Integrated Tokamak Code: TASK

Transcription

1 2010/01/18 Integrated Tokamak Code: TASK A. Fukuyama Department of Nuclear Engineering, Kyoto University Acknowledgments M. Honda, H. Nuga, and BPSI working group

2 Outline 1. Integrated Simulation of Fusion Plasmas 2. Integrated Tokamak Modeling Code: TASK 3. Transport Modeling 4. Source Modeling 5. ITER Modeling 6. Summary

3 Integrated Simulation of Toroidal Plasmas Why needed? To predict the behavior of burning plasmas in tokamaks To develop reliable and efficient schemes to control them What is needed? Simulation describing: Whole plasma (core & edge & divertor & wall-plasma) Whole discharge (startup & sustainment & transients events & termination) Reasonable accuracy (validation by experiments) Reasonable computer resources (still limited) How can we do? Gradual increase of understanding and accuracy Organized development of simulation system

4 Simulation of Tokamak Plasmas Broad range of time scale: 100GHz 1000s Broad range of Spatial scale: 10 μm 10m Energetic Ion Confinement Transport Barrier Core Transport MHD Stability Non-Inductive CD LENGTH 1m 1mm Heating LH EC IC MHD Transport NBI Turbulence Turbulent Transport Impurity Transport SOL Transport RF Heating NBI Heating Neutral Transport 1GHz 1MHz TIME 1kHz 1Hz Divertor Plasma Plasma-Wall Int. One simulation code never covers all range. Blanket, Neutronics, Tritium, Material, Heat and Fluid Flow Integrated simulation combining modeling codes interacting each other

5 Integrated Tokamak Simulation

6 Desired Features of Integrated Modeling Code Modular structure: forflexible extension of analyses Core modules (equilibrium, transport, source, stability) Various levels of models (quick, standard, precise, rigorous) New physics models (easier development) Standard module interface: for efficient development-of modules Interface with experimental data: for validating physics models Unified user interface: for user-friendly environment Scalability in parallel processing of time consuming modules High portability Open source of core modules Visualization included

7 Integrated Modeling Activities Japan: BPSI (Burning Plasma Simulation Initiative) Research collaboration among universities, NIFS and JAEA Framework for code integration Physics integration Advanced computing technology EU: ITM-TF (EFDA Task Force: Integrated Transport Modeling) Code Platform Project: code interface, data structure Data Coordination Project: verification and validation Five Integrated Modeling Projects: EQ, MHD, TR, Turb., Source US: FSP (Fusion simulation project) a part of SciDAC Simulation of Wave Interactions with Magnetohydrodynamics (SWIM) Center for Plasma Edge Simulation (CPES) Framework Application for Core-Edge Transport Simulations (FACETS) ITER: IMEG (Integrated Modeling Expert Group)

8 BPSI: Burning Plasma Simulation Initiative Research Collaboration of Universities, NIFS and JAEA Targets of BPSI Framework for collaboration of various plasma simulation codes Physics integration with different time and space scales Advanced technique of computer science All Japan collaboration Integrated code development

9 Integrated Modeling Code: TASK Transport Analysing System for TokamaK Features Core of Integrated Modeling Code in BPSI Modular structure Reference imprementation of standard data set and interface Various Heating and Current Drive Scheme EC, LH, IC, AW, NB High Portability Most of library routines included (except LAPACK, MPI, MDS+) Original graphic libraries (X11, Postscript, OpenGL) Development using CVS (Version control) Open Source: Parallel Processing using MPI Library (partially)

10 Components of the TASK Code Developed since 1992, now in Kyoto University BPSD Data Exchange Standard data set, Program interface EQ 2D Equilibrium Fixed boundary, Toroidal rotation TR 1D Transport Diffusive transport, Transport models TX 1D Transport Dynamic Transport, Rotation and E r FP 3D Fokker-Planck Relativistic, Bounce-averaged WR 3D Ray Tracing EC, LH: Ray tracing, Beam tracing WM 3D Full Wave IC: Antenna excite, Alfvén Eigenmode DP Wave Dispersion Dielectric tensor, Arbitrary f (u) FIT3D NBI Physics Birth, Orbit width, Deposition PL Utilities Interface to BPSD and profile database LIB Libraries Commonlibraries,MTX,MPI GSAF Graphics Original graphic library for Fortran TASK3D: Extension to 3D Helical Plasmas (NIFS and Kyoto U)

11 Original Structure of the TASK code

12 Present Structure of TASK and Related Codes in BPSI

13 Inter-Module Collaboration Interface Role of Module Interface Data exchange between modules: BPSD Standard dataset: Specify set of data Specification of data exchange interface: set/get/load/save Execution control: BPSX Specification of execution control interface: initialize, setup, exec, visualize, terminate Uniform user interface: parameter input, graphic output Role of data exchange interface Keep present status of plasma and device Save into file and load from file Store history of plasma Interface to experimental data base

14 Policies of BPSD Minimum and Sufficient Dataset To minimize the data to be exchanged Mainly profile data Routines to calculate global integrated quantities separately provided Minimum Arguments in Interfaces To maximize flexibility Use derived data type or struct Only one dataset in the arguments of an interface Minimum Interfaces To make modular programming easier Use function overloading

15 Data Exchange Interface: BPSD Standard dataset: Specify data to be stored and exchanged Data structure: Derived type (Fortran95): structured type e.g. Specification of API: Program interface e.g. time plasmaf%time number of grid plasmaf%nrmax number of species plasmaf%nsmax square of grid radius plasmaf%s(nr) plasma density plasmaf%data(nr,ns)%pn plasma temperature plasmaf%data(nr,ns)%pt Set data bpsd set data(plasmaf,ierr) Get data bpsd get data(plasmaf,ierr) Save data bpsd save(ierr) Load data bpsd load(ierr)

16 BPSD Standard Dataset (version 0.6) Category Name EQ TR TX FP WR WM DP Shot data bpsd shot type Device data bpsd device type in in in in 1D equilibrium data bpsd equ1d type out in in in 2D equilibrium data bpsd equ2d type out in in in in 1D metric data bpsd metric1d type out in in in 2D metric data bpsd metric2d type out in in in in Plasma species data bpsd species type in in in in in Fluid plasma data bpsd plasmaf type in out out i/o in Kinetic plasma data bpsd plasmak type out in Transport matrix data bpsd trmatrix type i/o Transport source data bpsd trsource type i/o i/o i/o out out Dielectric tensor data bpsd dielectric type in in out Full wave field data bpsd wavef type in out Ray tracing field data bpsd waver type in out Beam tracing field data bpsd waveb type in out User defined data bpsd 0/1/2ddata type

17 Execution Control Interface BPSX: not yet decided Monolithic integrated code Script-driven element codes MPI-2 driven element codes Example of element codes for TASK/TR TR INIT Initialization (Default value) TR PARM(ID,PSTR) Parameter setup (Namelist input) TR SETUP(T) Profile setup (Spatial profile, Time) TR EXEC(DT) Exec one step (Time step) TR GOUT(PSTR) Plot data (Plot command) TR SAVE Save data in file TR LOAD load data from file TR TERM Termination

18 Equilibrium Analysis Shape of an axisymmetric plasma: poloidal magnetic flux ψ(r, Z) Grad-Shafranov equation R 1 ψ RR R + 2 ψ Z 2 = μ 0R 2dp(ψ) dψ Pressure profile: p(ψ) Poloidal current density profile: F(ψ) Plasma boundary shape (fixed boundary) or Poloidal coil current (free boundary) F(ψ)dF(ψ) dψ determines the poloidal plasma shape. Coupling with transport analysis Input: p(ψ), q(ψ) = F dv 1 dψ R 2 Output: Metric quantities, Flux surface averaged quantities

19 Various Levels of Transport Modeling

20 Transport Modeling in the TASK code Diffusive transport equation: TASK/TR Diffusion equation for plasma density Flux-Gradient relation Conventional transport analysis Dynamical transport equation: TASK/TX: Continuity equation and equation of motion for plasma density Flux-averaged fluid equation Plasma rotation and transient phenomena Kinetic transport equation: TASK/FP: Gyrokinetic equation for momentum distribution function Bounce-averaged Fokker-Plank equation Modification of momentum distribution

21 Diffusive Transport Equation: TASK/TR Transport Equation Based on Gradient-Flux Relation: Γ = M F/ ρ where V: Volume, ρ: Normalized radius, V = dv/dρ Particle transport 1 V t (n sv ) = ( ) V ρ n s V s V ρ 2 n s D s ρ ρ + S s Toroidal momentum transport 1 V t (n su φs V ) = ( ) V ρ n s u φs V Ms V ρ 2 u φs n z μ s ρ ρ Heat transport ( ) 1 3 V 5/3 t 2 n st s V 5/3 Current diffusion B θ = [ t ρ η FR 0 R 2 + M s = 1 ( V ρ 3 ) V ρ 2 n st s V Es V ρ 2 T s n s χ s ρ R 0 F 2 ( V B θ μ 0 V ρ F ) ρ 2 R 2 ] η FR 0 R 2 J B ext + P s

22 Transport Processes Neoclassical transport Collisional transport in an inhomogeneous magnetic field Radial diffusion: usually small compared with turbulent diffusion Enhanced resistivity: due to trapped particles Bootstrap current: toroidal current driven by radial pressure gradient Ware pinch: Radial particle pinch driven by toroidal electric field Turbulent transport Various transport models: GLF23, CDBM, Weiland, Atomic transport: charge exchange, ionization, recombination Radiation transport Line radiation, Bremsstrahlung, Synchrotron radiation Parallel transport: along open magnetic field lines in SOL plasmas

23 Turbulent Transport Models CDBM model: current diffusive ballooning mode turbulence model developed by K. Itoh et al. Marginal stability condition of the current diffusive ballooning mode including turbulent transport coefficients as parameters GLF23 model: Gyro-Landau-Fluid turbulence model developed by Waltz and Kinsey (GA) Linear growth rate from gyro-landau-fluid model (ITG, TEM, ETG) Evaluate transport coefficients based on mixing length estimate Calibrate coefficients by the linear stability code GKS Calibrate coefficients by the nonlinear turbulence code GYRO Weiland model: developed by J. Weiland Based on ITG turbulence model

24 GLF23 Transport Model Linear growth rate from gyro-landau-fluid model Calibrate coefficients by the nonlinear turbulence code GYRO Good agreement with experimental data

25 CDBM Transport Model: CDBM05 Thermal Diffusivity (Marginal: γ = 0) χ TB = F(s,α,κ,ω E1 ) α 3/2 c2 ω 2 pe v A qr s α dependence of F(s,α,κ,ω E1 ) ω E1 =0 ω E1 =0.1 Magnetic shear s r dq q dr Pressure gradient α q 2 R dβ dr Elongation κ b/a E B rotation shear ω E1 r2 d E sv A dr rb Weak and negative magnetic shear, Shafranov shift, elongation, and E B rotation shear reduce thermal diffusivity. F ω E1 =0.2 ω E1 = F(s,α,κ,ω E1 ) = s - α ( 2κ 1/2 1 + κ 2 ) 3/ G 1 ω 2 E1 2(1 2s )(1 2s + 3s 2 ) for s = s α< s 5/2 1 + G 1 ω 2 E1 2(1 2s + 3s 2 + 2s 3 ) for s = s α>0

26 TFTR #88615 (L-mode, NBI heating)

27 DIII-D #78316 (L-mode, ECH and ICH heatings)

28 Comparison of Transport Models: ITPA Profile DB Deviation of Stored Energy CDBM CDBM05 GLF23 Weiland

29 1D Dynamic Transport Code: TASK/TX Dynamic Transport Equations (TASK/TX) M. Honda and A. Fukuyama, JCP 227 (2008) 2808 A set of flux-surface averaged equations Two fluid equations for electrons and ions Continuity equations Equations of motion (radial, poloidal and toroidal) Energy transport equations Maxwell s equations Slowing-down equations for beam ion component Diffusion equations for two-group neutrals Self-consistent description of plasma rotation and electric field Equation of motion rather than transport matrix Quasi-neutrality is not assumed.

30 Model Equation of Dynamic Transport Simulation Flux-surface-averaged multi-fluid equations: n s t = 1 r r (rn su sr ) + S s t (m sn s u sr )= 1 r r (rm sn s u 2 sr ) + 1 r m sn s u 2 sθ + e sn s (E r + u sθ B φ u sφ B θ ) r n st s t (m sn s u sθ )= 1 r 2 r (r2 m s n s u sr u sθ ) + e s n s (E θ u sr B φ ) + 1 ( r 3 n r 2 s m s μ s r r +F NC sθ ( ) 1 ms n s u sφ = t r + F C sθ + F W sθ + F X sθ + F L sθ r (rm sn s u sr u sφ ) + e s n s (E φ + u sr B θ ) + 1 r r ( ) rn s m s μ s r u sφ ) u sθ r t 3 2 n st s = 1 r +F C sφ + FW sφ + FX sφ + FL sφ ( ) 5 r r 2 u srn s T s n s χ s r T e + e s n s (E θ u sθ + E φ u sφ ) +P C s + P L s + P H s

31 Typical Ohmic Plasma Profiles at t = 50 ms JFT-2M like plasma composed of electron and hydrogen R = 1.3m, a = 0.35 m, b = 0.4m, B φb = 1.3T, I p = 0.2MA, S puff = m 2 s 1 γ = 0.8, Z eff = 2.0, Fixed turbulent coefficient profile E r (r) [kv/m] E 0.2 θ [mv/m] E φ [kv/m] B θ [T] B φ [T] n e [10 20 m -3 ] T e [kev] T i [kev] n 01 [10 14 m -3 ] n 02 [10 10 m -3 ] u er [m/s] u eθ [km/s] u eφ [km/s] D e [m 2 /s] χ e =χ i =μ e =μ i [m 2 /s] u ir [m/s] u iθ [km/s] u iφ [km/s] q j φ [MA/m 2 ] 0.0

32 Density Profile Modification Due to NBI Injection Modification of n and E r profile depends on the direction of NBI E r [kv/m] u bφ [km/s] q Co/Counter with I p : Density flattening/peaking n e [10 20 m -3 ] u er [km/s] j φ [MA/m 2 ] no NBI co counter T e [kev] u eθ [km/s] u iθ [km/s] T i [kev] 0.0 u eφ [km/s] u iφ [km/s]

33 Toroidal Rotation Due to Ion Orbit Loss Ion orbit loss near the edge region drives toroidal rotation Ref. M. Honda et al., NF (2008)

34 Source Modeling Heat and momentum sources: Alpha particle heating: sensitive to fuel density and momentum distribution Neutral beam injection: birth profile, finite size orbit, deposition to bulk plasma Waves: IC ( 50 MHz): fuel ion heating, current drive, rotation drive(?) LH ( 10 GHz): current drive EC ( 170 GHz): current drive, pre-ionization Particle source Gas puff, recycling: Neutral beam injection: Pellet injection: penetration, evaporation, ionization, drift motion

35 Wave Modeling Ray tracing: EC, LH Spatial evolution of ray position and wave number Wave length λ much less than scale length L: λ L Beam size d is sufficiently large (Fresnel condition): L d 2 /λ Beam tracing: EC New variables: beam radius, curvature of equi-phase surface Beam propagation: EC Under development: Time evolution of E(r, t)e i ωt Full wave analysis: IC, AW, MHD Stationary Maxwell s equation as a boundary problem Wave length λ comparable with scale length L Evanescent region, strong absorption, coupling with antenna

36 Wave Dispersion Analysis : TASK/DP Various Models of Dielectric Tensor ɛ (ω, k; r): Analytic model Resistive MHD model Collisional cold plasma model Collisional warm plasma model Kinetic model (Maxwelian, non-relativistic) Drift-kinetic model (Maxwellian, inhomogeneity) Hamiltonian formulation (cyclotron, bounce, precession) Numerical integrateion Kinetic model (Relativistic, Maxwellian) Kinetic model (Relativistic, arbitrary f (u)) Drift-kinetic model (Inhoogeneity, arbitrary f (u)) Hamiltonian formulation (arbitrary f (u)))

37 Geometrical Optics Ray Tracing: TASK/WR Wave length λ Characteristic scale length L of the medium Plane wave: Beam size d is sufficiently large Fresnel condition is well satisfied: L d 2 /λ Evolution along the ray trajectory τ Wave packet position r, wave number k, dielectric tensor K Ray tracing equation: dr dτ = K / K k ω dk dτ = K / K r ω Equation for wave energy W, group velocity u g, damping rate γ (u g W ) = 2γW

38 Beam Tracing: TASK/WR Beam size perpendicular to the beam direction: first order in ɛ Beam shape : Hermite polynomial: H n ) [ ] E(r) = Re C mn (ɛ 2 r)e(ɛ 2 r)h m (ɛξ 1 )H n (ɛξ 2 )e i s(r) φ(r) mn Amplitude : C mn, Polarization : e, Phase : s(r) + i φ(r) s(r) = s 0 (τ) + k 0 α(τ)[r α r α 0(τ)] s αβ[r α r α 0(τ)][r β r β 0 (τ)] φ(τ) = 1 2 φ αβ[r α r α 0(τ)][r β r β 0 (τ)] Position of beam axis : r 0, Wave number on beam axis: k 0 Curvature radius of equi-phase surface: R α = 1/λs αα Beam radius: d α = 2/φ αα R 1 d 1 Gaussian beam : case with m = 0, n = 0 R 2 d 2

39 Beam Propagation Equation Solvable condition for Maxwell s equation with beam field dr α 0 dτ = K k α dk 0 α dτ = K r α ds αβ dτ = 2 K r α r β dφ αβ dτ = ( 2 K r α k γ + 2 K r β k γ s αγ 2 K k γ k δ s αδ 2 K s r α βγ k γ ) φ βγ ( 2 K r β k γ + 2 K k γ k δ s αγs βδ + 2 K k γ k δ s βδ 2 K k γ k δφ αγφ βδ ) By integrating this set of 18 ordinary differential equations, we obtain trace of the beam axis, wave number on the beam axis, curvature of equi-phase surface, and beam size. Equation for the wave amplitude C mn (u g0 C mn 2) = 2 ( γ C mn 2) Group velocity: u g0, Damping rate: γ (e ɛ A e)/( K/ ω) φ αγ

40 Full wave analysis: TASK/WM magnetic surface coordinate: (ψ, θ, ϕ) Boundary-value problem of Maxwell s equation Kinetic dielectric tensor: ɛ E = ω2 c 2 ɛ E + i ωμ0 j ext Wave-particle resonance: Z[(ω nω c )/k v th ] Fast ion: Drift-kinetic [ t + v + (u d + u E ) + e α (v m E + u d E) ] α ε f α = 0 Poloidal and toroidal mode expansion Accurate estimation of k Eigenmode analysis: Complex eigen frequency whichmaximize wave amplitude for fixed excitation proportional to electron density

41 Fokker-Planck Analysis : TASK/FP Fokker-Planck equation for multi-species velocity distribution function: f s (p, p,ψ,t) f s = E( f s ) + C( f s ) + Q( f s ) + L( f s ) t E( f ): Acceleration term due to DC electric field C( f ): Coulomb collision term Q( f ): Quasi-linear term due to wave-particle resonance L( f ): Spatial diffusion term Bounce-averaged: Trapped particle effect, zero banana width Relativistic: momentum p, weakly relativistic collision term Nonlinear collision: momentum and energy conservation Multi-species: conservation between species Three-dimensional: spatial diffusion (neoclassical, turbulent)

42 ITER Modeling Needs based on Dr. Campbell talk at Cadarache, Sept 2007 Plasma scenario development: Preparation for operation Detailed design of auxiliary systems: H&CD, Diagnostics, Fueling, Design of plasma control system: Development and optimization of integrated control strategies Preparation of ITER operational programme: End-to-end scenario development Detailed pulse definition Experimental data evaluation: Pulse characterization and physics analysis Refinement of operation scenario and performance predictions

43 Analysis of EC propagation in ITER BB = RR = RA = RKAP = Q0 = QA = RF = 1.700E+05 NPHII= PA(1) = PZ(1) = PN(1) = PNS(1) = PTPR(1)= PTPP(1)= PTS(1) = MODELP1= 25. PA(2) = PZ(2) = PN(2) = PNS(2)= PTPR(2)= PTPP(2)= PTS(2) = MODELP2= 1. 4 ABS POWER ABS WEIGHT POWER

44 Analysis of IC propagation in ITER Minority heating in D phase: D+H Radial profile Wave electric field P abs by minority ion

45 ITER Plasma Transport Simulation with CDBM05 Inductive operation I p = 15 MA P NB = 40 MW on axis β N = 2.65 Hybrid operation I p = 12 MA P NB = 33 MW on axis β N = 2.68 Time evolution Radial profile Time evolution Radial profile

46 Alfvén Eigenmode Analysis by TASK/WM Alfvén eigenmode driven by alpha particles Calculated by the full wave module TASK/WM Toroidal mode number: n = 1 TAE is stable: Eigen mode frequency = (95.95 khz, 1.95 khz) Alfvén continuum frequency Alfvén eigenmode structure Alpha particle density

47 ITER Inductive Operation 15 MA Current rise in 70 s Behavior of plasma depends on timing of heating

48 Future Plan of the TASK code

49 Remaining Issues in Tokamak Plasma Modling Pedestal temperature: Transport barrier formation: ELM physics: Nonlinear MHD events: modeling of L/H transition turbulence transport model nonlinear behavior of ELM on transport modeling of plasma profile change Energetic-ion driven phenomena: coupling of Alfvén mode and drift waves Kinetic analysis of transport and MHD phenomena: non-maxwellian distribution Wave physics: full wave analysis of Bernstein waves Divertor plasma: Start up: and so on... plasma-wall interaction Rapid change of equilibrium and control

50 Summary Integrated modeling of burning plasmas is indispensable for exploiting optimized operation scenario of ITER and reliable design of DEMO. Discussion on international collaboration for the development of integrated ITER modeling is on going. There are many remaining issues in constructing comprehensive integrated code; especially, L/H transition mechanism, turbulent transport model, ELM physics, nonlinear MHD events, energetic particle driven phenomena and plasma wall-interactions. Solving those remaining issues requires not only large-scale computer simulations but also intensive modeling efforts based on experimental observations.