Spray Drying in Cyclone Separators

1 Spray Drying in Cyclone SeparatorsThesis for the Degree of Master of Science in Chemical EngineeringNiklas PerssonDepartment of Chemical EngineeringDivision of Chemical Reaction EngineeringChalmers University of TechnologyGothenburg, Sweden 2014 Thesis for the Degree of Master of ScienceSpray Drying in Cyclone SeparatorsNiklas PerssonExaminer:Bengt AnderssonSupervisor:Mohammad El-AltiDepartment of Chemical EngineeringChalmers University of TechnologyGothenburg, Sweden 2014 Spray Drying in Cyclone SeparatorsNiklas Perssonc Niklas Persson, 2014 Department of Chemical EngineeringChalmers University of TechnologySE-412 96 G oteborgSwedenTelephone + 46 (0)31-772 1000 Gothenburg, Sweden 2014 AbstractA gas-solid Cyclone separator is a separation device that separates solid particles from a gas phaseby the use of a centrifugal force field. In traditional Spray Drying a Cyclone separator is oftenincluded in a succeeding separation step, after a Spray Drying chamber.

2 This thesis a study ofwhether Spray Drying in Cyclone Separators is possible. If the Spray Drying can be conducted in thecyclone, the Spray Drying chamber can be removed from process and the investment cost flow field in the studied Cyclone was simulated in both two and three dimensions, with goodagreement in the middle of the Cyclone , where the region of interest is located. Single particlesimulations, together with hand calculations and a factorial designed parameter study showedthat Spray Drying is possible in the studied Cyclone . However, for lower gas velocities than typicaloperational conditions and for droplets in the order of magnitude of a normalized diameter and smaller, where the droplet diameter is normalized against the largest droplets in thecurrent system. By increasing the normalized Cyclone diameter from to , droplets up tothe size of can be dried before impinging the wall.

3 The results were in compliance thesis has been performed at ALTEN AB and Chalmers University of Technology inG oteborg, Sweden, during the period September to December 2013. The work has been su-pervised by Dr. Mohammad El-Alti at ALTEN AB, Professor Bengt Andersson and AssistantProfessor Ronnie Andersson at Chalmers University of wish to acknowledge the help provided by my supervisors Dr. Mohammad El-Alti, ProfessorBengt Andersson and Assistant Professor Ronnie Andersson for their help and guidance through-out the but not least I would like to thank the members of the analysis and simulation group atALTEN AB and my thesis colleague Erik Eddy SimulationNSNavier Stokes EquationsPV CPrecessing Vortex CoreRNGM athematical Renormalization-group techniqueRSMR eynolds Stress ModelUDFUser Defined Function]RANS Reynolds Decomposed and Averaged Navier Stokes EquationsGreek Symbols Loading Dirac Delta Numerical Error Thermal conductivity [W/mK] Von K arm an Constant Surface Roughness [-]EDissipation of Turbulent Kinetic Energy [J/kgs=m2/kg3] Kinematic Viscosity [m2/s] TTurbulent Viscosity [Pas] Density [kg/m3] Surface Tension [N/m] kPrandtl-Schmidt Number Characteristic Time [s] ijStress [Pa] Angle [ ]Roman SymbolsAdProjected Area [m2]iCDDrag Coefficient [-]CpHeat Capacity [J/kgK]

4 DDispersed PhaseDi,mDiffusivity Coefficient for Specie i [m2/s]FForce [N]hConvective heat transfer Coefficient [W/m2K]hfgHeat of Evaporation [J/kg]i,j,kDirectionkTurbulent Kinetic Energy, [J/kg=m2/s2]kcMass transfer Coefficient [m/s]MMolality [mol/kg]mMass [kg]NFlux [Mol/m2s]nMoleNuNusselt Number [-]PPressure [Pa]pParticleP0 Vapor Pressure [Pas]PrPrandtl Number [-]RUniversal Gas Constant [J/molK]rRadius [m]ReReynolds Number [-]ScSchmidt Number [-]StStokes Number [-]TTemperature [K]tTime [s]UVelocity, [m/s]uFluctuating Velocity [m/s]VVolume [m3]WeWeber Number [-]iiXGrid Spacing [m]y+Dimensionless Distance [-]Subscripts Bulk ConditionsfFluidmMolarsSurfacewviscous, viscous stressiiiContents1 Spray Drying .. Purpose .. Constraints ..22 The Reversed-Flow Cyclone .. Modeling the Flow Field .. Turbulence .. Wall Modeling .. Modeling the Discrete Phase .. acting on the particles.

5 Of the Multiphase Flow .. wall interaction .. Pressure .. heat and Mass transfer .. transfer .. transfer ..143 Solving the Flow Field .. Conditions .. Procedure .. Convergence .. Discrete Phase .. of Droplets .. Parameter study .. Study .. Parameter Study ..194 Simulations of Flow Field .. profiles .. Discrete Phase Model and Simulations .. Wall Interaction .. Between 2D and 3D Simulations .. Parameter Study .. Experiments ..295 Conclusion and Modeling the flow field .. Discrete phase .. Parameter study ..326 Future Work33 Bibliography35 Appendix A Reference Case37 Appendix B Droplet Curvature Effect on Vapor Pressure39 Appendix C Estimating Distance to first Grid Point41 Appendix D factorial43 Appendix E Characteristic Time for Diffusion45 Appendix F User Defined Function - Multicomponent Droplet Vapor Pressure47 Appendix G Nozzle49vi1 IntroductionThis chapter describes the background to the thesis, beginning with a description of traditionalspray Drying .

6 Further on, the idea of Spray Drying in cyclones is introduced and the aim withthe project is process studied and some of the results in this thesis are classified. Therefore, the numericalvalues in this thesis are Spray DryingSpray Drying is a method of Drying a solution or a slurry, in order to produce a powder. Thisis done by dispersing the liquid into a gas phase. Major benefits of Spray Drying is that theparticle size can be controlled by the atomization of the liquid, and Spray Drying is a fast processcompared to other Drying methods. The typical Spray dryer consists of a Spray Drying chamber,where the Spray is dried, and succeeding separation equipment such as one or more Cyclone sep-arators. When it comes to the atomization device it varies between applications, however, thetwo most common types are rotary disks and single fluid high pressure nozzles. An example of anozzle is given in appendix G, which is also the nozzle used in the experimental part of this background for the thesis is a potential of reducing investment cost since the Spray dryingchamber is not needed, if it is possible to dry the liquid in the Cyclone .

7 Also the operational costhas a potential of being PurposeThe purpose for this thesis is to investigate whether Spray Drying is possible in Cyclone sepa-rators, and can be seen as a feasibility study. There will be a fundamental description of thedroplet behavior in the Cyclone and correlations to problems in conventional Spray Drying whenit comes to sticking of the discrete phase on the wall. A parameter study will be conducted inorder to determine the influence on the Drying efficiency of several factors, with the approachof avoiding wet particle impact on the Cyclone wall. This approach was chosen since one of themajor problems in traditional Spray Drying is sticking of the discrete phase on the complement to the theoretical study and simulations, experimental investigation was con-ducted, which can be seen as both complement to the theoretical study and as confirmation ofresults obtained from simulations and hand aim of the thesis is not to derive a concept for successful Spray Drying in cyclones, but ratherto investigate the possibility and point out the direction for future 1.

8 ConstraintsThe focus of the thesis is to describe single droplet behavior, both in the gas phase and at wallimpact. Due to limited computational recourses focus of the simulations will be in 2D, withverification in 3D. Since focus was on single droplet behavior, no simulations of a liquid spraywith collision and breakup of droplets was TheoryThe theory section starts with a description of a typical gas solid reverse flow Cyclone , which isgiven in section Section describes the fundamentals of CFD and also the modeling ofturbulence and the choice of turbulence model. In section the modeling of the discrete phaseis described, and finally section describes the heat and mass transfer from the The Reversed-Flow CycloneIn figure below, the different parts of the Cyclone is pointed : The different parts of the studied cycloneThe tangential gas inlets induce a strong swirling flow, moving downwards in the Cyclone .

9 Thedownward motion of the this vortex is reversed at the bottom of the Cyclone , and the gas flowis thus moving upwards forming an inner vortex, and exists through the gas outlet at the topof the Cyclone . For the inner part of the outer vortex the tangential velocity can be describedas a rigid body rotation with increasing velocity in the radial direction, whilst the outer regionhas a decreasing tangential velocity limited by the wall boundary layer. Further on, the strongswirling flow is affecting the pressure profile in the Cyclone , the maximum static pressure is foundat the outer walls and decreasing towards the axis of the Cyclone , where the static pressure mayfall below atmospheric pressure. In addition the flow field in the Cyclone exhibits low frequency3 Chapter 2. Theoryoscillations, where the formed vortex moves in three dimensions with a time dependence. Thisphenomenon is known asprecessing vortex core(PVC) [12].

10 Due to the swirling flow of the gas, the particles are subjected to a strong centrifugal force field,and the effect of the particles inertia is that they are transported radially towards the wall. Atthe wall, the particles will be transported to the dust outlet due to the motion of the boundarylayer and gravitational force [12]. The particle size is important in the separation, too smallparticles will exit with the gas whilst the larger particles exit through the dust Modeling the Flow FieldIncomputational fluid dynamicsthe governing equations for continuity, momentum, energy andspecies are solved. Generally there are no analytical solutions, so the equations must be solvednumerically [1]. Upon solving PDEs with numerical methods an error is introduced, and it isimportant to bear in mind that the error will always exist for numerical solutions. The error isdepending upon the step size, which in the CFD case is equivalent to the size of the computationalcells, and will thus decrease with a denser mesh.