nuclear fuel assembly thermal hydraulics analysis...
TRANSCRIPT
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta2013.06.18
TAA Lecture ANS-Atlanta
Hisashi NINOKATAPolitecnico di MilanoDepartment of EnergyCeSNEF-Nuclear Engineering DivisionNuclear Reactors Group
Nuclear Fuel Assembly Thermal Hydraulics Analysis – Past, Present, and Future
Prof. Hisashi NINOKATAPolitecnico di Milano
Centro Studi Nucleari Enrico Fermi - CeSNEFDipartimento di Energia
via La Masa 34, 20156 MilanoItaly
E-mail: [email protected]
ISSCA‐5 TO BE ANNOUNCED
ALSO SAVE THE DATE: NUTHOS‐10 IN OKINAWA, JAPAN (DEC 14‐18, 2014) AND NURETH‐16 IN CHICAGO, USA (AUG 31 TO SEP 5, 2015)
International Seminar on Subchannel Analysis (ISSCA)
1st in Tokyo, 1992
2nd in Palo Alto, 1993
3rd in Stockholm, 1995
4th in Tokyo, 1997
5th under consideration in Milan, Italy, 2014
covers subchannel analysis and CFD/CMFD rod bundle thermal hydraulics.
Note: International seminar on subchannel analysis, CFD modeling and verification, as well as CHF experiment and benchmarking (ISACC) is going to
be held in Xi’an, China, August 3‐4, 2013
3
NOTICE
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaIntroduction: CFD vs Subchannel Analysis
DNS
LES
RANS•RSM, ASM
•K-e•Two-fluid model for 2- flow
Subchannel analysis
1D system analysis
Multi-C multi- flow
Subassembly disintegration process
Fuel relocation and DHR
Post Acc. HR
Boiling flow
W-I-SVoid drift
BWR BT/post BT
3rd field (D)
1 flow
Inter-SC exchanges
Divergence X-flow
Turbulent mixing
Global Flow Pulsation
’60s~COBRA IIIC, D.S. Rowe
’70~ COBRA IV-I’80~ COBRA-TF’90~ NASCA
`70 TRAC
MP HEC availableBlugene, Earth-Simulator, …
’70 IBM 360, CDC, FACOM, NEC, etc
’80~ Vector machine (VP50, VP100,..)‘90~ Para comp.
Current and Future
W. Heisenberg at his death bed in 1976: “When I meet God, I am going to ask him two questions: Why relativity? And why turbulence? I really believe he will have an answer for the first.”[comment] “turbulence” should be changed to “two phase turbulence”
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaCFD vs Subchannel Analysis
DNS
LES
RANS•RSM, ASM
•K-•Two-fluid model for 2- flow
Subchannel analysis
1D system analysis
Wire spacer TH by LES on-going (Merzari, ANL)Grid spacer TH by LES
Why not by CFD, instead of subchannel analysis?
Model experiment37-rod Bundles Air flow inside the
bundle
P/D=1.06Re=38,750
~ 20 m/s
CFD
Wire-spacer model required
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaSubchannel analysis (SCA)
RANS is self-closed and not in a position to provide information to SCA
What SCA can do and CFD cannot
Based on the presmise that we could
Predict accurately the flow and temperature conditions in different portion of the whole rod bundle;
Predict − Intra-subassembly as well as inter-subassemblies transfer phenomena
− Partial and total blockage accident phenomena
− Space-dependent boiling flow dynamics behaviors
Link and correlate local flow conditions to a local CHF criterion --- use the tube CHF data or LUT; or
Directly calculate the film dryout (BT) and rewetting; with an appropriate additional droplet field
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Velocity vector
x [cm]210-1-2
z [c
m]
500
480
460
440
420
400
380
360
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
Forced to Mixed Convection Intra S/A Redistribution bySubchannel Analysis
100% power & flow (Re=16200)
Velocity vector
x [cm]210-1-2-3
z [c
m]
500
480
460
440
420
400
380
360
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
Flow recirculation
5 % power & flow (Re=810)
Power skew: qmax/qmin=3.0
Heat loss from the hexcan wall: 10%
Flow and temperature re-distributions
Onset of laminarization, flow reversals and recirculation
1D Modeling
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
MIXING
Single-phase flow subassembly TH characteristics
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Local Instantaneous Transport Equations
Integral balances
Definitions of the average:
u =- J ( )St
Volumeintegration
z
TransverseMomentumCV
Basic subchannelCV
Cross flowcomponent
1Volume integration
1 1 1Surface integration
1 1 (fluid-fluid); ] (fluid-solid)
f
ff fs ff fs
ff fs
Vf
A A A Af f f
A Af f
dVV
dA dA dAV V V
dA dAV V
≮ ≯ [
Subchannel formulation
Vol Integral of div and grad terms to surface integral: Gauss
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Time- and CV-Averaging N-S Eqs results in DRM and Turbulent Mixing = Constitutive equations
Momentum transfer at the fluid-fluid interface of the subchannel boundary
1[ ]
ffAf
M n dAV
≮ ≯ =
J p
-≮ ≯ ≮ ≯ ≮ ≯= ≮ ≯
Mixing:
fluid-fluid
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Time- and CV-Averaging N-S Eqs results in DRM and Turbulent Mixing
Local instantaneous velocity; Time- and area-averaged velocity
Mass, Momentum and Energy
Example: Turbulent shear stress tensor
For example;
must be evaluated on the CV surface and
expressed by
Calculation results from CFD/CMFD would be useful to evaluate all the tensor elements
'
'
'
u u u
u v v v
w w w
u
u v
w
≮ ≯
≮ ≯ ≮ ≯
≮ ≯
t t txx xy xz
t t t tyx yy yzt t tzx zy zz
' ', ' ', ..., .t txz yzu w v w etc
and u
≮ ≯ ≮ ≯
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Area averaging of time-averaged velocities: Remarks
Local instantaneous velocity; Time- and area-averaged velocity
The velocity components in subchannel analysis (SCA), i.e., axial and cross-flow components (u, w), always refer to the CV area-averaged and time-averaged velocity components.
It means that u and w appearing in SCA are implicitly understood as:
i.e., with over-bar and the area averaging being omitted.
In the case of square rod array configuration, the velocity components are (u, v, w), with u and v transverse components defined on the four CV surfaces at the rod-to-rod gap. For triangular arrays, the transverse component u of the velocity is defined on the three CV surfaces at the rod-to-rod gap.
Difference from the porous media formulation: All these transverse-flow components (i.e., cross-flow) are fixed in their flow direction perpendicular to the CV fluid-fluid surface at each gap. This is the major difference from the porous media formulation in 3-D space.
'
'
'
u u u
u v v v
w w w
u
u v
w
≮ ≯
≮ ≯ ≮ ≯
≮ ≯
u w≮ ≯and ≮ ≯..≮ ≯
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Turbulent Mixing – Boussinesq Eddy Diffusivity
;
(Boussinesq's Eddy Viscosity)
t txz
udw
vdx
w
u
≮ ≯
' ' ( )
( )
( )
t txz
M
M M
dwu w
dxdw
dxdw
dx
wj+wj’
uk+ uk’
wi+wi’
Wj: time- and space-averaged axial velocity of the subchannel jWj’: its fluctuation
uk: time- and space-averaged x-flow velocity betw subchannel j and Iuk’ : its fluctuation
(u,v,w)t : independent variables in SCA (lumped, not distributed)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Diffusion Terms (cont’d)
when the heat conduction can be neglected relative to turbulent diffusion (Water);This is not true for low Pr liquid metals
fluid-fluid
Time- and CV-Averaging Energy Eq results in wall heat transfer and Turbulent Mixing
" )
1
ff
tM
H H
Af
Q q k k T
h h ndAV
≮ ≯ ≮( ≯
≮ ≯
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTurbulent Energy Mixing including Conduction
* '
1 1 1
P r
H Hij ij ijL
P ij
H
ij L Tij ij
kW s W
c x
sx x
Definition ofMixing Coefficients(MIT Mixing Project)
j i j ik ij ijL L
j jfi ij fi P ij
T T h hk kQ s s
A x A c x
'
*
1( )
1( )
Hk M ij ij j iL
jfi P ij
Hij j i
jfi
kQ Q s W h h
A c x
W h hA
Heat Conduction
Total
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
ff
j iMM A K
kf f k
w w1 1M ndA A
V V x
M ba Re , a 0.0084 and b 0.875
j iHM K
kf f k
e e1 1Q q" ndA A
V V x
H M 0.45 0.2 0.4 0.20.14 Re Pr 1 exp 71.8Re pr
k: Gap numberj: adjacent subchannel number to the gap kAk: Gap k area; xk: Turbulence characteristic length between two subchannels
Turbulence Mixing by COBRA-IIIc
Momentumdiffusionat the fluid-fluidinterface
Energy diffusionat the fluid-fluidinterface
'Hij ij ij kW s G
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTurbulence Mixing by COBRA-IIIc
Introduction of dimensionless Stanton number M
and
Mixing parameter
'Hij
ijij i
WM
s G
0.1250.001Re (2 / )ij ij ijs l
'Hij ij ij kW s G
'Hij
ijij k
W
s G
Re / ;
2k h
h ij
G D
D s
Rowe and Angle 1967Rogers and Rosehart 1972Rogers and Tahir 1975Gonzalez-Santalo and Griffith 1972Rudzinski et al. 1972Kuldip and Pierre 1973Beus 1970
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTurbulence Mixing by COBRA-IIIc (Energy)
'
1 1" j iH
fi M K Tj ij
j iHij T
k ij
h hA Q q ndA A
z z x
h hW
x
COBRA-IIIc
'Hij ij ij kW s G
So far, mechanistic modeling; Single-phase mixing parameter correlation is appropriate in the range =0.0045-0.005.
However, is often adjusted to get the best fit to experiment.
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaA gap to close
Classical turbulent mixing has not been satisfactory in predicting temperature distributions
It has been pointed out some other mixing mechanisms are responsible for the gap between subchannel analysis and the experiment
Global Flow Pulsation (GFP) is proposed to close the gap
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaOscillations
We want to define a characteristic frequency and a characteristic length
When Reynolds averaged Secondary Flows
Oscillations appear early in the laminar-turbulent transition (Reynolds threshold) a particular mechanism of instability plays a role ?
Computational Approach
DNS LES URANS
Simpler eccentric annular channels bundle subchannels
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
37-rod Bundles Air flow inside the
bundle
Large scale oscillations when P/D < 1.1Experiments of Krauss and Meyer
Data from: Krauss, T. and Meyer, L., Nuc. Eng. And Design (1998)
Experimental facts
P/D=1.06Re=38,750
~ 20 m/s
Power spectra of the cross velocity
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
URANS for Global Pulsation in Pin Subassemblies
URANS test calc.Presented at NURETH-12 by E. Merzari + H. Ninokata
Cyclic B.C.s
600mm ~ 4
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaGlobal Flow Pulsation
Below is the single-phase flow phenomena; while Two-phase flow structure is in chaos and never as clean as that of single phase turbulent
flows. Void drift: Driving force? Probably due to anisotropic structure of two-phase flow between
subchannels – analogy from the single-phase secondary flow motion to Global Flow Pulsation
Section
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
24 Take into account the oscillation:
− Frequency, amplitude and wave length of the cross flow oscillation; and
− Traveling speed
both to be obtained by DNS/LES/URANS
Example for comparison:
Enhanced Mixing by Flow Pulsation
0
0
:
( ) sin 2 ( )
25
0.0125 (80 )
5% ax
The cross flow velocity is assumed to follow
z tu z u
Tcm
T s Hz
u w
Turbulence and Conduction only
Outlet temp. SC i and j are 438C and 462 C
Time average
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaAssumptions to Make CV Formulation and Analysis Tractable
Assumptions Liquid is treated incompressible; Transport due to turbulence is dealt with by effective eddy diffusivity
approach; Energy dissipation is neglected; and
Spatial fluctuations on the subchannel control volume surface areaare not accounted for.
Area average + spatial non-uniformity
Always we assume uniform distribution on the CV surfaces: this is not aunique problem in SCA
Three 1/6 rods share one subchannel
, ) ( , )x y x y ≮ ≯
+ Cross correlation termsu u ≮ ≯ ≮ ≯≮ ≯≮ ≯
u ≮ ≯ ?
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Flow distribution is never uniform on the SC CV surfaces
u ≮ ≯ ?
Cannot be neglected in many cases
Cross-flows
Axial-flows
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaSubchannel sub-division idea
A B
C
To represent better fuel cladding temperature distributionsHowever, this is no longer SCA …
Three 1/6 rods share one subchannel
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaConstitutive Equations
State Variables: a set of subchannel equations are solved for:
and fuel temperature distributions
Constitutive eqs must be expressed in terms of averaged state variables!
, ,
in case of the two-fluid modelk k k k k ku v w and e or T
k L and G
≮ ≯ ≮ ≯, ≮ ≯
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Physical models (constitutive models) for pressure drop ---- experimental correlation basis
P for 1D System and Subchannel Analysis code
P modeling state of the arts SOK* Responses
1D SystemAnalysis
Bundle P for Axial Flow (Re high) good
Bundle P for Axial Flow (Nat Conv) fair
Sub-channel Analysis
Subchannel P for Axial Flow (Re high) good
Subchannel P for Cross-Flow (Low Re) fair CFD
Subchannel P for Buoyancy Dominated Flows
Large Uncertainties
CFD
* State of knowledge
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaInter-Subchannel Exchange Mechanisms
Subchannel exchanges are a result of cross-flow convection as well as molecular and turbulent diffusion mechanisms. Convection includes those due to divergence cross flow, global flow pulsation and void drift.
There are several mechanisms of exchange phenomena that are known even in the absence of pressure gradient in the cross-flow direction such as GFP and VD.
Void drift phenomena of two-phase flows may be calculated by a diffusion model on the long-time-average basis but are not a diffusion phenomenon but a result of cross-flow convection to the equilibrium void redistribution
Single-phase flow Two-phase flow
1. Divergence Cross Flowpressure drop correlation
Two-phase multiplier applied
2. Turbulent Diffusioneddy diffusivity model (turbulent mixing)
Two-phase multiplier applied
3. Global flow pulsation phenomena
Void drift phenomena0
dp
dx
0dp
dx
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Heat Transfer Coefficient
In general, Nu=f(Re, Pr): Dittus-Boetler equation and its variants
Sodium flows in rod bundles: FFTF-CRBRP
Non-metallic flows in rod bundles
5.0 3.8 0.864.0 0.16 / 0.33 / ( /100)
/ , Re Pr
Nu P D P D Pe
where
Nu hD k Pe
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
WIRE-WRAPPED FUEL ROD BUNDLE
Bundle vs subchannel pressure drop model
Forcing function model and Distributed Resistance Model
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaPressure Drop Correlations
Subchannel p/z for wire-wrapped fuel subassemblies: Novendstern
Chen-Todreas Model
S/A inlet outlet form losses
1 smoothf Mf
0.8856.94 0.086
10.124 2.239
29.7( / ) Re1.034
( / )/
P DM
H DP D
Note: these integral bundle-based models do not account for the distributed resistance, in subchannel-wise
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
wj
Wire Forcing Function model and DRM
Wire forced flow assumption --- no consideration for momentum conservation; in the classic COBRA-IIIC
===================================
Was replaced by DRM
Built in subchannel analysis code ASFRE at PNC (JAEA)
DRM: Ninokata Efthimiadis and Todreas(Nucl Eng Des 104 (1987) 93-102
COBRA-IIIC: A digital computer program for steady state and transient thermal-hydraulic analysis of rod bundle nuclear fuel elements, D.S. Rowe, BNWL-1695, 1973tan ,
: wire wrap angle; : wire wrap pitch
ij j
Du w
PP
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Time- and CV-Averaging N-S Eqs results in DRM and Turbulent Mixing (Once again)
Distributed Resistance at the Fluid-Solid Interface
1[ ] [ ]
fs
DRAf
F p pn n dAV
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaDistributed Resistance Model for Wire Spacers 1
Ninokata et al., NED, 1987.v
w
u
Conventional flow resistance models: Employ average characteristics of SCs
Complete separation of axial and lateral components
Distributed resistance models: Employ local characteristics of each SC
Proper decomposition of flow resistance forces
Solve N-S eq with these forces; As a result of force balance:, the velocity is obtained
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaDistributed Resistance Model for Wire Spacers 2
Modified Blasius type friction model by Rehme, …, etc
For predominantly lateral flows, Gunther-Shaw models, taking account the wire position (Fig), etc.
For predominantly axial flows, modified G-S correlation by drag coefficient for the free stream flow over a cylinder
A TR WF and F components
L NR WF and F components
Need to provide local flow resistance models to
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Other Modeling of Wire-Wrap Effects – Momentum Source Model (Rui Hu, ANL)
38
φ
θ
npn
nn
nt
perpendicular to the paper plane
x
z
nt
nn
pin surface
Fill the gap between the traditional CFD method and the sub-channel method
Reduce the geometric mesh complexity
Directly derived from N-S Equations, and dependent on local velocity
Mesh flexible for implementation
The MSM blocks the flow in the wire normal directions:
fn
Cvn (vn vt cos vpn )
dw
(nn
) fpn
Cvpn (vn vt cos vpn )
dw
(npn
)
ft
f frvt
2
2dw
(nt
)
Introduce additional friction in the wire tangential direction:
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Number of Calculation FEM Meshes: 2.8 Million
Turbulence Model: k-/k-e Model for momentum/energy
Convection Term: SUPG (Brooks & Hughes)
Wall Boundary: Wall Function
(Reichardt Eq./Kader Eq.)
Calculation Mesh (Horizontal Section)
Simulation by the FEM code SPIRAL (JAEA)(Courtesy of Dr. Hideki Kamide, JAEA)
39
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
40
Axial Velocity
Lateral Velocity Streamline
Measured Temperature
Predicted Temperature
4.5mm below Top of Heated Region
Simulation by the FEM code SPIRAL (JAEA)(Courtesy of Dr. Hideki Kamide, JAEA)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
ハンドリングヘッド
上部スペーサパッド
中間部スペーサパッド
下部スペーサパッド
エントランスノズル
ワイヤースペーサ
炉心燃料棒
ラッパー管
<水平断面図>
<全体図>
Wire-SpacerFuel Pin
Wrapper Tube
Entrance Nozzle <Cross Section>
Fuel Assembly
Difficulties in Mockup Experiments Complicated Geometry, High Cost Numerical
Approach
Detailed Temperature Distributions
Structural Integrity of Fuel Pin
- Normal Operation
(Full Power ~ N.C. Decay Heat Removal)
- Transient / Accident Conditions (Local Fault, Boiling)
- Deformed Geometry Condition (Swelling, Creep)
Clarification of Thermal-Hydraulic
Phenomena in Fuel Assembly
High Performance Core of Sodium Cooled Fast Reactor (High Burn-up, High Power Density)
Experimental Approach
Subassembly Thermal Hydraulics Engineering (JAEA)(Courtesy of Dr. Hideki Kamide, JAEA)
41
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Fuel Pin Deformation Distribution by BAMBOO
図 3.2-8 169 本燃料ピンバンドル水平断面内冷却材温度分布変化
<変形前> <変形後>
Coolant temperature distribution by a subchannel analysis ASFRE
<Horizontal Cross Section> <Axial Cross Section>
< Before Deformation > < After Deformation >
SCA Applications to Fluid-Structure Interactions (Courtesy of Dr. Hideki Kamide, JAEA)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
TWO-FLUID MODEL SUBCHANNEL ANALYSIS
Formulation
Constitutive equations
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTwo-phase flow patterns in a boiling channel
Flow pattern diagram must be provided
BWR BT/Post-BT
SFR low pressure sodium boiling --- low Pr, high to low q’’
L/G ~ 103
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTwo-Phase Flow Models
HEM 2-fluid 2-field model: F+G 2-fluid 3-field model: F+G+D Multi-component 2-fluid 2- (3-) field model
Judgment of CHF is made based on experimental correlations Film dryout CHF can be well predicted
A number of constitutive equations (closure relations or physical models) required
Inter-phase transfer phenomena Interface area concentration Well-posed vs. ill-posed (pure mathematical)
q”
F G
D
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaTwo-fluid model (RANS)
PDE DNS
• Local instantaneous PDE• N-S eqs• Energy eq
• Local instantaneous PDE• N-S eqs• Energy eq
Time average
• (LES to RANS)• Time-smoothed-out conservation eqs with interface tracking • (LES to RANS)• Time-smoothed-out conservation eqs with interface tracking
Leibnitz rule & local CV
average
• Local volume average with Leibnitz rule applied to moving boundary surface at the phase interface (L-G)
• Two-fluid model equation system with void fraction (RANS level)
• Local volume average with Leibnitz rule applied to moving boundary surface at the phase interface (L-G)
• Two-fluid model equation system with void fraction (RANS level)
Large SC CV average with Gauss
• 2-fluid 2-field 10 equations for 3-D representation• Physical models (closure equations)• 2-fluid 2-field 10 equations for 3-D representation• Physical models (closure equations)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaNeglected local effects
Higher order tensor terms appear as a result of the space averaging in terms of the cross-products of deviations from the area averaged state variables.
Cross products terms are those with deviations of void fraction distribution, cross-flow and three axial-flow components of each phase and their time variations. Normally these terms are ignored.
, ) ( , )k k kx y x y ≮ ≯
+ Cross correlation termsk k k k k ku u ≮ ≯ ≮ ≯≮ ≯≮ ≯
k ku ≮ ≯ ?
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaNeglected local effects
Result in a loss of local information in the subchannel formulation. The information of these deviations should be obtained by experiment for the moment. Otherwise by CMFD.
These higher order tensor terms are to be expressed in terms of main subchannel state variables (surface averaged velocity components, pressure, density of each phase.) This step is undone yet.
k ku ≮ ≯ ?
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Numerical Solution Scheme: Needs implicit coupling for low pressure boiling
Mass, Momentum, Energy of each phase
Successive substitution-type iteration scheme
− Easy coordination among separate subprogram developments;
− User feedbacks easily implemented
should, nonetheless, be avoided.
All information should be concentrated in the Poisson eqmatrix
− Almost prohibits collaborations among code subprogram development teams
Mass
MomentumEnergy EOS
Mass
Momen-tumEnergy
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaSolution by Modified ICE (TRAC)
All equations are dicretized (averaged per computing cell, mesh, FE or CV)
Key 1. Need to implicit coupling of mass, momentum and energy
Key 2. Linearized convective terms
Key 3. ICE: Implicit Continuous Eulerian
Linearization of all the momentum equations
Obtain these linearized (new) velocity-pressure gradient equations and
substitute into both mass & energy equations
Eliminate energy terms from both equations and
construct a Poisson equation for the
pressure distribution
Solve the Poisson equation
Obtain new velocity as well as temperature
distributions [end of one time step]
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaPhysical models
Experimental correlations for pipes of various diameter
Most physical models of subchannel analysis are based on modified version
Major Physical models SOK
Wall-fluid momentum exchange (W-F), (W-G) … P models good
Wall-fluid energy exchanges …. hWF and hWG (Dittus-Boelter) good
Interfacial momentum exchanges … Fanning friction factor concept at the interface; Wallis 1D model for annular flows; Chawla-Ishii for slug-bubbly
Large uncertainties
Interfacial mass exchanges --- kinetic theory of gases; … evaporation-condensation process is always under non-equilibrium condition
Relativelygood
Interfacial heat transfer: q”GI and q”FI, interface htc Fair
Interfacial area and Flow regime identification in two-fluid model representation
Large uncertainties
q”
F G
D
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaVoid distribution in a channel two-phase flow
Simply described by
and S (= wG/wF) with quality x
and S are dictated by what?
Inter-phase Shear I
wG/wF
SAs I
At z0S and
Vapor concentration
z0
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaVoid distribution in a channel two-phase flow
Simply described by
and S (= wG/wF) with quality x
and S are dictated by what?
As I
relative to W
Inter-phase Shear I
wG/wF
SWall
shear W
f
z0
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaVoid distribution in a channel two-phase flow
Simply described by
and S (= wG/wF) with quality x
and S are dictated by what?
Inter-phase Shear I
wG/wF
SWall
shear W
f
As W or f
with the other fixed
z0
z0
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Inter-relationship among void fraction, wall friction and interphase friction
The void concentration distribution is very sensitive (low pressure boiling, in paticular):
Steady-state One-dimensional annular flow(In the following, the subscripts G and F are denoted by g and f)
q”
F G
D
(Triangular relationship)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Two-phase flow multiplier, interfacial friction, void fraction and slip ratio
(I) For constant (Cf)I=0.07 (II) For constant f=16;S and are very sensitive to small changes in f and (Cf)I, respectively
(III) As (Cf)I increases; or(IV) As liquid wall shear or f increases, at the location tends to increase or decrease, respectively
Eq. (c)
Eq. (b)
H. Ninokata and A. Deguchi, Assessment of the physical models in a two-fluid model code and interpretation of experiments, Nucl. Energy, 1989, 28, No. 3, 161-170
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
References to two-fluid model boiling flow subchannel analysis
H. Ninokata and A. Deguchi, Assessment of the physical models in a two-fluid model code and interpretation of experiments, Nucl. Energy, 1989, 28, No. 3, 161-170
Ninokata, H., Okano. T., SABENA: Subassembly boiling evolution numerical analysis, Nucl. Eng. Des. 120, 349-367, 1990.
Development of the NASCA code for predicting transient BT phenomena in BWR rod bundles, H. Ninokata, M. Aritomi, T. Anegawa, et al., Proc. OECD-CSNI Workshop on Advanced Thermal-Hydraulic and Neutronic Codes: Current and Future Applications, Barcelona, April 10-11, 2000.
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
TWO-FLUID REPRESENTATION WITH AN ADDITIONAL DROPLET FIELD
BWR new fuel bundle design – axially heterogeneous, with burnable poison (Gd) distributed concentration,
New spacer design
BT and post BT phenomena
Need to predict droplet motion and their influences
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaDroplets
Lagrangean treatment not appropriate
Another field added for a group of droplets
Size distribution
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
60
Liquid Film Model
Entrainment due to boilingin a film
Entrainmentdue to wavedisturbances
Deposition
Dryout
NASCA
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaInfluences of Spacers
61NASCA
Three major depositionenhancement effects due to
◆ turbulence generatedbehind spacers
◆ run-off of droplets collected on spacers
Enhanceddeposition
Turbulence Runoff
Enhanceddeposition
◆ change in flow directions due to spacer
Spacer
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Convection Diffusion Sources
Mass( )K K Ku F-G
D-GF-D
Ev.Cord
Entr & Deposit
MomentumF-GD-GF-D
F-GD-GF-D
( )K K K Ku u
( )K K K Ku e Energy
Distributed ResistanceTurbulent Mixing
Distributed heat transfer
Turbulent mixing
Two-fluid three-field formulation -1
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaExamples
NASCA
H. Ninokata, T. Anegawa, M. Aritomi, T. Hara, H. Kamo, S. Kusuno, K. Mishima, S. Morooka, K. Nishida, M. Sadatomi,
A. Sou, Y. Yabushita, and Y.Yamamoto
DEVELOPMENT OF THE NASCA CODE FOR PREDICTING TRANSIENT BT
PHENOMENA IN BWR ROD BUNDLES
NASCA
OECD/CSNI Workshop on Advanced Thermal-Hydraulicsand Neutronic Codes: Current and Future ApplicationsBarcelona, Spain, April 10-13, 2000
Introduction
• NASCA (Nuclear Reactor Advanced Sub-Channel Analysis)
• is aimed at simulating thermal-hydraulics phenomena in BWR rod bundles during transients, in particular BT and post BT
• is useful to understand the BT and post BT phenomena; and
• is going to be used as a standard code in Japanese utilities and BWR vendors.
65NASCA
NASCA Code Development• On the basis of available and well established concepts;• Two-fluid three-field subchannel formulation;• Basic version completed 1996 [ISSCA-4]• Major efforts (1997- ) centered on modeling:
- two-phase turbulent mixing and void drift;- spacer effects; - BT and post BT phenomena; and- validation for separate effects experiments as well as for bundle integral experiments.
66NASCA
Two-phase flow turbulent mixing
• : 1, w, h; k : G or L
• TM : Two-phase multiplier for turbulent mixing, due to Beus / Kelly and Kazimi
• Model verification for equilibrium two-phase flows in the inter-connected two-subchannel test sections (Kumamoto University)
67
jkkikkkgapTMTP
kij
,
,
NASCA
,
kTM
TP TM SP
Void drift• Void re-distribution to equilibrium distribution• Mechanisms not clearly identified• Modeling based on the inter-connected two-subchannel experimental data
(Kumamoto University)
68
EQjkkkikkkjkkkikkkVD
VDTPkij
,
)(
k
VDSPVDTP ,
iA
jA
jkkkikkkiAjAjGiG
jGiG
aKEQjkkkikkk )()(
)/(
(Lahey’s Model)
NASCA
Two-phase multipliers
• For Mixing TM < 5 (Beus / Kelly-Kazimi)
• For VD VD = 15 for churn-turbulent flow regime (0.5 < < 0.7)
69NASCA
Results for GE3x3 bundle experiments
70
18.7mm58.8mm
Heater rod(φ14.5mm)
Channel BoxCorner
Side
Center
Fig. 1 GE 3x3 rod bundle cross section
0
500
1000
1500
2000
2500
3000
3500
0 500 1000 1500 2000 2500 3000 3500
Experiment (kg/m2s)
Bundle average mass flux: 651-2672 kg/m2s
CornerSideCenter
Cal
cula
tion
(k
g/m
2s) Inlet subcooling
1174kJ/kg
Fig. 2 Comparison of subchannel mass fluxbetween calculation and experiment
NASCA
71
0.0
0.1
0.2
0.3
0.4
0.5
0.0 0.1 0.2 0.3 0.4 0.5
Quality [Experiment] (-)
Qual
ity
[Cal
cula
tion]
(-)
CornerSide
Center
+0.024
-0.024
Results for GE3x3 bundle experiments
NASCA
BT and Post BTDryout Phenomena
• Vaporization of liquid film in the forced convection mode
• Droplet entrainment and deposition models• Validation for single tube experiments including
Bennet’s cold patch experiments• Check consistency with the CISE correlation• Need to improve a dependency of the critical power
on the heated length (upstream effects)• Entrainment due to boiling in the liquid film (Ueda’s
model); tuning against the Bennet cold patch tests
72NASCA
BT and Post BTRewetting phenomena
• In general very complicated• Hydrodynamic-limited regime in annular-mist flows
73
For simplicity, in NASCA, Rewetting = Liquid film flow recovery
NASCA
BT and Post BTSpacer model tuning
• For BT/post BT tests in an annular channel with simulated spacers (Yokobori, et al)
• Tuning of the weighting factors for the three major independent deposition models with respect to BT timing, location, BT duration and rewetting
• Finally for 4x4, 8x8 bundle experiments
74NASCA
Results of Bundle Calculations
75NASCA
0.9 0.9 0.9 0.9
0.9 1.3 1.3 0.9
0.9 1.3 1.3 0.9
0.9 0.9 0.9 0.9
T/C
3.7m
Spacer
0.0 0.5 1.0 1.5Relative power
Axi
al p
ositi
on
76
Comparisons of steady-state critical power
0
1
2
3
0 500 1000 1500
質量流束[kg/m2/s]
限界
出力
[MW
]
測定データ
NASCA ・Good agreement between cal. and exp.
・Dependency on massflow rate
Measurement
Mass flux
Crit
ical
pow
er
NASCA
77
Too long BT duration by cal.Conservative pred. for low mass fluxes NASCA
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定第7スペーサ上流位置
0
0.2
0.4
0.6
0.8
1
1.2
10 15 20 25 30時間[s]
流動
パラ
メー
タ相
対値
流量出力
圧力
入口冷却材温度
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定
第6スペーサ上流位置
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定
第5スペーサ上流位置
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定第4スペーサ上流位置
Comparisons for Flow Transient
Time (s)Time (s)
Rod
sur
face
tem
pera
ture
[K]
Rod
sur
face
tem
pera
ture
[K]
Inlet coolant temp
pressurePower
Flow
Upstream of 4th spacer
Upstream of 5th spacer
Upstream of 7th spacer
Upstream of 6th spacer
meas.
meas.
meas.
meas.
78
Good agreement for BT timing, temperature behaviors and rewetting NASCA
0
0.5
1
1.5
2
10 15 20 25 30時 間 [s]
流動
パラ
メー
タ相
対値
出 力
圧 力
流 量入 口 温 度
500
600
700
800
900
10 15 20 25 30時 間 [s]
燃料
棒表
面温
度[K
]
NASCA
測 定第 7スペー サ 上 流 位 置
500
600
700
800
900
10 15 20 25 30時 間 [s]
燃料
棒表
面温
度[K
]
NASCA
測 定第 6スペー サ 上 流 位 置
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定第5スペーサ上流位置
500
600
700
800
900
10 15 20 25 30時間[s]
燃料
棒表
面温
度[K
]
NASCA
測定第4スペーサ上流位置
Comparisons forPower Transient
Power
FlowInlet coolant temp
pressure
Time (s)Time (s)
Rod
sur
face
tem
pera
ture
[K]
Rod
sur
face
tem
pera
ture
[K]
Upstream of 4th spacer
Upstream of 5th spacer
Upstream of 7th spacer
Upstream of 6th spacer
meas.
meas.
meas.
meas.
79
Changes in liquid volume fraction and fuel rod surface temperature distributions (1/6) [Power transient]
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
Steady-statebefore transient
T=16 s
F
ilm
Axial Node PositionR
od S
urfa
ce T
emp
[K]
Axial Node PositionNASCA
Spacer position
Spacer position
80
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
T=17 s
Film thinning
NASCA
Changes in liquid volume fraction and fuel rod surface temperature distributions (2/6) [Power transient]
Rod
Sur
face
Tem
p[K
]
F
ilm
Axial Node Position
Axial Node Position
Spacer position
Spacer position
81
NASCA
Changes in liquid volume fraction and fuel rod surface temperature distributions (3/6) [Power transient]
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
T=18 s
BT BT BT
Film Film
Spacer position
Spacer position
Rod
Sur
face
Tem
p[K
]
Axial Node Position
F
ilm
Axial Node Position
82
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
T=19 s
Film thickening
NASCA
Changes in liquid volume fraction and fuel rod surface temperature distributions (4/6) [Power transient]
Rod
Sur
face
Tem
p[K
]
F
ilm
Axial Node Position
Axial Node Position
Dryout
Spacer position
Spacer position
83
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
Thick film propagation
Film thickness increase downstream of a spacer
Rewetting due to the propagation of a film down-stream of the spacer
T=20 s
NASCA
Changes in liquid volume fraction and fuel rod surface temperature distributions (5/6) [Power transient]
Rod
Sur
face
Tem
p[K
]
F
ilm
Axial Node Position
Axial Node Position
Spacer position
Spacer position
84
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
15 20 25 30 35 40 45 50
軸方向ノード位置
液膜
体積
割合
スペーサ位置
500
550
600
650
700
750
800
850
900
15 20 25 30 35 40 45 50
軸方向ノード位置
燃料
棒表
面温
度[K
]
スペーサ位置
Rewetting completed
T=21 s
NASCA
Changes in liquid volume fraction and fuel rod surface temperature distributions (6/6) [Power transient]
Rod
Sur
face
Tem
p[K
]
F
ilm
Axial Node Position
Axial Node Position
Spacer position
Spacer position
85
Summary of Bundle Calculations• Critical power prediction:flow rate dependency
• Steady-state critical power: good agreement with experiment
• Flow transient: good agreement for BT timing;late for rewetting
• Power transient: good agreement for BT timing; earlier rewet timing
• Additional subdivision of subchannels yields good results for high burnup 8×8 fuel rod bundles with unheated water rod(s) [not shown here]
• Good results obtained for NUPEC 8x8 rod bundle experiments [not shown here]
NASCA
DEVELOPMENT OF THE NASCA CODE FOR PREDICTING TRANSIENT BTPHENOMENA IN BWR ROD BUNDLES
NASCA
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaRecent results
Application of NASCA to OECD/NRC NUPEC BWR Full Size Fine-Mesh Bundle Test, K. Nozaki, et al. NURETH-12 (2007)
Analytical Study on Boiling Transition under Flow-Power Oscillating Condition for Hyper ABWR, K. Nozaki, et al, 13th JSME Symposium on Power and Energy Technology, June (2008)
Multi-component multi-phase flow subchannel analysis
Relevant to Nuclear Reactor Thermal Hydraulics
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
MULTI-PHYSICS PHENOMENA MODELING
(the following notes consist of information from JAEA and R&D results at the Tokyo Institute of Technology)
Fuel S/A degradation and CDAs
Calculation quality depends on the physical models
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaComputational model
SAS/SIMMER code system for CDAs since 1970’s KAMUI – for fuel S/A degradation by subchannel analysis
Multi-component multi-phase flow Multi-component multi-field formulation In case of fuel S/A degradation: 3 components (fuel, steel, Na), 3-phases and 2-
or 3-velocity fields (mixture velocity fields):
Mixture fields required mixture material properties (viscosity, heat capacity, conductivity, .. etc.)
Phase interfaces --- topology
Component Solid-phase Liquid-phase Vapor-phase
Fuel X X X
Steel X X X
Sodium X X
2 velocity fields Mixture velocity field Gas-phase v
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaIn-Pile Experiments
International cooperation: PNC (JAEA)-DeBeNe-CEA, PNC(JAEA)-USNRC/DOE, …, etc CABRI
SCARABEE
TREAT, SLSF
EBR-II
…
IGR-EAGLE (Experimental Acquisition of Generalized Logic to Eliminate criticalities)
Phebus
NSRR, …
91
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
CEC
Reactor core
Coolingwater cavity
Control rod channel
Cross-section of IGR core(NNC in Kazakhstan)
CDA Evaluation Methods & Mitigation Measures- Upward Discharge Experiment in EAGLE Project of JAEA -
PERFORMANCEMax. thermal neutron flux density:Max. thermal neutron fluence:Min. half-width of pulse:Max. energy release:Central Experimental Channel (CEC):
7×1016 n/cm2s3.7×1016 n/cm2
0.12 s5.2 GJφ228mm×L3825mm
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Insertion of test section into IGR core
Test section for upward discharge
Inner duct
SA can wall
Cross section
Core
Discharge path
Closed end
FAIDUS option (reference for JSFR)
IGRcore
Fuel pins to be molten
Simulated core part
Discharge path
Simulated upper plenumSodium
CDA Evaluation Methods & Mitigation Measures- Upward Discharge Experiment in EAGLE Project of JAEA -
93
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Validation of subassembly degradation and core meltdown, relocation models by TIT
CABRI hodo-scope data
SCARABEE TIB temperature flow data
TREAT/SLSF
ACRR
1
2
3
4 5
Flow blockage at the start of transient
Coolant Wall
Fuelpin
6 5 4 3
1 2
14 13 12 11 10 9 8 7
19 18 17 16 15
Fissile length 60cm
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Multi-component multi-field model application to SCARABEE-BE+2 and APL Experiments: VIDEO
Subchannel analysis for interpretation of SCARABEE in-pile experiments by the KAMUI code
KAMUI BE+2 : Instantaneous total inlet blockage Fuel assembly at full power What happened if the inlet nozzle were totally blocked and if
scram failed?
KAMUI APL : Unprotected loss of flow Fuel assembly at full power What happened if the pump stopped and if the scram failed?
The multi-fluid multi-phase subchannel analysis code KAMUI for subassembly accident analysis of an LMFR, Fumio Kasahara and Hisashi Ninokata, Journal of Nuclear Science and Technology, 37, No.8 (2000) 654-669.
Agreement? Excellent, good, fair, poor? Trend agreement is important but meaningless if the users don’t try to catch physics
To minimize subjective judgment on modeling multi-physics, we need:
Identification and estimation of uncertainties
Only visual comparisons are not sufficient
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
H. Ninokata and E. Merzari 97
How do you catch physics? I. In case of DNS or LES
So much information from DNS or LES
Many new phenomena, detailed turbulent structure through visualization …
Done by visualization thanks to rapid progresses in CG technology …. Fancy -- but it’s a subjective approach
Objective education techniques, to avoid possible controversy and to identify nature and significance of the structure
Proper Orthogonal Decomposition (POD) Technique (Elia Merzari, NED 241 (2011) 4621-4632; PhD thesis at Tokyo Inst Tech)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
H. Ninokata and E. Merzari 98
How do you catch physics? II. In case of multi-physics simulation
As more multi-physics involved, more complex calculation system with so many physical models representing the interactions
Physical models are based on known knowledge and a result of assumptions, approximations, compromises
With the CV sizes larger, more uncertainties
Comparisons must be done with experiment (and theory if any), Done by visualization – Not sufficient
Needs to identify modeling uncertainties, to avoid possible controversy and to identify nature and significance of the structure
An attempt to quantify uncertainty
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Uncertainty identification in physical modeling -1 Marco Pellegrini, PhD thesis, TIT (2012)
Erroneous example: stratification in sodium flow
turbulence heat flux model should take into account the gravity
We would like to know how erroneous the predictions are when the turbulent heat flux is modeled w/ or w/o gravity effects
We follow the Bayesian rule P(B|A)={P(A|B)*P(B)}/P(A)
Prior probability P(B) [calculation] can be updated to P(B|A) with P(A), probability of A by experimentation, where P(A|B) a likelihood function;
Noted that the likelihood P(A|B) is given a’priori but subjective; should be improved by optimal estimation-control theories
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Uncertainty identification in physical modeling -2 Marco Pellegrini, PhD thesis, TIT (2012)
Assume a degree of being subjective for a certain model, P(B),
P(B) could be updated based on a direct comparison of the model prediction with experiment, to P(B|A)
By carrying out as many as calculations as possible with different model parameter values, we obtain P(B|A)
P(B|A) accounts also for the uncertainty in the experimental results P(A) and provides statistical information on the mean value, standard deviation, tolerance limits, ..
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Uncertainty identification in physical modeling -3 Marco Pellegrini, PhD thesis, TIT (2012)
A Simple Example:
Suppose the model for the turbulent heat flux in a CFD code is expressed in terms of velocity gradient (C1) and the gravity effect (C3)Run as many cases for C1 and C3 as possible (Monte Carlo or economical Latin Hypercube Sampling) to construct a response surfaceMean value of C1 and C3 represent optimal values while the standard deviation could be interpreted as a subjective degree of belief in C1 and C3 model parameters.C1 trustable; C3 questionable …….. Note: this is just an example
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-AtlantaComments
• In practice, validation of engineering multi-physics phenomena is likely to be made on rather qualitative basis, often relying on many subjective judgments in comparison with the results from large-scale integral tests or mock-up experiments
• In validation processes, although an eventual subjective judgment cannot be ruled out but should be made minimal. To make it more quantitative and rational, a proposal has been made of the identification of errors and/or uncertainties inherent in computations based on the Bayesian rule
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
103
CFD Applications to LMFRs, LWRs
Replacement of subchannel analysis by LES would be desired eventually --- Less modeling required
Computational results are substantially dependent on:
Specified mesh schemes and boundary conditions;
Numerical schemes in general, model selections which require users’ knowledge on turbulence and expertise in creating specific simulation models and interpreting the results of the simulations.
Note: there is no universal approach in CFD other than directly
solving the N-S equations (DNS)
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Principle of physics modeling for subchannel analysis
Physical models incorporated into the subchannel analysis (as well as other computational methods) must be as simple as possible
Only small scale channel experiments are good enough Similarity is better to be established
Up to now, classical correlations have been the substantial ingredients of subchannel analysis
However, most correlations were developed when a large CPU was not available
Mechanistic modeling Need large expensive mock-ups? No Clarify mechanisms of the phenomena, identify key parameters
and express them in terms of the state variables of SCA Some regulatory authorities, project leaders or management often
don’t understand the importance and therefore the need from computations
H. Ninokata CeSNEF - Dipartimento di Energia
2013.06.18TAA lecture
ANS-Atlanta
Profile of Hisashi NINOKATABA in physics (1970), MS (‘72) and Dr. Eng (‘77) in NE, The University Tokyo
TEPCO (’77‐’80), PNC (’80‐’93), TITech (’93‐’12), POLIMI (’12‐ )
Rotary International Fellowship Exchange Student to MIT NE where he has started his rod bundle TH since 1972: MIT Mixing Project ; VELASCO code;
… in ’80s to ’90s,
Involved in Liquid Metal Boiling Working Group (LMBWG) organized by JRC Ispra, KfK, CEA and PNC in ’80s, IAHR Liquid Metal WG
Fast breeder TH and safety: subchannel analysis code (ASFRE and SABENA) development for Na boiling; COMMIX‐AQUA code development (ANL);
SIMMER‐AFDM development (LANL); CABRI project
… from ’90’s to present,
BWR BT and post BT analysis; SMR design (IRIS, SFR)
NRx core neutronics‐TH design, self‐controllability, Risk‐informed Design with PSA
CFD applications to turbulent flows in fuel subassemblies
Fukushima Dai‐ichi accident evaluation in AESJ, ANS, …communication with the public
ENDand
Thank you