ANSYS - Multiphase Flow

2010 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary 2010 ANSYS, Inc. All rights reserved. 1 ANSYS, Inc. Proprietary

Development and Validation of Eulerian

Multiphase Flow Models in ANSYS CFD

Development and Development and Validation of Eulerian Validation of Eulerian

Multiphase Flow Multiphase Flow Models in ANSYS CFDModels in ANSYS CFD

Thomas FrankANSYS Germany

[email protected]

Thomas FrankThomas FrankANSYS GermanyANSYS Germany

[email protected]@ansys.com

2010 ANSYS, Inc. All rights reserved. 2 ANSYS, Inc. Proprietary

Outline

Introduction

MPF model validation for adiabatic air-water flows

Polydisperse MPF model validation

MUSIG model

Validation of the RPI wall boiling model

The FRIGG testcase for boiling flow in rod bundles

RPI & MUSIG model

Summary & Outlook


ANSYS as Part of the German CFD Network in Nuclear Reactor Safety

GRS

FZ Dresden- Rossendorf /

TOPFLOW

ANSYS Germany

FZ Karlsruhe

AREVA

TU Mnchen: TD, Nucl. Energy

Univ. Applied Sciences Zittau-Grlitz: IPM

Univ. Stuttgart: Nuclear Energy

Becker Technologies ThAI test facility

International Guest Visitors


Methodology of CFD Model Development & Validation

ExperimentExperiment

phys.phys.--math.math.ModelModel

ValidationValidation

3d CFD Model3d CFD Model

Complex Complex GeometryGeometry

Complex Flow Complex Flow ConditionsConditions

Combination with Combination with other Modelsother Models


Eulerian MPF Modeling - The Particle ModelMass weighted averaged conservation equations

Mass, momentum, energy transport equations for each phase

1

N

k k k k k klll k

r rt

U kk k k k k k k k k k kr r r P rt U U U F I

turbulence models for each phase (e.g. k-

/ k-

SST model, 0-eq. disp. phase turb. model, Sato model)

heat transfer equations for each phase with interfacial transfer

closure

interfacial forces need empirical closure

high void fraction effects, bubble induced turbulence, etc.

secondary drag lift turbulentwall virtual mass

mom. transfer dispersionlubrication

L WL TD Mk D V FF F FI F F

PresenterPresentation NotesThe main governing multiphase flow equations for mass and momentum transport in a full N-phase multiphase flow framework are presented. The RHS of the momentum transport equations contain additional interfacial momentum transfer terms, which require additional closure. The mass and momentum transfer equations have to be accompanied by at least two turbulence model equations for the continuous phase and heat transfer equations for each of the phases, providing additional closure laws for interfacial heat and mass transfer (bulk condensation/evaporation) and turbulence modification due to the presence of the disperse phase (e.g. by the Sato model).


Lift force, Wall lubrication force & turbulent dispersion

Lift force:

due to asymmetric wake and deformed asymmetric particle shapeTomiyama CL

correlation

Wall lubrication force:

surface tension prevents bubbles from approaching solid wallsAntal, Tomiyama & Frank W.L.F. models

Turbulent dispersion force:

turbulent dispersion = action of turb. eddies via interphase drag

( )L G L LL G LrC F U U U (Re , Re , )L L PC C Eo

2WL G L rel rel W W WwallC r F U U n n n P(Eo, y/d )wall WC C

34

tFD P FTD F F P P

P rF P F

C r rU U rd r r

F FAD model by Burns et al. (ICMF04)

PresenterPresentation NotesFor the interfacial momentum transfer additional closure laws for the lift force, the wall lubrication force and the turbulent dispersion force have to be provided. For bubbly flows these can be specified by using Tomiyamas correlation for the lift force coefficient, Franks generalized correlation for the wall lubrication force coefficient and by usig the Favre averaged drag (FAD) turbulent dispersion force model by Burns et al.


Bubbly Flow Model Validation FZR MT-Loop and TOPFLOW Database

TOPFLOW

MT-Loop


CFX Model Validation MT-Loop & TOPFLOW Test Matrix

M01 experimental test series on MT-Loop

evaluation based on air volume fraction profiles at L/D=59,2 (z=3.03m) from the sparger system

-

numerically investigated test case conditions

superficial air velocity J_G [m/s]

s

u

p

e

r

f

i

c

i

a

l

w

a

t

e

r

v

e

l

o

c

i

t

y

J

_

L

[

m

/

s

]

finely disperse bubbly flow

bubbly flow with near wallvoid fraction maximum

bubbly flow in the transitionregime

bubbly flow with void fraction maximum at pipe center

bubbly flow with void fraction maximum at pipe center, bimodal

slug flow

PresenterPresentation NotesModel verification and validation is one of the most important steps in model development and implementation. In validating physical-mathematical models it is of special importance that the validation simulations are carried out in accordance to highest CFD standards, as they are outlined in the so-called Best Practice Guidelines (ERCOFTAC, ECORA), in order to clearly differentiate between sources of numerical errors (which has to be minimized by all means in a validation study) and remaining model errors. Furthermore it has to made sure, that suitable, high-resolution and high-quality experimental data are finally used for results comparison in order to avoid wrong judgment about the model accuracy.Model verification and validation is carried out on very different levels, starting from very simple configurations which allow under certain circumstances even analytical flow solutions. Next validation steps are undertaken for isolated phenomena tests, where special experiments are conducted in simplified geometries to reduce the level of flow complexity and phenomena interaction, e.g. by setting up quasi 1-dimensional experiments. Finally more combinations of physical models are validated in so-called demonstration tests, where flow complexity and physical phenomena interaction is comparable to the industrial size flow application.For the validation of the multiphase flow models FZ Rossendorf has established a large database of wire-mesh sensor measurements for air-water and vapor-water vertical pipe flows of different inner diameter (DN=55mm and DN=198mm). On the slide above the test matrix of the MT-Loop test facility is outlined, where the coloredcombinations of superficial air and water velocities mark the experiments for which cross-sectional velocity and gas volume fraction profiles are available. Numerical validation tests with ANSYS CFX have been carried out for the test conditions marked in red frames and results are compared to the experimental data.


Validation: Bubbly Flows Turbulent Dispersion Force

0,00

1,00

2,00

3,00

0 5 10 15 20 25

Radius, mm

N

o

r

m

.

A

i

r

V

o

l

u

m

e

F

r

a

c

t

i

o

n

k-eps + RPI TD (0.5)

k-eps + FAD TD

SST + RPI TD (0.5)

SST + FAD TD

Data

Model improvement


Monodispersed Bubbly Flow MT-Loop Test Case FZR-019

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.00 5.00 10.00 15.00 20.00 25.00Radius [mm]

n

o

r

m

a

l

i

z

e

d

v

o

l

u

m

e

f

r

a

c

t

i

o

n

[

-

]

Experiment FZR-019

Antal W.L.F., Grid 2

Tomiyama W.L.F., Grid 2

Frank W.L.F., Grid 2

FZD-019:

JL =1.017 m/sJG =0.004 m/sdP =4.8 mm

Grace dragTomiyama liftT./A./F. Wall L. ForceFAD Turb. Disp.SST turb. modelSato modelt=0.002s2210 Iterations

PresenterPresentation NotesOnly a few results of this extensive validation study can be shown here. Further results are available in technical reports and publications. After verification tests aimed to minimization of numerical errors (grid refinement studies, investigations on proper convergence level, integration time scale and discretization scheme for advection terms and time derivatives), different closure assumptions and correlations for the non-drag forces on bubbles have been compared to each other. It was found, that the lateral gas volume fraction distribution at the upper most measurement cross section in the MT-Loop test facility depends not only on the lift force, but to a large degree on the formulation of the so-called wall lubrication force too. It could further been shown, that the popular and widely used Antal correlation fails for a wide range of flow conditions, while the Tomiyama correlation and the generalized formulation by Frank was in good agreement to data.


Monodispersed Bubbly Flow MT-Loop Test Case FZR-052

FZD-052:

JL =1.017 m/sJG =0.0151 m/sdP =4.4 mm

Grace dragTomiyama liftT./A./F. Wall L. ForceFAD Turb. Disp.SST turb. modelSato modelt=0.002s2400 Iterations

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

0.00 5.00 10.00 15.00 20.00 25.00Radius [mm]

n

o

r

m

a

l

i

z

e

d

v

o

l

u

m

e

f

r

a

c

t

i

o

n

[

-

]

Experiment FZR-052

Antal W.L.F., Grid 2

Tomiyama W.L.F., Grid 2

Frank W.L.F., Grid 2

PresenterPresentation NotesIn CFD simulations using the Antal correlation for the wall lubrication force the gas volume fraction reached non physical high values close to the pipe wall, since the wall lubrication force amplitude predicted by the Antal correlation was to small in order to balance the lift force directed towards the wall as predicted by Tomiyamas law.


movable diaphragmwire-mesh sensor

movable diaphragm

TOPFLOW Test Facility @ FZDTOPFLOW Test Facility @ FZD

gas injection

PresenterPresentation NotesThe present investigation on the simulation of complex 3-dimensional bubbly flows is based on experiments, which had been carried out on the TOPFLOW test facility at FZ Rossendorf, Germany. A movable diaphragm (or half-moon shaped orifice) has been installed into the DN 200 vertical pipe of TOPFLOW. The mechanic drive allows for a 500mm displacement of the obstacle in the vertical direction. The wire-mesh sensor technique developed by the FZR is used in order to measure volume fractions, air and water velocities in the air-water dispersed bubbly flow around the obstacle. The wire-mesh sensor is thereby mounted in a fixed location, while the obstacle can be moved up- and downwards in order to allow for different measurement distances between the obstacle and the measurement plane between -500mm and +500mm. A sparger system is used for generation of a monitored and almost constant bubble size distribution.


TOPFLOW-074 Test Case Conditions from Test Matrix

Selection of test case conditions:

TOPFLOW-074 test case was subject of validation in the past

Superficial velocities: JG

=0.0368 m/s

JL

=1.017 m/s

Wire-mesh sensor measurements at locations:

z=10, 15, 20, 40, 80, 160, 250, 520mm

PresenterPresentation NotesAfter the experimental campaign measurement data for the flow around the movable obstacle are now available not only for the initially selected flow conditions of TOPFLOW test case 074, but furthermore for a larger umber of different air and water superficial velocities. In a second series of measurements the flow of multiphase flow mixture of saturated vapor bubbles at 65 bar and 280o C with water has been investigated.The experiments provide very detailed measurement data at 16 different cross sections of the vertical pipe up- and downstream of the obstacle. Together with the very dense cross-sectional spatial resolution of the 64x64 wire-mesh sensors this results in a 3-dimensional experimental data set for volume fractions, bubble size distributions and air/water velocities, which is excellently suited for validation of 3-dimensional CFD simulations.


3d Bubbly Flow Around Obstacle Water Velocity Comparison

Comparison CFD Experiment

Absolute water velocity distribution in symmetry plane

Import of exp. data into CFX-Post

Pre-interpolation of exp. data to z=0.01m

CFD Exp.

PresenterPresentation NotesAfter the experiments had been carried out at FZ Rossendorf, the 3-dimensional data set has been provided for a detailed comparison with the ANSYS CFX simulation. A special data conversion and interpolation program has been written for the wire-mesh sensor data by Prasser & Al Isssa at FZ Rossendorf. This turned out to be necessary (or at least beneficial), since the spatial resolution of the sensor data in z-direction on the one hand side and x-/y-direction otherwise was quit different without pre-processing. Therefore the experimental data were interpolated with z=0.01m between the 16 available measurement cross sections as given on slide 7. Afterwards the experimental data were imported as values of additional variables into CFX-Post, which allows to use the same set of flow visualization tools and equal color range for variable value visualizations as in a standard CFD post-processing. In the result the results of the CFX simulation and experimental data can be directly compared in all available detail. As a unique feature streamlines and variable value isosurfaces can be applied to the analysis of the experimental data as well.The above picture show the comparison for the absolute water velocity with the CFD result on the left and experimental data from TOPFLOW experiment on the right. Remarkable agreement in the flow structure can be observed. Furthermore the length scale of the recirculation area behind the obstacle, the location of the vortex core (to be seen by the green color directly downstream the obstacle) and the re-attachement length of the downstream flow are in good agreement.


3d Bubbly Flow Around Obstacle Air Void Fraction Comparison

Comparison CFD Experiment

Air void fraction distribution in symmetry plane

CFD Exp.

PresenterPresentation NotesThe same method of data comparison has been applied in the above picture to the air volume fraction. Both data sets show the entrainment and accumulation of air volume fraction behind the obstacle in the core of the vortex system, also the experimental data show a slightly lower level of air bubble accumulation. This is explained by bubble coalescence, which takes place in the experiments in the region of high air volume fraction and which was not taken into account by the numerical simulation. The formed larger bubbles are able to escape from the recirculation zone leading to less pronounced maximum void fraction in this area.


1) z=10mm

2) z=15mm

3) z=20mm

4) z=40mm

5) z=80mm

6) z=160mm

7) z=250mm

8) z=520mm

1)

2)

3)

4)

6)

7)

8)

5)

3d Bubbly Flow Around Obstacle Air Void Fraction Comparison

PresenterPresentation NotesThe comparison of the cross-sectional distribution of air void fraction downstream of the obstacle is of special interest, since we can observe strong separation effects and secondary motion in the experimental data. Almost identical patterns can be observed in the CFD simulation, which is mainly addressed to the action of the lateral lift force acting on the bubbles in region of high water velocity gradients. This can be remarkably observed in cross sections 4-6 in the above images, where regions of high air volume fractions can be observed in identical locations and with identical intensity in both the numerical and experimental data set.

Remark on Scaling : Air VF in the legend set to a maximum value of 15%. Only for Cross sections z=40mm and z=80mm a max. value of 25% was used.


3d Bubbly Flow Around Obstacle Cross-Sectional Air Void Fraction

5) z=80mm TOPFLOW ExperimentCFX Simulation

PresenterPresentation NotesFor a better view of the driving secondary fluid motion the cross-sectional images from plane 3 and 6 (CFD data) are reproduced in enlargement. Especially on plane 6 the region of high air volume fraction directly corresponds to a strong secondary fluid motion, which is directed inwards towards the center of the vortex system above and behind the obstacle. In contrary on plane 3, which is at z=+20mm downstream of the obstacle, the stagnation point flow on the obstacle surface leads to an almost homogeneous value of the air volume fraction in the right half of the measurement plane 20mm on top of the obstacle surface.


ANSYS CFX

Experimental Data Quantitative Comparison

Quantitative data comparison @ cross sections z=10, 15, 20, 40, 80, 160, 250, 520mm:

absolute water velocity air volume fraction

y=0mm

x=-35mm x=+35mm

z=520mm

z=-520mm

obstacle


ANSYS CFX


z=-80mmy=0mm

0.0

0.5

1.0

1.5

2.0

2.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

N

o

r

m

.

A

i

r

V

o

l

u

m

e

F

r

a

c

t

i

o

n

[

-

]

Experiment (z=-80mm)

CFX Simulation (z=-80mm)

0.0

0.5

1.0

1.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

A

b

s

o

l

u

t

e

W

a

t

e

r

V

e

l

o

c

i

t

y

[

m

/

s

]




ANSYS CFX


z=-20mmy=0mm

0.0

0.5

1.0

1.5

2.0

2.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

N

o

r

m

.

A

i

r

V

o

l

u

m

e

F

r

a

c

t

i

o

n

[

-

]



0.0

0.5

1.0

1.5

2.0

-100 -75 -50 -25 0 25 50 75 100

x [mm]

A

b

s

o

l

u

t

e

W

a

t

e

r

V

e

l

o

c

i

t

y

[

m

/

s

]




ANSYS CFX


z=80mmy=0mm

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

N

o

r

m

.

A

i

r

V

o

l

u

m

e

F

r

a

c

t

i

o

n

[

-

]

Experiment (z=80mm)

CFX Simulation (z=80mm)

0.0

0.5

1.0

1.5

2.0

2.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

A

b

s

o

l

u

t

e

W

a

t

e

r

V

e

l

o

c

i

t

y

[

m

/

s

]

Experiment (z=80mm)



ANSYS CFX


z=250mmy=0mm

0.0

0.5

1.0

1.5

2.0

2.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

N

o

r

m

.

A

i

r

V

o

l

u

m

e

F

r

a

c

t

i

o

n

[

-

]

Experiment (z=250mm)


0.0

0.5

1.0

1.5

2.0

2.5

-100 -75 -50 -25 0 25 50 75 100

x [mm]

A

b

s

o

l

u

t

e

W

a

t

e

r

V

e

l

o

c

i

t

y

[

m

/

s

]

Experiment (z=250mm)



Polydispersed Bubbly Flow Caused by Breakup & Coalescence

Transition from disperse bubbly flow to slug flow:

Balance between:

coalescence

of bubbles

turbulent bubble

breakup

bubble size distribution; polydisperse bubbly flowcounter-current radial motion of

small and large bubbles;

more than one velocity field

new population balance model (inhomogeneous MUSIG)

PresenterPresentation NotesFor gas-liquid flows of higher volume fraction the bubble size distribution is establishing in a balance between a) bubble breakup and coalescence processes and b) bulk condensation or evaporation taking place at the interface between the gaseous and liquid phase. Due to different lift excerted on bubbles of different size, small bubbles are driven to the geometry walls while large bubbles move to the geometry center like it can be observed e.g. in pipe and channel flows. This leads to further changes in the radial volume fraction distribution enhancing either bubble fragmentation or coalescence in dependence on local fluid velocity gradients and turbulence. In order to take these physical effects into account the inhomogeneous MUSIG model has been developed by FZ Rossendorf and ANSYS.


Inhomogeneous MUSIG Model

momentum equations are solved for N gas phases

(vel. groups)

size fraction equations for Mi

bubble size classes in each vel. group

bubble coalescence and break-up over all Mi

MUSIG groups

dPdP,krit

N(dP

)

coalescence/

evaporation d

1

d3

d4

d5

d6

d7

d8

d2

v1 v2 v3

Velocity group mass transfer

Break up Coalescence

breakup/

conden-

sation


Basic population balance equations

Size fraction equations

B B C C

dn(m, r, t) n(m, r, t) U(m, r, t)n(m, n(m, r, t) m(r, t)dt

r, t)m tt r

B D B D

MUSIG Model Extension

B B C C

ji d i i d i i B D B Dj i( r f ) ( r U f ) S S S St x

S

i ii 1 i

i i 1 i 1 i

m mm m m m

i ii i 1

i i 1 i 1 i

m mm m m m

Si =

for evaporation

for condensation

Bubble number density

New terms

Breakup/Coalescence terms

iii

i

ii


TOPFLOW Test Facility @ FZD

Courtesy of FZD

L

~

8

m

D=195mm


Validation of 3x7 Inhomogeneous MUSIG Model on TOPFLOW-074

good agreement at levels A, L through R

too fast spreading of the bubble plume from inlet due to too intensive turbulent dispersion

0.0

5.0

10.0

15.0

20.0

25.0

0.0 25.0 50.0 75.0 100.0

x [mm]

A

i

r

v

o

l

u

m

e

f

r

a

c

t

i

o

n

[

-

]

Exp. FZR-074, level AExp. FZR-074, level CExp. FZR-074, level FExp. FZR-074, level IExp. FZR-074, level LExp. FZR-074, level OExp. FZR-074, level RCFX 074-A, Inlet level (z=0.0m)CFX 074-A, level CCFX 074-A, level FCFX 074-A, level ICFX 074-A, level LCFX 074-A, level OCFX 074-A, level R

PresenterPresentation NotesFor the TOPFLOW (DN=198mm) test facility the gas/vapor inlet boundary conditions are different from MT-Loop, since the gaseous phase enters the flow domain through ring-shaped wall nozzles. Different levels of gas injection allow for the variation of the distance between gas injection and the measurement plane without changing the sensor location, which is difficult and time consuming in a high-pressure and high-temperature environment (as for water-vapor experiments). But in the result the two-phase flow can no longer be regarded as monodisperse, since the gas volume fractions at least in the vicinity of the gas injection location can exceed 25-30% leading to strong bubble coalescence. Therefore the inhomogeneous MUSIG model has been applied for the validation against TOPFLOW data in order to account for the polydisperse character of the bubbly flow. The above diagram shows the comparison of lateral gas volume fraction profiles from a 3x7 inhomogeneous MUSIG model simulation to wire-mesh sensor data. From the data it can be seen, that the inhomogeneous MUSIG model is able to predict the flow transition from a strong near wall peak in the lateral gas volume fraction profile (A-level injection) to a flow with a core peak for the largest distance between gas injection and the measurement plane (R-level injection)


Condensation Test Case

P=2 [MPa]

Jw

=1.0 [m/s]

Js

=0.54 [m/s]

Ts

=214.4 [C]

Tw

=210.5 [C] Tw

=3.9 [K]

Dinj

= 1 [mm]

Detailed experimental data: Bubble size distribution Radial steam volume fraction distribution

Dirk Lucas, Horst-Michael Prasser: Steam bubble condensation in sub-cooled water in case of co-current vertical pipe flow,Nuclear Engineering and Design, Volume 237, Issue 5, March 2007,

Pages 497-508


Physical Model Setup

Standard MUSIG & Extended MUSIG 25 bubble size classes 3 velocity groups:

03 [mm],36 [mm], 630 [mm] Arranged in accordance with critical Tomiyama bubble

diameter for bubble size dependent lift force Break up model: Luo & Svendsen (FB

=0.025) Coalescence model: Prince & Blanch (FC

=0.05)


TOPFLOW Condensation Testcase

Inlet BC Inlet Position WLF TD Force Heat Transfer

Config 1 Dinj = 4mmSource point @ Wall - CTD=1.0 Nu=2+0.6Rep

0.5Pr0.3

Config 2 Dinj

= 4 mm Source point @ 75 mm FWLF CTD=1.5 Nu=2+0.15Rep0.8Pr0.5

Config 3 Dinj

= 1 mm Source point @ 75 mm FWLF CTD=1.5 Nu=2+0.15Rep0.8Pr0.5


Results: Vapor Volume Fraction

Config 2 Config 3Config 1


Results: Vertical Averaged Steam Distribution


Results: Radial Steam Distribution


Results: Bubble Size Distribution


CFD Simulation for Fuel Assemblies in Nuclear Reactors

Material PropertiesMaterial Properties

Multiphase Flow Multiphase Flow ModelingModeling

Wall Boiling & Wall Boiling & Bulk CondensationBulk Condensation

Conjugate Heat Conjugate Heat Transfer (CHT)Transfer (CHT)

FSI: Stresses & FSI: Stresses & DeformationsDeformations

Validation againstValidation against

ExperimentsExperiments

TurbulenceTurbulence

PresenterPresentation NotesIn order to successfully apply CFD simulation to the prediction of flow through nuclear reactor fuel assemblies a large number of submodels are involved. This starts from the provision of accurate material properties (e.g. steam tables for a wider range of operating conditions for temperature and pressure). Further on submodels are required for single- and multiphase flow turbulence modeling as well as multiphase flow modeling for different flow regimes from liquid flow (subcooled region), bubbly flow (initially subcooled, then saturated), slug flow and annular flow (still saturated). The latter involves further submodels for bulk condensation and evaporation, wall boiling and conjugate heat transfer (CHT) in solids adjacent to the fluid flow domain where thermal boundary conditions have to be applied e.g. at the Uranium core of a fuel rod. High robustness, convergence efficiency and interoperability of all submodels of a certain CFD software code are necessary in order to make CFD simulations applicable to flow prediction for nuclear reactor fuel assembly flows. Finally submodels as well as the whole CFD software package have to be thoroughly validated against data from simplified isolated phenomena experiments and integrated tests.


Multiphase Flow Regimes for Boiling Water Flow

subcooled flow

bubbly flow slug flow

ONB OSB

Tsat

T

x

wall temperature

mean fluid temperature

subcooled boiling

nucleate boiling (saturated boiling)

annular flow

spray flow

PresenterPresentation NotesIf we consider flow conditions in a pipe or channel with heated walls, then we observe a change from single-phase subcooled liquid flow, to bubbly flow (ONB Onset of Nucleate Boiling, OSB Onset of Significant Boiling), slug flow regime with nucleate boiling, annular flow and finally the formation of droplet flow under dry-out conditions. The lower schematic diagram shows the behavior of wall and mean fluid temperature in comparison to the fluid saturation temperature in correspondence to the changing flow regimes.


Flows with Subcooled Boiling (DNB) RPI-Wall Boiling Model

EQFWall qqqq convective heat flux

1 ( )F F W Lq A h T T evaporation heat flux

E G Lq m (h h ) quenching heat flux

2 ( )Q Q W Lq A h T T

u

t*m *m

y

1A 2A2A

Quenching heat flux

*m

Convective heat flux

Mechanistic wall heat partioning model:

PresenterPresentation NotesThe Rensselaer Polytechnical Institute (RPI) developed the so-called RPI wall boiling or heat partitioning model. In this model the overall heat flux from the heated wall to the two-phase flow (the subcooled liquid) is divided into 3 parts: a convective, quenching and evaporation heat flux. Furthermore the heat flux partitioning model associates each of the heat flux contributions with a dimensionless wall area ratio in order to define the ratio between heat flux contributions.


RPI-Wall Boiling Model Submodels for Model Closure

Submodels for closure of RPI wall boiling model: Nucleation site density:

Lemmert & Chawla , User Defined Bubble departure diameter:

Tolubinski & Kostanchuk, Unal, Fritz, User Defined Bubble detachment frequency:

Terminal rise velocity over Departure Diameter, User Defined Bubble waiting time:

Proportional to Detachment Period, User Defined Quenching heat transfer:

Del Valle & Kenning, User Defined Turbulent Wall Function for liquid convective heat transfer coefficient

Correlation for bulk flow mean bubble diameter required: e.g. Kurul & Podowski

correlation via CCL

Supported combination of wall boiling & CHT in the solid GGI & 1:1 solid-fluid interfaces


RPI Wall Boiling Model in the ANSYS CFX-Pre 12.0 GUI


Investigated Boiling Testcases

Bartolomei et al. (1967,1982)

Lee et al. (ICONE-16, 2008)

OECD NEA PSBT subchannel benchmark (1987-1995, 2009)

Bartolomei with recondensation

(1980)R = 7.7 mm

Z

=

2

m

q

=

0

.

5

7

M

W

/

m

2

Gin

=900 kg/(s m2)

2010 ANSYS, Inc. All rights reserved. 45 ANSYS, Inc. Proprietary 2010 ANSYS, Inc. All rights reserved. 45 ANSYS, Inc. Proprietary

FRIGG-6a Test Case (Anglart & Nylund, 1967, 1996 & 1997)

FRIGGFRIGG--6a Test Case6a Test Case (Anglart & Nylund, (Anglart & Nylund, 1967, 1996 & 1997)1967, 1996 & 1997)


FRIGG-6a Test Case Description

13.8mm

21.6mm

35.5mm

Geometry (FT-6a) Six electrically heated

rods placed in a vertical adiabatic pipe

Flow Parameters Upward directed

subcooled water flow Mass flux @Inlet

Gin = 1163 kg m-2 s-1 Pressure @Inlet

pin = 5 MPa

Adiabatic Wall

Heated Rod

Rod wall heat flux qRod = 0.522 MW m-2

Liquid subcooling @Inlet Tsub = 4.5 K

LH

=4.2 m


FRIGG-6a Test Case Experimental Data

Determination of experimental data by gamma ray attenuation method: Measurements of area

averaged gas volume fraction in different cross-sectional zones along the test section

Defintion of Zones:

Zone1 (r < 14.6 mm)

Zone2 (14.6 mm < r < 28.6 mm)

Zone3 (r > 28.6 mm)

Zone1

Zone2

Zone3r


FRIGG-6a Test Case Mesh Refinement Hierarchy

Mesh01 Mesh02 Mesh03

No. Elements699 x 150(104 850)

2796 x 300(838 800)

11184 x 600(6 710 800)

No. Nodes 116 421 884 639 6 892 869

Max y+ 180 94 51

Min Angle [deg] 51.9 50.4 49.64

Min Determinant 0.84 0.91 0.98

Numerical Effort

~ 90 minutes@ 6 CPUs

~ 17 hours@ 16 CPUs

~ 6 days@ 40 CPUs


FRIGG-6a Test Case Baseline Setup: SST

Steam VF

Outlet Outlet

Two cross-sectional distributions of steam volume fraction (Mesh03,SST)

Plot of steam volume fraction (Mesh03, SST)


FRIGG-6a Test Case Baseline Setup: SST

TL

Outlet Outlet

Two cross-sectional distributions of liquid temperature (Mesh03,SST)

Plot of liquid temperature (Mesh03,SST)


FRIGG-6a Test Case Mesh Comparison


Turbulence Modeling in Rod Bundles

So far good comparison, but Wall friction in rod bundles leads to secondary flows Anisotropic turbulence SST BSL RSM Does not influence so much cross-sectional averaged

flow properties Secondary flows affect steam & temperature distributions

on heated wall surfaces Can be relevant for safety!


FRIGG-6a Test Case Turbulence Model Comparison



Outlet

Steam volume fraction

SST model

Outlet

BSL RSM model



Contour plot of steam volume fraction (Outlet)Liquid velocity in cross section (Outlet)

SST model NO secondary flows



Contour plot of steam volume fraction (Outlet)Liquid velocity in cross section (Outlet)

BSL RSM model secondary flows


New R&D Consortium

R&D Initiative: Modeling, Simulation & Experiments for Boiling Processes in Fuel Assemblies of PWR


Coupling of RPI Wall Boiling Model with Homog./Inhomg. MUSIG

Bubble departureWall Boiling Model (RPI)

Bubble breakup & coalescence(MUSIG Model)

Condensation(MUSIG model extension)

C

o

u

p

l

i

n

g

Future model extensions:Bubble departure diameter computed from equilibrium of forcesInclude further phenomena

CFX 12.1

CFX 12.1

Cust. Solv. v3

Cust. Solv. v1

Cust. Solv. v2


DEBORA Testcase: RPI & MUSIG

0 0.002 0.004 0.006 0.008 0.01R [m]

0.0

0.2

0.4

0.6

[

-

]

ExpCFX-12Gas1Gas2

0 0.002 0.004 0.006 0.008 0.01R [m]

1.5

2.0

2.5

3.0

3.5

[

m

/

s

]

UGAS ExpUGAS1 CFX-12UGAS2 CFX-12ULIQUID CFX-12

0 0.002 0.004 0.006 0.008 0.01R [m]

350

355

360

365

370

[

K

]

ExpCFX-12TSAT

0 0.002 0.004 0.006 0.008 0.01R [m]

0.0000

0.0002

0.0004

0.0006

[

m

]

ExpCFX-12

Inhomog. MUSIG

Phase change

Breakup &

Coales-

cence

RPI

Volume Fraction

Velocity

Temperature

Mean Sauter diam.

dashed lines dB

=f(Tsat

-TL

) ; solid lines dB

as mean Sauter diam. from MUSIG group

By courtesy of E. Krepper, FZD


Modeling, Simulation & Experiments for Boiling Processes in Fuel Assemblies of PWR

Ultrafast electron beam X-ray CT of fuel rod bundle in titanium pipe on TOPFLOW @ FZD:

Images by courtesy of U. Hampel, FZD


Modeling, Simulation & Experiments for Boiling Processes in Fuel Assemblies of PWR

Wall boiling simulation in 3x3 rod bundle with spacer grid:

Wall superheat TW

-TSat


Summary & Outlook

Overview on ANSYS CFD multiphase flow model development and validation

Continuous effort in model improvement, R&D

Emphasis in validation on BPG, comparison to data, geometry & grid independent modeling

High interoperability of physical models

Outlook: Ongoing & customer driven CFD model development Research cooperation with Industry & Academia More & more complex MPF phenomena Coupling of wall boiling model to inhomogeneous MUSIG Extension of the wall heat partitioning in wall boiling model


Thank You!

Development and Validation of Eulerian Multiphase Flow Models in ANSYS CFDOutlineANSYS as Part of the German CFD Network in Nuclear Reactor SafetyMethodology of CFD Model Development & ValidationEulerian MPF Modeling- The Particle ModelLift force, Wall lubrication force & turbulent dispersionBubbly Flow Model Validation FZR MT-Loop and TOPFLOW DatabaseCFX Model Validation MT-Loop & TOPFLOW Test MatrixValidation: Bubbly FlowsTurbulent Dispersion ForceMonodispersed Bubbly Flow MT-Loop Test Case FZR-019Monodispersed Bubbly Flow MT-Loop Test Case FZR-052Slide Number 12TOPFLOW-074 Test Case Conditions from Test Matrix3d Bubbly Flow Around ObstacleWater Velocity Comparison3d Bubbly Flow Around ObstacleAir Void Fraction Comparison3d Bubbly Flow Around ObstacleAir Void Fraction Comparison3d Bubbly Flow Around ObstacleCross-Sectional Air Void Fraction ANSYS CFX Experimental Data Quantitative ComparisonANSYS CFX Experimental Data Quantitative ComparisonANSYS CFX Experimental Data Quantitative ComparisonANSYS CFX Experimental Data Quantitative ComparisonANSYS CFX Experimental Data Quantitative ComparisonPolydispersed Bubbly Flow Caused by Breakup & CoalescenceInhomogeneous MUSIG ModelMUSIG Model ExtensionTOPFLOW Test Facility @ FZDValidation of 3x7 Inhomogeneous MUSIG Model on TOPFLOW-074Condensation Test CasePhysical Model SetupTOPFLOW Condensation TestcaseResults: Vapor Volume FractionResults: Vertical Averaged Steam DistributionResults: Radial Steam DistributionResults: Radial Steam DistributionResults: Bubble Size DistributionResults: Bubble Size DistributionCFD Simulation for Fuel Assemblies in Nuclear ReactorsMultiphase Flow Regimes for Boiling Water FlowFlows with Subcooled Boiling (DNB) RPI-Wall Boiling ModelRPI-Wall Boiling Model Submodels for Model ClosureRPI Wall Boiling Model in the ANSYS CFX-Pre 12.0 GUI Investigated Boiling TestcasesFRIGG-6a Test Case(Anglart & Nylund, 1967, 1996 & 1997)FRIGG-6a Test CaseDescriptionFRIGG-6a Test Case Experimental DataFRIGG-6a Test CaseMesh Refinement HierarchyFRIGG-6a Test CaseBaseline Setup: SST Steam VFFRIGG-6a Test CaseBaseline Setup: SST TLFRIGG-6a Test CaseMesh ComparisonTurbulence Modeling in Rod BundlesFRIGG-6a Test CaseTurbulence Model ComparisonFRIGG-6a Test CaseTurbulence Model ComparisonFRIGG-6a Test CaseTurbulence Model ComparisonFRIGG-6a Test CaseTurbulence Model ComparisonNew R&D ConsortiumCoupling of RPI Wall Boiling Model with Homog./Inhomg. MUSIGDEBORA Testcase: RPI & MUSIGSlide Number 60Modeling, Simulation & Experiments for Boiling Processes in Fuel Assemblies of PWRSummary & OutlookSlide Number 63

ANSYS - Multiphase Flow

Documents

Transcript of ANSYS - Multiphase Flow