© 2008 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary Multiphase Flow Developments in ANSYS CFX-12 Thomas Svensson Medeso
© 2008 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary
Multiphase Flow Developments inANSYS CFX-12
Thomas SvenssonMedeso
© 2008 ANSYS, Inc. All rights reserved. 2 ANSYS, Inc. Proprietary
Outline
• Euler-Euler– Wall Boiling Model– Non-Drag Forces
• Euler-Lagrange– Particle collision model– Wall film Modeling– Particle-Wall Interaction
• Other news/improvements
© 2008 ANSYS, Inc. All rights reserved. 3 ANSYS, Inc. Proprietary
RPI Wall Boiling Model
• Determines Heat Flux Partitionat Wall:– Q = Qc + Qq + Qe
– Qc = Convective Heat Transfer• Determined by Turbulent Wall Function
– Qq = Quenching Heat Transfer• Departure of a bubble from heated
surface cooling of surface by freshwater.
– Qe = Evaporative Heat Transfer• Determined by physical sub-models on
the sub-grid scale.
Qw
all
G
QC
QE
© 2008 ANSYS, Inc. All rights reserved. 4 ANSYS, Inc. Proprietary
Wall Boiling Validation:Bartolomej Test Case
• MMT Validation Test Case– Subcooled Boiling in Pipe with Heated Wall Bartolomej
et al. (1967, 1982) (Conxita Lifante, 2008)• large number of experimental testcase conditions with data• steam-water pipe flow with wall boiling• liquid sub-cooling defined to have steam inception
always at the same wall height– Different configurations were studied in the paper.
Main parameters:• Mass inflow rate• Pressure• Wall heat flux• Pipe diameter
© 2008 ANSYS, Inc. All rights reserved. 5 ANSYS, Inc. Proprietary
Bartolomej Test Case: Description
• 2D axial symmetry,steady simulation• 1 degree extrusion• Specified heat flux at
the wall• Symmetry b.c. at
planes and axis• Inlet b.c. with given
inlet mass flow• Outlet b.c. with
average staticpressure
X=X/100
X=2m
R
© 2008 ANSYS, Inc. All rights reserved. 6 ANSYS, Inc. Proprietary
Results: Grid3
Axial development of water temperatur and steamvolume fraction
© 2008 ANSYS, Inc. All rights reserved. 7 ANSYS, Inc. Proprietary
Comparison to Experimental Data- Grid Independent Solution
© 2008 ANSYS, Inc. All rights reserved. 8 ANSYS, Inc. Proprietary
Wall Boiling Verification:Rod Bundle Geometry
• 3×3 rod symmetry sectionfrom a nuclear reactor fuelassembly with guide vanes– Periodic BC’s at all sides– Wall heat flux of
qwall = 106 W/m2
– Reference Pressurep = 15.7 MPa
– Water inlet temperatureTInlet = 607K
• FZ Dresden andAnsys Germany
• Validation with comparison toexperimental data is work in progress
© 2008 ANSYS, Inc. All rights reserved. 9 ANSYS, Inc. Proprietary
Non-Drag Forces
• Motivation– Bring to release excellent progress made in validation of non-drag
force models in bubbly flow• Joint work between Ansys and Forschungszentrum Dresden
(FZD)• Utilised in conjunction with wall boiling in most validation
studies of boiling flow.– Also make available well validated models for spherical solid and
liquid droplet lift forces– Numerics improvements to mitigate poor robustness of Virtual
Mass Force implementation in previous releases (fundeddevelopment)• See later section on numerics improvements.
© 2008 ANSYS, Inc. All rights reserved. 10 ANSYS, Inc. Proprietary
Non-drag Forces
• Mass weighted averaged conservation equations
( ) ( ) 0k k k k k
t! " ! "
#+$ =
#U
( ) ( ) ( )kk k k k k k k k k k kP
t! " ! " " "
#+$% = & $ &$% ' + +
#U U U F I
Turbulence models for each phase(k-ε, k-ω, SST, 0-eq. disperse phase turbulence model)
Interfacial forces need empirical closure
{ { { { {drag lift turbulentwall virtual mass
dispersionlubrication
L WL Tk D D VM= + + + +F F FI F F
© 2008 ANSYS, Inc. All rights reserved. 11 ANSYS, Inc. Proprietary
Non-Drag Forces: Lift
• Tomiyama Model– Well validated model for bubbly flow.– Takes into account change of sign of lift force due to change in
bubble shape as bubble size increases.– Depends on Eotvos number, hence requires specification of
surface tension and gravitational force.
• Saffman Mei– Applicable to rigid spheres.– Generalises Saffman’s anaytical model to extend applicability to
higher particle Reynolds numbers.
• Legendre Magnaudet– Applicable to liquid drops.– Takes account of induced circulation inside drops.
© 2008 ANSYS, Inc. All rights reserved. 12 ANSYS, Inc. Proprietary
Non-Drag Forces: Wall Lubrication
• Tomiyama– Like Tomiyama lift force, depends on Eotvos number, hence
accounts for dependence of wall lubrication force on bubbleshape.
– In conjunction with Tomiyama lift force, produces excellent resultsfor bubble flow in vertical pipes.
– However, requires pipe diameter as input parameter, hencegeometry dependent .
• Frank– Generalises Tomiyama’s model to be geometry independent.– Model constants calibrated and validated for bubbly flow in
vertical pipes.
© 2008 ANSYS, Inc. All rights reserved. 13 ANSYS, Inc. Proprietary
Non-Drag Forces Validation:Bubbly Flow in Vertical Pipe
• Forschungszentrum Dresden (FZD) MT-Loop testfacility.– Wiremesh sensor with 24x24 electrodes.– Database to test CFD predictions.– Length, L = 4 m, Inner Diameter, D = 51.2 mm.
• Air-Water at atmospheric pressure, and 30 C.• Measurements carried out for stationary flows of
various superficial velocity ratios.– 10 different cross sections located between L/ D = 0.6 and 59.2
from gas injection.– Select test cases in bubbly flow regime with a near-wall peak in
gas volume fraction.
© 2008 ANSYS, Inc. All rights reserved. 14 ANSYS, Inc. Proprietary
Non-Drag Forces Validation:Bubbly Flow in Vertical Pipe
Test dp[mm]
Ul,sup[m/s]
Ug,sup[m/s]
017 4.8 0.405 0.0040
019 4.8 1.017 0.0040
038 4.3 0.225 0.0096
039 4.5 0.405 0.0096
040 4.6 0.641 0.0096
041 4.5 1.017 0.0096
042 3.6 1.611 0.0096
074 4.5 1.017 0.0368
© 2008 ANSYS, Inc. All rights reserved. 15 ANSYS, Inc. Proprietary
Validation: Bubbly Flow in VerticalPipe
© 2008 ANSYS, Inc. All rights reserved. 16 ANSYS, Inc. Proprietary
Particle Collision Model - Conception
• Statistical collision model (Sommerfeld)– Computational effort of simultaneously tracing all
particles is not required– Instead an iterative approach is used:
• Sequential calculation of particle trajectory• Compute statistical particle properties (mean and standard
deviation of droplet diameter and velocities)• Creation of a virtual collision partner according to local
statistical mean particle properties• Random process decides whether or not a collision takes
place
© 2008 ANSYS, Inc. All rights reserved. 17 ANSYS, Inc. Proprietary
Description of Validation Experiment
• Enforced crossing oftrajectories
• Flow induced bygravitation
• Glass particles, dP = 3mm
• ρP = 2500 kg/m3
• Collision effects dominate
Validation by experiment of Fohanno & Oesterlé
© 2008 ANSYS, Inc. All rights reserved. 18 ANSYS, Inc. Proprietary
Particle Trajectories Without / WithCollision Model
without collision model with collision model
© 2008 ANSYS, Inc. All rights reserved. 19 ANSYS, Inc. Proprietary
Notes on Particle-Particle Collision
• Validation report available“A stochastic particle-particle collision model for dense gas-particleflows implemented in the Lagrangian solver of ANSYS CFS and itsvalidation”, 6th International Conference on Multiphase Flow, ICMF2007, Leipzig, Germany, July 9-13, 2007, Paper No. 148, pp 1-16
• This is an ‘expensive’ modelParticle integration time step may become very small compared tonon-collision simulation (up to several (~2 - 4) orders of magnitude)
© 2008 ANSYS, Inc. All rights reserved. 20 ANSYS, Inc. Proprietary
Particle Wall Film
• Major physical phenomena
• Film movement due to external forces isneglected for CFX 12– Still film can move if on a moving wall
Conduction
Convection
Impinging
SplashingEvaporation External Forces Separation
© 2008 ANSYS, Inc. All rights reserved. 21 ANSYS, Inc. Proprietary
Wall Film Modeling – Approach andMain Assumptions
• Modeling Approach– Wall film is modeled using a Lagrangian
approach• I.e.: Wall film made of a special type of particles “Wall particles”
• Assumptions– Thin film approach (no displacement effect)– Neglect influence of film on fluid drag– No film movement due to external forces Quasi Static Wall Film
© 2008 ANSYS, Inc. All rights reserved. 22 ANSYS, Inc. Proprietary
Wall Film Modeling Example:Evaporating Droplets
conductQ
Water droplets (Tp = 293 K) hit a heatedwall
Assumptions:
• Droplets sticks to wall, i.e. norelative movement between particleand wall
• Energy is transferred fromwall/surrounding to film
• Film evaporates into ambient
Twall = 350 [K]
convectiveQ
© 2008 ANSYS, Inc. All rights reserved. 23 ANSYS, Inc. Proprietary
Wall Film Modeling Example:Evaporating Droplets (2)
© 2008 ANSYS, Inc. All rights reserved. 24 ANSYS, Inc. Proprietary
Particle-Wall Interaction Model
• Particle-Wall Interaction Model– This model describes how particles interact with film
covered walls and under which conditions a wall film isformed Prerequisite for wall film model
• Droplet-wall interaction is complex and not all aspectsare well understood.– Dimensional analysis shows that droplet-wall interaction
depends on:• Particle quantities (Weber Number), existence of a wall film, wall
roughness, wall temperature (and much more)
© 2008 ANSYS, Inc. All rights reserved. 25 ANSYS, Inc. Proprietary
Notes on the Wall Interaction Models
• Elsässer Model– Accounts for wall temperature effects, wall roughness and
particle-wall material combination, …– Targeted towards IC-E applications (~ Gasoline injection)
• Stick to Wall– Simplest possible model: all particles that hit a wall become part
of the wall film
© 2008 ANSYS, Inc. All rights reserved. 26 ANSYS, Inc. Proprietary
Spin-off of Wall Interaction Extension
• Child droplet generationmodel– Parent droplet can
create more than onechild
© 2008 ANSYS, Inc. All rights reserved. 27 ANSYS, Inc. Proprietary
Other Multiphase developments
• New Turbulence Induced Atomization Model• Improvements / Added Functionality for– Robustness of Coupled Volume Fraction for
inhomogeneous multiphase flows– Discretisation of Virtual Mass Force for more robustness– More user control of Particles and Particle Output– Particle Injection Options– Secondary Break-Up Models
© 2008 ANSYS, Inc. All rights reserved. 28 ANSYS, Inc. Proprietary
Multiphase Fluent 12
• Coupled multiphase solver• Multi-Fluid VOF• Cavitation model• DDPM
© 2008 ANSYS, Inc. All rights reserved. 29 ANSYS, Inc. Proprietary
Coupled Solver
• Simultaneous solution of the equations of amultiphase system would offer a more robustalternative to the segregated approach.
• The memory usage would be larger than the PCSIMPLE but the gains in convergence make thisapproach attractive for steady-state solution
• FLUENT has already an AMG coupled solver withILU smoother used for single phase
• Description below uses velocity and pressurecorrection. Volume fraction is solved segregated
• Can be extended to volume fraction correction
© 2008 ANSYS, Inc. All rights reserved. 30 ANSYS, Inc. Proprietary
Validation IV
• Three-dimensionalturbulent mixing tank,Montante and Bakker(2004).– The system under
investigation is a four-baffled vessel with fourRushton turbines solvedwith the multiple referencemodel
– For a converged solutionthe CPU time ratiobetween the PC-SIMPLEsolver and the coupledsolver was about 2.3.
© 2008 ANSYS, Inc. All rights reserved. 31 ANSYS, Inc. Proprietary
Multi-Fluid VOF
• Multi-Fluid VOF involves the following features:– Interface sharpening schemes (such as Geo-Reconstruct, CICSAM,
Modified HRIC) in the framework of Eulerian multiphase. This givesaccess to non-shared velocity and temperature fields for the problemsinvolving sharp interface treatment.
– Variable time stepping for Explicit schemes in the framework ofEulerian multiphase.
– Modeling of• Surface tension• Wall adhesion• Marangoni convection in the framework of Eulerian multiphase.
– Modeling of Anisotropic drag, especially for free surface flows.– Compatibility of Explicit schemes with other models such as
Turbulence, Energy, Species and Mass transfer, Dynamic mesh,Granular flow.
– Immiscible fluid option to model free surface flows. This option enablesGeo-Reconstruct and CICSAM schemes for Explicit VOF. Drag lawoptions with this model are “Symmetric” and “Anisotropic”.
© 2008 ANSYS, Inc. All rights reserved. 32 ANSYS, Inc. Proprietary
Bubbles with Multi-Fluid VOF
Bubbles rising through a slurry of granular solids in water. DPM isused to track the red particles with the granular phase velocity
© 2008 ANSYS, Inc. All rights reserved. 33 ANSYS, Inc. Proprietary
Summary of Cavitation Models inFLUENT 12
• Cavitation models developed under the general multiphase,pressure-based numerical framework– Not available for the density-based solver
• The cavitation models can be applied to any geometric system,all grid types supported in FLUENT, non-conformal / slidinginterfaces, and moving/deforming mesh
• The models have been extended to multiphase and multi-species systems
• The models can be solved with mixture (mixture model) orphase (Eulerian multifluid) temperature equations
• They are fully compatible with all the turbulence models inFLUENT, ranging from simple length scale models to LES
• Both liquid and vapor phase can be incompressible orcompressible.
• The input material properties (vaporization pressure, density,viscosity, and etc.) can be constants or functions oftemperature.
...),,,,,,,1{ !" kYiTWVU= ...),,,,,,,1{ !" kYiTWVU= ...),,,,,,,1{ !" kYiTWVU=
© 2008 ANSYS, Inc. All rights reserved. 34 ANSYS, Inc. Proprietary
Dense Dispersed Particle Model
• The dense dispersed particle model (DPPM)is a Lagrangian technique to modelparticulate flows
• Provides an efficient treatment for sizedistributions in multiphase problems
• In FLUENT, this model is an extension fromDPM to account for dense phase effects.– Account for the effect of blockage on the fluid
• Introduce calculation of volume fraction– Account for the effect of collisions on the
motion of particles• Use particle pressure and particle kinetic energy
from Granular Kinetic Theory