recent progress in integrate code taskfukuyama/pr/2007/... · 2007-07-23 · us-japan jift workshop...
TRANSCRIPT
US-Japan JIFT Workshop onIntegrated Modeling on Fusion Plasmas
ORNL, Oak Ridge, USA2007/01/29
Recent Progress in Integrate Code TASK
— Integrated Modeling of —
– RF Heating and Current Drive in Tokamaks —
A. FukuyamaDepartment of Nuclear Engineering, Kyoto University
in collaboration withS. Murakami, M. Honda, A. Sonoda, T. Yamamoto
Outline
• Integrated Simulation of Fusion Plasmas
• Progress of TASK Code Development
• Self-Consistent Analysis of RF Heating and Current Drive
• Full Wave Analysis Including FLR Effects
• Future Plan of Integrated Modeling
• Summary
Burning Plasma Simulation
Broad time scale:100GHz ∼ 1000s
Broad Spatial scale:10 μm ∼ 10m
1mm
1m
1Hz1kHz1MHz1GHz
LH
EC
IC
MHD
One simulation code nevercovers all range.
Energetic Ion Confinement
Transport Barrier
MHD Stability
Turbulent Transport
Impurity Transport
SOL Transport
Divertor Plasma
Core Transport
Non-Inductive CD
RF Heating
NBI Heating
Neutral Transport
Plasma-Wall Int.
Blanket, Neutronics, Tritium, Material, Heat and Fluid Flow
Integrated simulation combining modelingcodes interacting each other
Integrated Tokamak Simulation
Fixed boundary
Free boundary
Time evolution
Diffusive transport
Dynamic transport
Kinetic transport
MHD stability
Kinetic stability
Ray/Beam tracing
Full wave
Kinetic transport
Fluid transport
Orbit following
Equilibrium:
Core Plasma Transport:
Macroscopic Stability:
Wave Heating Current Drive:
Peripheral Plasma Transport:
Turbulent Tranport
Plasma Deformation
Neutral Transport
Impurity Transport
Velocity Distribut. Fn
Neoclassical Tranport
Wave Despersion
Magnetic Island
Turbulent Transport
Plasma-Wall Interaction
Integrated Simulation
Anisotropic Pressure
Energetic Particles
Control model
Diagnostic model
Neutral Transport
Impurity Transport
Radiation Transport
Multi-Hierarchy and Integrated Approaches
Equilibrium:
Core Plasma Transport:
Macroscopic Stability:
Wave Heating Current Drive:
Peripheral Plasma Transport:
Turbulent Tranport
Plasma Deformation
Neutral Transport
Impurity Transport
Velocity Distribut. Fn
Neoclassical Tranport
Wave Despersion
Magnetic Island
Turbulent Transport
Plasma-Wall Interaction
Integrated Simulation
Anisotropic Pressure
Energetic Particles
Control model
Diagnostic modelMulti-Hierarchu Simulation
Device size
Magnetic island size
Ion gyroradius
Collsionless skin depth
Electron gyroradius
Particle distance
NonlinearMHD Simulation
ExtendedMHD Simulation
TurbulenceSimulation
MolecularDynamics
Neutral Transport
Impurity Transport
Radiation Transport
Debye length
BPSI: Burning Plasma Simulation Initiative
Integrated code: TASK and TOPICS
TASK Code
• Transport Analysing System for TokamaK
• Features◦ Core of Integrated Modeling Code in BPSI— Modular structure— Reference data interface and standard data set— Uniform user interface
◦ Various Heating and Current Drive Scheme◦ High Portability◦ Development using CVS (Concurrent Version System)
◦ Open Source: http://bpsi.nucleng.kyoto-u.ac.jp/task/)
◦ Parallel Processing using MPI Library◦ Extension to Toroidal Helical Plasmas
Recent Progress of TASK
• Fortran95◦ TASK V1.0: Fortran95 compiler required (g95, pgf95, xlf95, ifort,...)◦ TASK/EQ, TASK/TR: Fortran95 (Module, Dynamic allocation)
• Module structure◦ Standard dataset: partially implemented◦ Data exchange interface: prototype◦ Execution control interface: prototype
• New module: from TOPICS by M. Azumi
◦ TOPICS/EQU: Free boundary 2D equilibrium◦ TOPICS/NBI: Beam deposition + 1D Fokker-Planck◦MHD stability component: coming
• Self-consistent wave analysis• Dynamic transport analysis: by M. Honda
Modules of TASK
EQ 2D Equilibrium Fixed/Free boundary, Toroidal rotationTR 1D Transport Diffusive transport, Transport modelsWR 3D Geometr. Optics EC, LH: Ray tracing, Beam tracingWM 3D Full Wave IC, AW: Antenna excitation, EigenmodeFP 3D Fokker-Planck Relativistic, Bounce-averagedDP Wave Dispersion Local dielectric tensor, Arbitrary f (u)PL Data Interface Data conversion, Profile databaseLIB Libraries LIB, MTX, MPI
Under Development
TX Transport analysis including plasma rotation and Er
Collaboration with TOPICSEQU Free boundary equilibriumNBI NBI heating
Structure of TASK
Inter-Module Collaboration Interface: TASK/PL
• Role of Module Interface◦ Data exchange between modules:
— Standard dataset: Specify set of data (cf. ITPA profile DB)— Specification of data exchange interface: initialize, set, get◦ Execution control:— Specification of execution control interface:
initialize, setup, exec, visualize, terminate— Uniform user interface: parameter input, graphic output
• Role of data exchange interface: TASK/PL◦ Keep present status of plasma and device◦ Store history of plasma◦ Save into file and load from file◦ Interface to experimental data base
Standard Dataset (interim)
Shot dataMachine ID, Shot ID, Model ID
Device data: (Level 1)RR R m Geometrical major radiusRA a m Geometrical minor radiusRB b m Wall radiusBB B T Vacuum toroidal mag. fieldRKAP κ Elongation at boundaryRDLT δ Triangularity at boundaryRIP Ip A Typical plasma current
Equilibrium data: (Level 1)PSI2D ψp(R, Z) Tm2 2D poloidal magnetic fluxPSIT ψt(ρ) Tm2 Toroidal magnetic fluxPSIP ψp(ρ) Tm2 Poloidal magnetic fluxITPSI It(ρ) Tm Poloidal current: 2πBφRIPPSI Ip(ρ) Tm Toroidal currentPPSI p(ρ) MPa Plasma pressureQINV 1/q(ρ) Inverse of safety factor
Metric data1D: V ′(ρ), 〈∇V〉(ρ), · · ·2D: gi j, · · ·3D: gi j, · · ·
Fluid plasma dataNSMAX s Number of particle speciesPA As Atomic massPZ0 Z0s Charge numberPZ Zs Charge state numberPN ns(ρ) m3 Number densityPT Ts(ρ) eV TemperaturePU usφ(ρ) m/s Toroidal rotation velocityQINV 1/q(ρ) Inverse of safety factor
Kinetic plasma dataFP f (p, θp, ρ) momentum dist. fn at θ = 0
Dielectric tensor dataCEPS ↔ε (ρ, χ, ζ) Local dielectric tensor
Full wave field dataCE E(ρ, χ, ζ) V/m Complex wave electric fieldCB B(ρ, χ, ζ) Wb/m2 Complex wave magnetic field
Ray/Beam tracing field dataRRAY R(�) m R of ray at length �ZRAY Z(�) m Z of ray at length �PRAY φ(�) rad φ of ray at length �CERAY E(�) V/m Wave electric field at length �PWRAY P(�) W Wave power at length �DRAY d(�) m Beam radius at length �VRAY u(�) 1/m Beam curvature at length �
Data Exchange Interface
• Data structure: Derived type (Fortran95): structured type
e.g.
time plasmaf%timenumber of grid plasmaf%nrmaxsquare of grid radius plasmaf%s(nr)plasma density plasmaf%data(nr)%pnplasma temperature plasmaf%data(nr)%pt
• Program interface
e.g.Initialize bpsd init data(ierr)Set data bpsd set data(’plasmaf’,plasmaf,ierr)Get data bpsd get data(’plasmaf’,plasmaf,ierr)
• Other functions:
◦ Save data into a file, Load data from a file, Plot data
Execution Control Interface
• Example for TASK/TRTR INIT Initialization (Default value) BPSX INIT(’TR’)TR PARM(ID,PSTR) Parameter setup (Namelist input) BPSX PARM(’TR’,ID,PSTR)TR SETUP(T) Profile setup (Spatial profile, Time) BPSX SETUP(’TR’,T)TR EXEC(DT) Exec one step (Time step) BPSX EXEC(’TR’,DT)TR GOUT(PSTR) Plot data (Plot command) BPSX GOUT(’TR’,PSTR)TR SAVE Save data in file BPSX SAVE(’TR’)TR LOAD load data from file BPSX LOAD(’TR’)TR TERM Termination BPSX TERM(’TR’)
• Module registrationTR STRUCT%INIT=TR INITTR STRUCT%PARM=TR PARMTR STRUCT%EXEC=TR EXEC· · ·BPSX REGISTER(’TR’,TR STRUCT)
Example of data structure: plasmaf
type bpsd_plasmaf_data
real(8) :: pn ! Number density [mˆ-3]
real(8) :: pt ! Temperature [eV]
real(8) :: ptpr ! Parallel temperature [eV]
real(8) :: ptpp ! Perpendicular temperature [eV]
real(8) :: pu ! Parallel flow velocity [m/s]
end type bpsd_plasmaf_data
type bpsd_plasmaf_type
real(8) :: time
integer :: nrmax ! Number of radial points
integer :: nsmax ! Number of particle species
real(8), dimension(:), allocatable :: s
! (rhoˆ2) : normarized toroidal flux
real(8), dimension(:), allocatable :: qinv
! 1/q : inverse of safety factor
type(bpsd_plasmaf_data), dimension(:,:), allocatable :: data
end type bpsd_plasmaf_type
Examples of sequence in a module
• TR EXEC(dt)call bpsd get data(’plasmaf’,plasmaf,ierr)
call bpsd get data(’metric1D’,metric1D,ierr)
local data <- plasmaf,metric1D
advance time step dt
plasmaf <- local data
call bpsd set data(’plasmaf’,plasmaf,ierr)
• EQ CALCcall bpsd get data(’plasmaf’,plasmaf,ierr)
local data <- plasmaf
calculate equilibrium
update plasmaf
call bpsd set data(’plasmaf’,plasmaf,ierr)
equ1D,metric1D <- local data
call bpsd set data(’equ1D,equ1D,ierr)
call bpsd set data(’metric1D’,metric1D,ierr)
Example: Coupling of TASK/TR and TOPICS/EQU
• TOPICS/EQU: Free boundary 2D equilibrium
• TASK/TR Diffusive 1D transport (CDBM + Neoclassical)
• QUEST parameters:
◦ R = 0.64 m, a = 036 m, B = 0.64 T, Ip = 300 kA, OH+LHCD
ψp(R,Z) jφ Te(ρ, t) j‖(ρ, t)psi(R,Z)
0.2 0.4 0.6 0.8 1.0 1.2-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
rcu(R,Z)
0.2 0.4 0.6 0.8 1.0 1.2-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
TE [keV] vs t
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.9
1.0
0.0
0.2
0.4
0.6
0.8
1.0
1.21.4
1.61.8
2.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
AJ [A/m^2] vs t
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.9
1.0
0.0
0.2
0.4
0.6
0.8
1.0
1.21.4
1.61.8
2.0
0 100
1 106
2 106
3 106
4 106
5 106
6 106
7 106
8 106
9 106
Transport simulation
• OH + off-axis LHCD: 200 kW
• Formation of internal transport barrier (equilibrium not solved)PIN,POH,PNB,PRF,PNF,PEX [MW] vs t
0.0 0.4 0.8 1.2 1.6 2.00.00.10.20.30.40.5
POH
PLH
PTOT
WF,WB,WI,WE [MJ] vs t
0.0 0.4 0.8 1.2 1.6 2.00.000
0.002
0.004
0.006
0.008Wtot
We
TE,TD,<TE>,<TD> [keV] vs t
0.0 0.4 0.8 1.2 1.6 2.00.00.20.40.60.81.0 Te(0)
Ti(0)
IPPRL,IOH,INB,IRF,IBS,IP [MA] vs t
0.0 0.4 0.8 1.2 1.6 2.00.000.050.100.150.200.250.30
IOH ILH
IBS
TE,TD [keV] vs r
0.0 0.2 0.4 0.6 0.8 1.00.0
0.1
0.2
0.3
0.4
0.5
Te
Ti
AKD,AKNCD,AKDWD [m 2/s] vs r
0.0 0.2 0.4 0.6 0.8 1.00
2
4
6
8
χTOT
χi
χCDBM χNC
JTOT,JOH,JNB,JRF,JBS [MA/m 2] vs r
0.0 0.2 0.4 0.6 0.8 1.00.0
0.4
0.8
1.2
1.6 jTOT
jLH jOH
jBS
QP vs r
0.0 0.2 0.4 0.6 0.8 1.00
1
2
3
q
Self-Consistent Wave Analysis with Modified f (u)
• Modification of velocity distribution from Maxwellian
◦ Absorption of ICRF waves in the presence of energetic ions◦ Current drive efficiency of LHCD◦ NTM controllability of ECCD (absorption width)
• Self-consistent wave analysis including modification of f (u)
Self-Consistent ICRF Minority Heating Analysis
• Analysis in TASK◦ Dielectric tensor for arbitrary f (v)◦ Full wave analysis with the dielectric tensor◦ Fokker-Plank analysis of full wave results◦ Self-consistent iterative analysis: Preliminary
• Energetic ion tail formation◦ Broadening of power deposition profile
Momentum Distribution Tail Formation Power deposition
-10 -8 -6 -4 -2 0 2 4 6 8 100
2
4
6
8
10
PPARA
PPERP
F2 FMIN = -1.2931E+01 FMAX = -1.1594E+00
0 20 40 60 80 10010-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
p**2
Pabs with tail
Pabs w/o tail
Pab
s [a
rb. u
nit]
r/a
FLR Effects in Full Waves Analyses
• Several approaches to describe the FLR effects.
• Differential operators: k⊥ρ→ iρ∂/∂r⊥◦ This approach cannot be applied to the case k⊥ρ � 1.◦ Extension to the third and higher harmonics is difficult.
• Spectral method: Fourier transform in inhomogeneous direction
◦ This approach can be applied to the case k⊥ρ > 1.◦ All the wave field spectra are coupled with each other.◦ Solving a dense matrix equation requires large computer resources.
• Integral operators:∫ε(x − x′) · E(x′)dx′
◦ This approach can be applied to the case k⊥ρ > 1◦ Correlations are localized within several Larmor radii◦ Necessary to solve a large band matrix
Full Wave AnalysisUsing an Integral Form of Dielectric Tensor
• Maxwell’s equation:
∇ × ∇ × E(r) +ω2
c2
∫↔ε (r, r′) · E(r′)dr = μ0 jext(r)
• Integral form of dielectric tensor: ↔ε (r, r′)◦ Integration along the unperturbed cyclotron orbit
• 1D analysis in tokamaks (in the direction of major radius)
◦ To confirm the applicability of an integral form of dielectric tensor◦ Another formulation in the lowest order of ρ/L— Sauter O, Vaclavik J, Nucl. Fusion 32 (1992) 1455.
• 2D analysis in tokamaks◦ In more realistic configurations
One-Dimensional Analysis
ICRF minoring heating with α-particles (nD : nHe = 0.96 : 0.02)
Differential form Integral form[V/m]
Ex
Ey
[W/m 2]
PePD
PHe
[V/m]
Ex
Ey
[W/m 2]
PD
PHe
Pe
R0 = 3.0m
a = 1.2m
B0 = 3T
Te0 = 10keV
TD0 = 10keV
Tα0 = 3.5MeV
ns0 = 1020m−3
ω/2π = 45MHz
Absorption by α may be overestimated by differential approach.
2-D Formulation
• Coordinates◦Magnetic coordinate system: (ψ, χ, ζ)◦ Local Cartesian coordinate system: (s, p, h)◦ Fourier expansion: poloidal and toroidal mode numbers, m, n
• Perturbed currentj(r, t) = − q
m
∫du qu
∫ ∞−∞
dt′[E(r′, t′) + u′ × B(r′, t′)
] · ∂ f0(u′)∂u′
• The time evolution of χ and ζ due to gyro-motion⎧⎪⎪⎪⎨⎪⎪⎪⎩χ(t − τ) = χ(t) +
∂χ
∂sv⊥ωc{cos(ωcτ + θ0) − cos θ0} + ∂χ
∂pv⊥ωc{sin(ωcτ + θ0) − sin θ0} − ∂χ
∂hv‖τ
ζ(t − τ) = ζ(t) +∂ζ
∂sv⊥ωc{cos(ωcτ + θ0) − cos θ0} + ∂ζ
∂pv⊥ωc{sin(ωcτ + θ0) − sin θ0} − ∂ζ
∂hv‖τ
Variable Transformations
• Transformation of Integral Variables◦ Transformation from the velocity space variables (v⊥, θ0) to the
particle position s′ and the guiding center position s0.
◦ Jacobian: J =∂(v⊥, θ0)∂(s′, s0)
= − ω2c
v⊥ sinωcτ.
• Integration over τ◦ Integral in time calculated by the Fourier series expansion with
cyclotron period, 2π/ωc
• Integration over v‖◦ Interaction between wave and particles along the magnetic field
lines described by the plasma dispersion function.
Final Form of Induced Current
• Induced current:↔μ −1 ·
⎛⎝Jmn1 (ψ)Jmn2 (ψ)Jmn3 (ψ)
⎞⎠ = ∫ du′
∫du0↔σ (u, u′, u0) ·
⎛⎝Em
′n1 (ψ)
Em′n
2 (ψ)Em
′n3 (ψ)
⎞⎠
• Electrical conductivity:
↔σ (u, u′, u0) = −i(
12π
)72
n0q2
m
∑m′n′
∑l
∫ 2π
0dχ∫ 2π
0dζ exp i
{(m′ − m)χ + (n′ − n)ζ
}↔H (u, u′, χ)
• Matrix coefficients:
H1i =(Q3μ
−11i − u′Q1μ
−12i
) √πk‖vT‖
Z(ηl) +
[−Q1vT⊥
v2T‖
μ−13i +
(1 − v
2T⊥v2T‖
){Q3κ2i + Q1u
′κ1i}] √π
21k‖Z′(ηl)
H2i =(−Q2uμ
−11i + Q0uu
′μ−12i
) √πk‖vT‖
Z(ηl) +
[Q0uvT⊥v2T‖
μ−13i −
(1 − v
2T⊥v2T‖
)u{Q2κ2i + Q0u
′κ1i}] √π
21k‖Z′(ηl)
H3i =
(−Q2
vT⊥μ−1
1i +Q0u′
vT⊥μ−1
2i
) √π
21k‖Z′(ηl) +
[−Q0
vT‖μ−1
3i +vT‖vT⊥
(1 − v
2T⊥v2T‖
){Q2κ2i + Q0u
′κ1i
}] √πk‖ηlZ′(ηl)
Kernel Functions
• Kernel functions
Q⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩
0123
⎫⎪⎪⎪⎪⎪⎬⎪⎪⎪⎪⎪⎭
(u, u′, χ0, l) =∫ 2π
0dλ
1| sin λ|
⎧⎪⎪⎨⎪⎪⎩
1V1
V2
V1V2
⎫⎪⎪⎬⎪⎪⎭
× exp i
[k⊥vT⊥ωc
{(V2 − V1) cosα − (u − u′) sinα
}+ lλ − V
20
2
]
V20 =
(u + u′
2
)2 1
cos2 12λ+
(u − u′
2
)2 1
sin2 12λ
V1 =u − u′
21
tan 12λ− u + u
′
2tan
12λ
V2 =u − u′
21
tan 12λ+u + u′
2tan
12λ
↔h =
1Jω
⎛⎜⎝ 0 −n m′
n 0 i ∂∂ψ
−m′ −i ∂∂ψ
0
⎞⎟⎠
u ≡ s − s0
vT⊥ωc
u′ ≡ s′ − s0
vT⊥ωc
where↔κ =↔μ −1 ·↔g ·↔h ,↔μ is the transformation matrix for (s, p, h)→(ψ, χ, ζ), and↔g is the metric tensor.
Consistent Formulation of Integral Full Wave Analysis
• Analysis of wave propagation◦ Dielectric tensor:∇ × ∇ × E(r) − ω
2
c2
∫dr0
∫dr′
p′
mγ∂ f0(p′, r0)
∂p′· K1(r, r′, r0) · E(r′) = iωμ0 jext
where r0 is the gyrocenter position.
• Analysis of modification of momentum distribution function◦ Quasi-linear operator∂ f0∂t+
(∂ f0∂p
)E+∂
∂p
∫dr∫
dr′E(r) E(r′) ·K2(r, r′, r0) · ∂ f0(p′, r0, t)∂p′
=
(∂ f0∂p
)col
• The kernels K1 and K2 are closely related and localized in the re-gion |r − r0| � ρ and |r′ − r0| � ρ.
Extension to TASK/3D
• 3D Equilibrium:
◦ Interface to equilibrium data from VMEC or HINT◦ Interface to neoclassical transport coefficient codes
• Modules 3D-ready:
◦WR: Ray and beam tracing◦WM: Full wave analysis
• Modules to be updated:
◦ TR: Diffusive transport (with an appropriate model of Er)◦ TX: Dynamic transport (with neoclassical toroidal viscosity)
• Modules to be added: (by Y. Nakamura)
◦ EI: Time evolution of current profile in helical geometry
Road map of TASK code
Fixed/Free Boundary Equilibrium EvolutionEqulibrium
Core Transport 1D Diffusive TR
1D Dynamic TR
Kinetic TR 2D Fluid TR
SOL Transport 2D Fluid TR
Neutral Tranport 1D Diffusive TR Orbit Following
Energetic Ions Kinetic Evolution Orbit Following
Ray/Beam Tracing Beam PropagationWave Beam
Full Wave Kinetic ε Gyro Integral ε Orbit Integral ε
Stabilities Tearing ModeSawtooth Osc.
Resistive Wall Mode
Turbulent Transport
ELM Model
CDBM Model Linear GK + ZF
Diagnostic Module
Control Module
Plasma-Wall Interaction
Present Status In 2 years In 5 years
Systematic Stability Analysis
Start Up Analysis
Nonlinear ZK + ZF
Summary
•We are developing TASK code as a reference core code for inte-grated burning plasma simulation based on transport analysis.
•We have developed a part of standard dataset, data exchangeinterface and execution control and implemented them in TASKcode. An example of coupling between TOPICS/EQU and TASK/TRwas shown, though not yet completed. Some other modules ofTOPICS will be incorporated soon.
• Preliminary results of self-consistent analysis of wave heatingand current drive describing the time evolution of the momentumdistribution function and Integro-differential full wave analysis in-cluding FLR effects have been obtained.
• Further continuous development of integrated modeling is neededfor comprehensive ITER simulation.