stripper Petrochemical Technology.of petroleum coke combustor. Powder Technology, 192, 143-15111.850-854(n Chinese).Yang, N. Wang w Ge, w.& Li, J(2003a). CFD simulation of concurrent-uplid flow in circulating fluidized beds with structure-dependent drag coevel gas-solid annular flow stripper Petrochemical Technolog 31(10).8in ChineselLiu, M, Lu, C.& Shi, M.(2004) Hydrodynamic behaviour of a gas-solid air-loopfficient in modeling gas-solid two-phase flow. China Particuology. 1(1).Chinese Journal of Chemical Engineering. 121) 55-59Lu, C, Xu, G Lu, S, Shi, M(2002). Study and industry application of a pre- Zhang P(2004). Numerical simulation of hydrodymamic and coke combustionsin FCC regenerators. Master thesis. China University of Petroleum, BeijingR=m1说哪ammm③m可mngation of mix. Zhang. Y Lu, C,& Shi, M. (2004), Large-scale cold pilot experiment on a new annul-McKeen,T,& Pugsley, T. S(2003) Simulation and experimental validation Zimmermann, S& Taghipour, F (2005). CFD modeling of the hydrodynamics andof a freely bubbling bed of FCC catalyst. Powder Technology, 129, 139-reaction kinetics of FCC fluidized-bed reactors, industral and Engineering Chem152istry Research, 44, 9818-982中国煤化工CNMHG