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
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
2010 ANSYS, Inc. All rights reserved. 3 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 4 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 5 ANSYS, Inc. Proprietary
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).
2010 ANSYS, Inc. All rights reserved. 6 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 7 ANSYS, Inc. Proprietary
Bubbly Flow Model Validation FZR MT-Loop and TOPFLOW Database
TOPFLOW
MT-Loop
2010 ANSYS, Inc. All rights reserved. 8 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 9 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 10 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 11 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 12 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 13 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 14 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 15 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 16 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 17 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 18 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 19 ANSYS, Inc. Proprietary
ANSYS CFX
Experimental Data Quantitative Comparison
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
]
Experiment (z=-80mm)
CFX Simulation (z=-80mm)
2010 ANSYS, Inc. All rights reserved. 20 ANSYS, Inc. Proprietary
ANSYS CFX
Experimental Data Quantitative Comparison
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
[
-
]
Experiment (z=-20mm)
CFX Simulation (z=-20mm)
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
]
Experiment (z=-20mm)
CFX Simulation (z=-20mm)
2010 ANSYS, Inc. All rights reserved. 22 ANSYS, Inc. Proprietary
ANSYS CFX
Experimental Data Quantitative Comparison
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)
CFX Simulation (z=80mm)
2010 ANSYS, Inc. All rights reserved. 23 ANSYS, Inc. Proprietary
ANSYS CFX
Experimental Data Quantitative Comparison
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)
CFX Simulation (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)
CFX Simulation (z=250mm)
2010 ANSYS, Inc. All rights reserved. 24 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 25 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 26 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 27 ANSYS, Inc. Proprietary
TOPFLOW Test Facility @ FZD
Courtesy of FZD
L
~
8
m
D=195mm
2010 ANSYS, Inc. All rights reserved. 28 ANSYS, Inc. Proprietary
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)
2010 ANSYS, Inc. All rights reserved. 29 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 30 ANSYS, Inc. Proprietary
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)
2010 ANSYS, Inc. All rights reserved. 31 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 32 ANSYS, Inc. Proprietary
Results: Vapor Volume Fraction
Config 2 Config 3Config 1
2010 ANSYS, Inc. All rights reserved. 33 ANSYS, Inc. Proprietary
Results: Vertical Averaged Steam Distribution
2010 ANSYS, Inc. All rights reserved. 34 ANSYS, Inc. Proprietary
Results: Radial Steam Distribution
2010 ANSYS, Inc. All rights reserved. 35 ANSYS, Inc. Proprietary
Results: Radial Steam Distribution
2010 ANSYS, Inc. All rights reserved. 36 ANSYS, Inc. Proprietary
Results: Bubble Size Distribution
2010 ANSYS, Inc. All rights reserved. 37 ANSYS, Inc. Proprietary
Results: Bubble Size Distribution
2010 ANSYS, Inc. All rights reserved. 39 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 40 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 41 ANSYS, Inc. Proprietary
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.
2010 ANSYS, Inc. All rights reserved. 42 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 43 ANSYS, Inc. Proprietary
RPI Wall Boiling Model in the ANSYS CFX-Pre 12.0 GUI
2010 ANSYS, Inc. All rights reserved. 44 ANSYS, Inc. Proprietary
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)
2010 ANSYS, Inc. All rights reserved. 46 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 47 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 48 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 49 ANSYS, Inc. Proprietary
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)
2010 ANSYS, Inc. All rights reserved. 50 ANSYS, Inc. Proprietary
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)
2010 ANSYS, Inc. All rights reserved. 51 ANSYS, Inc. Proprietary
FRIGG-6a Test Case Mesh Comparison
2010 ANSYS, Inc. All rights reserved. 52 ANSYS, Inc. Proprietary
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!
2010 ANSYS, Inc. All rights reserved. 53 ANSYS, Inc. Proprietary
FRIGG-6a Test Case Turbulence Model Comparison
2010 ANSYS, Inc. All rights reserved. 54 ANSYS, Inc. Proprietary
FRIGG-6a Test Case Turbulence Model Comparison
Outlet
Steam volume fraction
SST model
Outlet
BSL RSM model
2010 ANSYS, Inc. All rights reserved. 55 ANSYS, Inc. Proprietary
FRIGG-6a Test Case Turbulence Model Comparison
Contour plot of steam volume fraction (Outlet)Liquid velocity in cross section (Outlet)
SST model NO secondary flows
2010 ANSYS, Inc. All rights reserved. 56 ANSYS, Inc. Proprietary
FRIGG-6a Test Case Turbulence Model Comparison
Contour plot of steam volume fraction (Outlet)Liquid velocity in cross section (Outlet)
BSL RSM model secondary flows
2010 ANSYS, Inc. All rights reserved. 57 ANSYS, Inc. Proprietary
New R&D Consortium
R&D Initiative: Modeling, Simulation & Experiments for Boiling Processes in Fuel Assemblies of PWR
2010 ANSYS, Inc. All rights reserved. 58 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 59 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 60 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 61 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 62 ANSYS, Inc. Proprietary
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
2010 ANSYS, Inc. All rights reserved. 63 ANSYS, Inc. Proprietary
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