THEORY AND COMPUTATION - General Atomics PERFORMANCE DISCHARGES WITH ELMING EDGE SUSTAINED AND...
Transcript of THEORY AND COMPUTATION - General Atomics PERFORMANCE DISCHARGES WITH ELMING EDGE SUSTAINED AND...
NATIONAL FUSION FACILITYS A N D I E G O
DIII–D003–00/VSCwj
THEORY AND COMPUTATION
byV.S. Chan
Presented toDIII–D Program Advisory Committee Meeting
January 20–21, 2000
PHYSICS ISSUES FOR LONG-PULSE ADVANCEDTOKAMAK PLASMAS REQUIRE FURTHER THEORY
AND MODELING DEVELOPMENT
� Non-ideal MHD modes have become prominent performance limiting factors
— What roles do resistive wall mode, neoclassical tearing mode, fast ioninstabilities play?
� Plasma edge behavior can affect global performance
— The effect of ELMs on edge pressure gradient— Interplay between SOL/divertor and core
� Localized current profile control and bootstrap alignment are essential forsteady-state— Confidence in local current drive physics
� Transport barrier control is being explored to optimize both confinementand stability— Opportunity for electron transport understanding— Physics based modeling of ITB formation a useful tool
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
HIGH PERFORMANCE DISCHARGES WITH ELMING EDGESUSTAINED AND PERFORMANCE LIMITED BY NON-IDEAL MODES
003-00/VSC./jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
� Previous high-performancedischarges terminated by first large ELM
� Beta rollover correlated withthe onset of a localized perturbation
1218.8 1219.0 1219.2 1219.4 1219.6
–1002567°
97°
137°
157°
277°
307°
322°
340°
150B from magneticprobes at:
–1000
100
–10025
150
–10025
150
–10025
150
–10025
150
–100100300
–2000
2001.5 2.0 2.5 3.0 3.5
t (s)
0.0
2.0
4.0
β N
Dα
#75124VH Mode
βN
� Recent high-performancedischarges robust to ELMs
4
2
00.8 1.3 1.8
Time (s)
Dα
βN
Time (ms)
99510
Onset of RWM •
THE SOFT BETA COLLAPSE AT THE END OF THE LONG-PULSEHIGH PERFORMANCE PHASE IS CAUSED BY RESISTIVE WALL MODES
� Interaction of small RWM with rotation and angular momentumtransport (unique GA)
— Incorporate RWM drag profile in transport simulations and validate resultsagainst experiment
� Nonlinear saturation and identification of small RWM in slowdown phase— Investigate Gimblett and Finn theories of RWM stability, possibly using
NIMROD and compare with observed mode structure (with Culham)
� Feedback stabilization of RWM— Incorporate active feedback in vacuum code and couple to GATO,
PEST and MARS; also investigate alternate feedback scheme (with Bondeson,Chance, Columbia U.)
Key Physics Issues
003-00/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
BETA ROLLOVER OBSERVED IN LONG-PUSLE HIGHPERFORMANCE PLASMAS MAY BE CAUSED BY HIGH
FREQUENCY FAST ION INSTABILITIES
� Mode identification: *AE vs. KBM vs. fast particle branch— Use CONT code with analytic theory code from UCI to compare
predictions for candidate modes (with Chance, L. Chen)
� Effect of fast particle instabilities on ion confinement: expulsion vs. redistribution— Utilize ORBIT code and UCI code to determine effect of magnetic fluctuations
from fast particle modes on fast ions
� Stability of fast particle modes— Investigate relative drive vs. damping for various modes, and validate against
experiment (with L. Chen)
Key Physics Issues
003-00/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
CLASSICAL AND NEOCLASSICAL TEARING MODES MAY SETβ LIMIT FOR LONG-PULSE DISCHARGES
� Physical interpretation and quantitative determination of ∆'
— Use PEST-III and develop TWIST-R to cross-check ∆', improve inner layermodels, and develop finite b theory to relate ∆' to free energy in generalcase (with Pletzer, Glasser)
� Physics of polarization term and dependence on mode frequency— Check Wilson (Culham) and Waelbroeck (IFS) theory against DIII–D experiment
� ECCD stabilization— Investigate relative efficiencies of stabilization using ECH and ECCD
(with Sauter, ASDEX-U)
� Identification of classical vs. neoclassical tearing modes— Compare predictions from linear and nonlinear models using diagnostic codes
DIAG PAK and VACUUM with measurements (with Wisconsin, Chance)
Key Physics Issues
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
PHYSICS ISSUES FOR LONG-PULSE ADVANCEDTOKAMAK PLASMAS REQUIRE FURTHER THEORY
AND MODELING DEVELOPMENT
� Non-ideal MHD modes have become prominent performance limiting factors
— What roles do resistive wall mode, neoclassical tearing mode, fast ioninstabilities play?
� Plasma edge behavior can affect global performance
— The effect of ELMs on edge pressure gradient— Interplay between SOL/divertor and core
� Localized current profile control and bootstrap alignment are essential forsteady-state— Confidence in local current drive physics
� Transport barrier control is being explored to optimize both confinementand stability— Opportunity for electron transport understanding— Physics based modeling of ITB formation a useful tool
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
EDGE STABILITY FOCUSES ON ENHANCING PREDICTIVECAPABILITIES TO TEST AND IMPROVE ELM MODEL
� Advance understanding of edge modes and enhance predictive capabilitiesthrough development of new computational tools and theory
� Working ELM model
— Intermediate to low n ballooning/kink/peeling modes frominteraction among MHD modes with various n and evolutionand growth of edge P′ and J
— ELM amplitude determined by radial width of unstable modes
� Issues
— Lacking computational tools to evaluate intermediate n > 6modes and the current drive and non-ideal effects onthese modes
— Lacking tools to evaluate non-linear evolution and mode coupling
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
more JBS
2nd stability accessdiamagnetic stabilization
Stable?
Stable?
J edgeP′ edge
INTERMEDIATE n STABILITY ANALYSIS TOOLS ARE CRUCIALTO TEST AND IMPROVE THE WORKING ELM MODEL
� Intermediate n modes tendto be most unstable
� Unstable n determinedby edge J, shape, FLRstabilization, and 2ndregime access
� Need to developcomputational tools toevaluate intermediaten > 6 modes— Current and non-ideal
effects, general geometry(BOUT, ELITE)
— Ballooning representation(GATO)
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
0 O(40)0
5
10
15Schematic of Ideal MHD Edge Instability Thresholds
∞n, Toroidal Mode Number
CalculatedWith GATO
Squareness
FLRStabilization
High (δ2 = 0.5)
δ2 = 0.5 — ε
δ2 = 0.05
P edge
Thr
esho
ld (a
.u.)
′
ANALYSIS OF n > 6 MODES USING EXISTING IDEALSTABILITY CODES IS COMPUTATIONALLY INTENSIVE
� Evaluation of n > 6modes using GATOrequires very finemesh and hours ofCRAY time
� A ballooningrepresentation isbeing implementedinto GATO to improvethe accuracy andefficiency of thecalculations
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
2 4 6 8 100
5
10
15
20
Toroidal Mode Number (n)
Ballooning 1st Regime Limit
Stable
Threshold Boundary
Unstable
Pedge′
[106 P
a/(W
b/ra
dian
)]
CURRENT EFFECTS MAY LIMIT SECONDBALLOONING STABILITY ACCESS
� Current effects on ballooning modesevaluated using a simplified S-α model
� 2nd regime access is possible for high nand a deep magnetic well
� A general geometry high npeeling/ballooning mode code ELITE isbeing developed in collaborationwith Culham
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
2 3 4 5 6 70
2
4
6
s (M
agne
tic S
hear
)
n = 40
n = 40
n = 20
n = 10
Peeling ModeUstable
Stable
Stable
Ballooning Unstable
α (Normalized Pressure Gradient)
UNDERSTANDING AND CONTROLLING EDGE INSTABILITIESARE ESSENTIAL FOR SUSTAINING HIGH PERFORMANCE PLASMAS
� Determination of finite n ballooning stability from non-ideal corrections
— Use full geometry gyrokinetic code to determine ncrit when pressuregradient is limited by first regime ballooning (small ELMs)
� Stability of large ELMs at intermediate n— Develop ELITE code and ballooning representation in GATO and match for
intermediate n; add non-ideal corrections to ELITE (with Wilson)
� Role of diamagnetic effects, p' and j in determining instability thresholdand penetration
— Add diamagnetic effects into ELITE, and explore parameter space using fullsuite of codes
� Nonlinear ELM evolution— Develop and use BOUT code (LLNL) (with Xu, Galkin)
Key Physics Issues
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
CORE/EDGE COUPLING SIMULATION PROVIDESFIRST DEMONSTRATION OF STABLE ALGORITHM
� Initial UEDGE simulationwith 4 MW radial power flowin electrons and ions
� Coupling iterated untilradial power flow andtemperatures at 96%surface consistent withNBI heating
003-00/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
Electrons
Electrons
Thermal Ions
Ions
Beam Ions
Core (Corsica)0.400.35
0.250.200.150.100.05
00 0.1 0.2 0.3 0.4 0.5
ΨN
0.6 0.7 0.8 0.9 1.0
0 0.1
94002, 1250 ms
0.2 0.3 0.4 0.5ΨN
0.6 0.7 0.8 0.9 1.0
n (1
020 m
–3) 0.30
0.35
0.250.200.150.100.05
0
T (k
eV)
0.30
Edge(UEDGE)
ADDITION PHYSICS WILL BE ADDED FOR NEXT STEP SIMULATIONS
� Turn on coupling for density and gas variables
� Add consistent model for toroidal rotation in CORSICA and UEDGE
� Model L to H transition in DIII–D discharge with simple model for edge transport
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
Key Physics Issues
— Decrease D and χ in edge only
— Use different models for core transport
PHYSICS ISSUES FOR LONG-PULSE ADVANCEDTOKAMAK PLASMAS REQUIRE FURTHER THEORY
AND MODELING DEVELOPMENT
� Non-ideal MHD modes have become prominent performance limiting factors
— What roles do resistive wall mode, neoclassical tearing mode, fast ioninstabilities play?
� Plasma edge behavior can affect global performance
— The effect of ELMs on edge pressure gradient— Interplay between SOL/divertor and core
� Localized current profile control and bootstrap alignment are essential forsteady-state— Confidence in local current drive physics
� Transport barrier control is being explored to optimize both confinementand stability— Opportunity for electron transport understanding— Physics based modeling of ITB formation a useful tool
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
0.0
0.5
1.0
1.5
2.0
2.5
0.0 0.2
⟨JφR
0/R⟩
(MA/
m3 )
0.4 0.6 0.8 1.0
Total
Hirsh 88ν∗=0Finite ν∗
ψ0.5
COMPUTED BOOTSTRAP CURRENT PROFILEIS SENSITIVE TO THEORETICAL MODEL
� New analytic formulas forbootstrap current wereconstructed based on 3-Dkinetic calculations forgeneral tokamak geometryand arbitrary collisionality
� NCLASS 1999 (ORNL)is being implementedto model edgebootstrap current
Sauter et al., Phys. Plasma 4(1999) 2834
003-00/VSC/jy
QTYUIOPS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
–1
0
1
2
3
4
5
–1
Theoretical ECCD Efficiency (1018 A/W-m2)
0 1 2 3 4 5
Exp
erim
ental(101
8AW-m
2 )
MEASURED OFF-AXIS ECCD EXCEEDS PREDICTIONS FROMBOUNCE-AVERAGED FOKKER-PLANCK CALCULATIONS
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
� Experimental ECCD obtainedfrom internal loop-voltageanalysis based on EFITequilibrium reconstructionusing MSE data
� Standard modeling approachis based on bounce-averageapproximation; it gives a lowerestimate of ECCD efficiency
� Improved understanding andresolving the discrepancyare needed
Solid Circles: ExperimentalOpen Circles: ECCD efficiency computed
using CQL3D with themeasured Ohmic electric field Ell
REFINED DATA ANALYSIS AND IMPROVED THEORETICALMODELING ARE PLANNED TO RESOLVE DISCREPANCIES
� Refine data analysis technique
— 1-1/2-D transport simulation study of current profile evolution indicated thatbetter spatial resolution in equilibrium reconstruction is crucial
— Introduce new current profile representation into EFIT to allow localizedfeatures with strong gradient
— Perform consistency check using both the refined internal loop-voltageanalysis and 1-1/2-D transport simulations
� Improve theoretical modeling of ECCD
— Finite collisionality effects were examined by a linear theory based onGreen's function formulation; only modest enhancement in ECCD efficiencywas observed
— Extend the finite collisionality theory to include quasi-linear and Ell effects
— Generalize theoretical modeling of ECCD to non-axisymmetric configurations
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
PHYSICS ISSUES FOR LONG-PULSE ADVANCEDTOKAMAK PLASMAS REQUIRE FURTHER THEORY
AND MODELING DEVELOPMENT
� Non-ideal MHD modes have become prominent performance limiting factors
— What roles do resistive wall mode, neoclassical tearing mode, fast ioninstabilities play?
� Plasma edge behavior can affect global performance
— The effect of ELMs on edge pressure gradient— Interplay between SOL/divertor and core
� Localized current profile control and bootstrap alignment are essential forsteady-state— Confidence in local current drive physics
� Transport barrier control is being explored to optimize both confinementand stability— Opportunity for electron transport understanding— Physics based modeling of ITB formation a useful tool
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
ION THERMAL TRANSPORT IN AT LONG-PULSE DISCHARGE ISREDUCED THROUGHOUT THE DISCHARGE BUT ELECTRON
THERMAL TRANSPORT REMAINS ANOMALOUS
� L–mode: χi is already reduced toChang-Hinton neoclassical in theinterior of the discharge, confirmingthat an ITB is forming there
� ELM-free H–mode: reduced ionthermal transport regions at the edgeand core combine to reduce χi to near-neoclassical values throughout theentire plasma volume
003/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
� ELMing H–mode: ion thermal transportincreases slightly throughout the plasma,but remains within a factor of ~2 of theneoclassical level
� Electron transport remains at L–modelevels in the plasma interior, with somereduction seen in the H–mode edge region
0.0 0.2 0.4 0.6 0.8 1.0ρ
1.497 s (ELMing)
1.257 s (ELM-free)
1.106 s (L–mode)
0.1
1
10
m2 /s
0.1
1
10
m2 /s
0.1
1
10
m2 /s
χi
χe
χi
χiχe
χi
χiχe
χi
tottot
nc
tottot
nc
tot
tot
nc
LINEAR GYROKINETIC STABILITY ANALYSIS YIELDSINFORMATION ON ION AND ELECTRON TRANSPORT
� GKS code (originally developed by Kotschenreuther) is used for analysis ofDIII–D discharges
— Maximum ITG growth rate profiles for comparison with ExB velocity shear rateto find transport barrier location
— Critical electron temperature gradient profile for high wavenumber ETG mode forelectron transport studies with an ion transport barrier
� Recent extensions by Waltz include:
— Shaped local magnetic equilibrium (R. Miller)— Fully electromagnetic response— Debye shielding term for ETG modes— Flux tube code, cannot treat shear-flow
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
TRANSPORT MODELING WITH GLF23 IS USED TO STUDYTRANSPORT BARRIER FORMATION AND CONTROL
� The Gyro-Landau-Fluid transport model GLF23 is a comprehensive theory basedtransport model
— Tested against transport profile database
— Used to study heat and cold pulse propagation
— Is beginning to be used for reactor design studies
— A faster more robust numerical scheme is under development for transportbarrier evolution
— A full GLF model for one impurity species has been added to the code formodeling of impurity injection
� Recent improvements to the model:
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY
QTYUIOP
RELATED DEVELOPMENTS
� In 1999, we have constructed Luna - an 18-processor mini-supercomputer(Beowulf-class) running Linux OS
� The Plasma Science Advanced Computing Initiative will begin this year
— Provides an easy-to use, robust, stable platform for selected simulation codeswith significant gain in performance e.g. Monte-Carlo fast ion package
— Stimulated innovative ideas to use parallel clusters e.g. multiple-time slices,parallel EFIT reconstruction
— Plan to construct a 44-processor system (10x faster) this year
— Nonlinear gyrokinetic simulations will provide fundamental insights in theformation and dynamics of internal transport barriers
— 3-D Macroscopic simulations will yield better understanding ofnon-ideal instabilities
— Diagnostics of simulations results can be validated against experiments
003-00/VSC/jyS A N D I E G O
DIII–DNATIONAL FUSION FACILITY