Post on 08-Apr-2015
FLUENT 6.2Product Presentation
Multiphase and Chemical Processes
Mikael Stallgård, FLUENT SE
mas@fluent.se
Multiphase Modeling Capability- Overview -
Current Multiphase Models in FLUENTDispersed Phase Model
Steady and Unsteady particle trackingStochastic and Particle cloud model for turbulent dispersion Ability to include particle size distribution and include various forces and physics for particles (reaction), many UDF’s and post processing capabilities
Volume of Fluid ModelSolves for the interface between g/l and l/lAdvanced front tracking to resolve interfaceEffective for modeling the motion of large ( much bigger than grid) bubbles
Current Multiphase Models in FLUENTMixture model
A multi fluid model: Time dependent and 3D extension of drift flux modelCan effectively replace Eulerian Fluid/Fluid model where phases reach equilibrium fast.
Eulerian-Eulerian/Eulerian-Granular Multiphase ModelSolves momentum, enthalpy, and continuity equations for N phasesTurbulence models for dilute and dense phase regimes.
DEM3rd party integration (EDEM) for Lagrarian particle tracking withparticle-particle integration, may now be coupled with Fluent6.2
Multiphase – New features in 6.2
FLUENT 6.2
Multiphase Flows/Major Enhancement
Multiphase species transport and chemical reactionsSupport for all multiphase models Allows mass transfer between species in different phasesHeterogeneous reaction rates specified with UDF
Ozone Decomposition in a Fluidized Bed
Particles in a fluidized catalytic bed are used to convert ozone (O3) to oxygen (O2)FLUENT is used to study fluidization and reaction in a single simulationThe Eulerian granular multiphase (EGM) model is used with reactions in the gas phase
Ozone Decomposition in a Fluidized Bed
Rising bubbles of gas are predicted by the transient EGM calculationBubbles pass through bed surface and enter the gas space above
Ozone Decomposition in a Fluidized Bed
Gas phase reaction takes place in bed region onlyCout/Cin of ozone (1 –conversion) (pink) vs. gas velocity compares well with data (blue)
Saturation in gas holdup occurs when bed can no longer hold more gas, even at higher velocitiesFLUENT captures complex physics in a single simulation for this important industrial application
FLUENT 6.2
Multiphase Flows
Multiphase species transport and chemical reactionsSupport for all multiphase models Allows mass transfer between species in different phasesHeterogeneous reaction rates specified with UDF
Reynolds stress (RSM) turbulence model for dilute Eulerian multiphase flows
FLUENT 6.2
Multiphase Flows
Multiphase species transport and chemical reactionsSupport for all multiphase modelsAllows mass transfer between species in different phasesHeterogeneous reaction rates specified with UDF
Reynolds stress (RSM) turbulence model for dilute Eulerian multiphase flows
Extensions for the granular modelFull PDE for granular temperature Johnson and Jackson boundary conditionAdditional new options for granular property specificationSupport for granular model with mixture model
Multiphase/Mixture Model …
Mixture modelSecondary phase can now be granular
Applicable for solid-fluid simulationsSecondary phase cannot be “packed bed”
Granular physics added as followsAdd total granular pressure to momentum equationSolids viscosity available for dispersed solid phasePacking limit for solids Only algebraic granular temperature is availableApplicability is mainly for liquid-solids multiphase systems. Density difference should be small.
Slip velocity now contains dispersion due to turbulence.
Return
FLUENT 6.2
Multiphase/VOFVolume-of-fluid (VOF) model enhancements
Species transport and chemical reactions
More efficient front tracking, HRICHigh Resolution Interface Capturing (HRIC) schemeAllows larger time steps than geo-recon scheme
More diffuse than geo-reconProvides sharper interface than QUICKAvailable for structured and unstructured meshesApplication
Prediction of standing free surface waves in steady state
New open channel boundary condition
Support for inviscid flows
Improved surface tension robustness
Multiphase/VOF …
New HRIC scheme example of swirling fuel injector
First Order Implicit HRIC Implicit
Second Order Implicit
Interface Capture Using Implicit VOFVelocity Vectors Colored by Volume Fraction
Multiphase/VOF …
Experimental PhotographCourtesy of IIHR, U. of Iowa
Contours of Z Coordinate on Free Surface
Front tracking/open channel BC: Hydrofoil
FLUENT 6.2
Multiphase Flows
Discrete phase model (DPM) enhancementsNew wall film model
Particles impinge on a surface, splash and/or form a thin film
FLUENT 6.2
Wall Film Model: IC Engine
Wall Film Height (mm)Spray Colored by Particle Velocity
DPM …
Wall film modelParticles impinge on a surface and form a thin film
ImpingingFuel Droplet
Splashing
Fuel Film
Wall Conduction
ConvectiveHeat Transfer
Evaporation Shear Force Film Thickness
Flow separationand
Sheet Breakup
Ref. StantonInt. J. of Heat &
Mass Transfer (1998)
FLUENT 6.2
Multiphase Flows
Discrete phase model (DPM) enhancementsNew wall film model
Particles impinge on a surface, splash and/or form a thin film
Two-way particle-turbulence coupling
More accurate and efficient particle trackingNew tracking schemesError-controlled adaption of integration time-stepAutomated tracking scheme selection
FLUENT 6.2
Multiphase Flows
Target applications include:Bubble Column ReactorsFluidized Bed ReactorsAutomotive In-Cylinder FlowsSpray Nozzles and AtomizersShip HydrodynamicsFuel Injector Pumps
HydrocycloneSeparation of Solids
from Water
(Courtesy of M Slack
Fluent Europe)
Hydrocyclone Application
The experimental cyclone study carried out by Mondron et. al (1990) is used as a case study to validate the simulation approach
75mm diameter cycloneLimestone and water slurry with 10.47 % by weight solids with a size distributionOpen to atmosphere Air is drawn into the underflow and exits through the overflow via a stable air core.
Best Practice Modelling specific to Cyclones
The modelling challenges3 dimensional flow patternsAnisotropic turbulenceDispersed secondary phase
Both high and low volume loadingsSize distributionsParticle to Particle interactions
Low pressure central coreMay result in a backflow of gas (air core) in hydrocycloneUnstable flow structures may develop
Mesh Numerical sensitivityRadial Pressure distribution on over and underflow boundaries
Anisotropic Turbulence
+ k-epsilon, � RNG, –— RSM, ∆ experiment [LDA]
Tangential velocity Axial velocity
LES Cyclone Animations
Velocity magnitude Axial vorticity Velocity vectors
Transient flow structures
Multiphase Models
Air Core and Particle trajectories shown.
Multi-Phase Modelling (1) Lagrangian particle tracking
Effective method for quantifying separation performance.Can be carried out as a coupled calculation or as a post processing exercise.Must use sufficient particles to ensure insensitive result.Release position/distribution at inlet will is reflected in the predicted separation performance. In a steady flow field particles which may have only spent a fraction of a second in a toriodal re-circulation are held there indefinitely.Stochastic random kicks can be delivered to represent the effect of turbulent flow structuresDoes not account for the volume occupied by dispersed materialNo particle to particle interaction
Multi-Phase Modelling (2)
Algebraic Slip Mixture modelSolves a single set of momentum equations for both continuous and dispersed phases, (same velocity).Accounts for the occupied volume Can not simulate phases travelling in different directionsDispersed phase must achieve terminal velocity quickly
Computationally cheap Can be used to solve for air core but does not account for any interface effects, fluffy interface – cheap robust.Adequate when phases remain suspendedStruggles to resolve high dispersed phase concentrations at walls
Multi-Phase Modelling (3)
Eulerian-Eulerian-Granular modelThe most definitive multiphase model which solves a separate set of momentum equations for each phase.Accounts for high volume loadingsParticle to Particle interactionAccurate predictions of air core development
High CPU cost when combined with RSM turbulence equations.
RSM and Eulerian simulation results Cokljat et. al. 2003
Mondron cyclone simulated using RSM and Eulerian-Eulerianapproach, 5 phases simulated on 70,000 cells took 4 days to solve on 6 parallel CPU.
Validation (Test Case)
Tangential velocity at 2 locations measured using LDA compared to CFD predictions.
DPM particles (limestone) predicted separation efficiency curve compared to measured cyclone performance
Flow split
CFD 9%
Experiment 8.48
Templates
What are Templates?
Templates = process automation toolsEasy-to-use, customized interfacesRun GAMBIT and FLUENT in the backgroundAddress specific applications, customized and fine-tuned for your specific processesFacilitate the CFD process for both CFD engineers and non-CFD engineers
Templates do not replace the role of the expert analyst in defining the process, exploring the limits and fixing any problems.
Why Templates?
To increase productivity To enable broader use of CFDTo standardize processes
Increased Productivity
Automate & streamline any repetitive portion of the CFD analysis processSave time of CFD experts for more advanced CFD analysis projectsReduce turnaround time, to enable many more CFD calculationsMake parametric calculations possible, for studying design modifications
Broader Use of CFD
Use CFD in the design processUse CFD with optimization
Objective functions can be implemented in a template to perform optimization tasks
Enable any degreed engineer to use CFDNo knowledge or training of the supporting packages (GAMBIT,FLUENT,TGrid) is requiredA few analysts can support or supervise the work of many design engineers or process engineers
Standardize Processes
Capture engineering / process knowledgeExpert knowledge is embedded in the templateBest practices are built in, to ensure a better solution and faster convergenceOptimum mesh topologies are hardwired
Ensure process consistencyA standardized CFD process produces results independent of the userGet comparable results from every regional office
Example: Cyclone template
Which methodology do you want ?Physics vs. computational cost, complexity & stabilityFor design
Appropriate geometry description (circular inlet for demonstration only) RSMConstant slurry density with ASMM air core predictionDPM post processing used to calculate the separation performance.The tool automates the set up running and reporting of the results
Variety of cyclone geometries easily created using scripts
Maintenance & Evolution
Initial template version is based on current ‘best practices’ for given objectivesTemplates will need to evolve over time
‘Best practices’ may improve because of added knowledge from using the templateObjectives may changeGAMBIT and FLUENT will change too
Template maintenance is part of the service provided by Fluent Consulting
Future Modelling Developments
Future Activities
New solver technologiesInvestigation on different matrices - coupling
AccuracyHigher order schemes for time dependents problemNeed to reproduce convection without excessive numerical smearingNeed to account for time limitation in higher order spatial discretization.
Future Activities
Investigation on time dependent algorithm Need to reduce dramatically the computational time for large industrial cases.
Fractional steps methodMixed Implicit-Explicit SchemesPISO
Future Activities (DPM)
Improve robustness of current DPMExtend to dense multiphase flows
Interpolate particle properties to the gridInclude volume fraction in the continuous phase
Include particle normal stressImprove coupling with the continuous phase for a robust solution
Intelligent ParticlesHandle Polydispersed granular flows
Single Fluid Approach
Mixture, VOF and Cavitation Model will benefit from the transient Non Iterative Algorithm (NITA) Continuous improvement on physical modelling, accuracy and boundary conditions
Granular Flow Regimes
Kinetic theorySolid mechanicsElasticity
Stress is strain dependent
Stress is strain independent
Stress is strain dependent
Rapid flowSlow flowStagnant
ViscousPlasticElastic
Future Activities
Currently the granular model is applicable to the viscous regimeExtend validity of the model for frictional and elastic regime
Frictional ViscositySolids pressure and volume fraction relationshipHypoplasticity models
Binary models for viscous regimePolydispersed models for viscous regime
Turbulence in Multiphase
Models available in FLUENTMixture model, solves two equation turbulence model based on the mixture of all phasesDispersed Model, solves two equation turbulence model for primary phase and assumes turbulence quantities for dispersed phaseFull Κ−ε model for each individual phaseMixture Reynolds stress model Dispersed Reynolds stress model
Developments in Turbulence
Most model are just a simple extension of single phase modelsStarting from fundamental equation for the Reynolds stresses in multiphase develop Explicit Algebraic Stress Models capable of capturing the connection between the phases in the eddy viscosity formulation.Improvement on the modelling of dispersionImprovement in the RSMLES for multiphase
Population Balance
This is a critical component in modeling particulate systemsDispersed phase systems play an important role in many industrial production processesThree models are available for one dispersed phase
Discretization MethodStandard Method of MomentQMOM
Improvements on the kernels for aggregation, breakage and growthExtend to n phase
Data Structure for Multiphase ModelsData Structure in multiphase models involve multiple domains:
Super Domain: This is the top-level domain contains all phase-independent and mixture data: geometry, connectivity, propertySub-Domains: Each phase has a sub-domain that inherits the mixture-specific data and maintains the phase-specific dataInteraction Domain: To activate the phase interaction mechanisms
Sub-Domains
Threads
InletFluid-2
Fluid-1
Solid-1
OutletPorousMedium
Solid-2Wall
Threads
InletFluid-2
Fluid-1
Solid-1
OutletPorousMedium
Solid-2Wall
Threads
InletFluid-2
Fluid-1
Solid-1
OutletPorousMedium
Solid-2Wall
Interaction Domain
InletFluid-2
Fluid-1
Solid-1
OutletPorousMedium
Solid-2Wall
SuperDomain
Threads
Primary PhaseDomain
Secondary PhaseDomains
Sub-Threads
Coming Soon: FLUENT 6.3
FLUENT 6.3: Current Status
New feature versionImportant new core numerics functionality Enhanced physical modeling capabilitiesCustomer-requested enhancements
Development is currently underwayRelease is estimated for late 2005
Let’s preview the new core numerics functionality that we are working on in FLUENT 6.3…
FLUENT 6.3
New Core Numerics Functionality
Support for polyhedral meshesAutomatic cell agglomeration in the solverReduces original tetrahedral mesh by 3-5x
Results in faster convergence
Original tet/hybrid mesh: 51,467 cellsNew polyhedral mesh: 16,908 cells
FLUENT 6.3
Polyhedral Mesh: Simplified Sedan
Static PressureMesh
FLUENT 6.3
New Core Numerics Functionality
Support for polyhedral meshesAutomatic cell agglomeration in the solverReduces original tetrahedral mesh by 3-5x
Results in faster convergence
Pressure-based coupled solverImproved convergence and robustness for skewed/stretched meshes and “stiff” problemsConvergence rates not sensitive to mesh sizeLittle need to change relaxation factorsAbility to switch on-the-fly to fully segregated solver
FLUENT 6.3
Pressure-Based Coupled Solver: Propeller
2482101300Segregated (FLUENT 6.2)
4141360P-V Coupled (FLUENT 6.3)
Memory (MB)CPU (min)Iterations
Static Pressure
Solver Performance ComparisonMesh size: 172,000 Cells