Benchmark activity of particle transport modelling within IOS 1 Spokesman: Y.S. Na 1 (ysna@snu.ac.kr) Contributors (16): A. Fukuyama 3, J. Garcia 4, N. Hayashi 5, C.E. Kessel 2, K. Kim 1,8, F. Koechl 6, T. Luce 7, D. H. Na 1, A. Pankin 8, J.M. Park 9, A.R. Polevoi 10, F. Poli 2, A.C.C. Sips 11, I. Voitsekhovitch 12, A. Wisitsorasak 13, X. Yuan 2 Institutions (13): 1 Seoul National University, Seoul, Korea 2 Princeton Plasma Physics Laboratory, Princeton, New Jersey, USA 3 Kyoto University, Kyoto, Japan 4 Association Euratom-CEA/Cadarache, Saint-Paul-Lez Durance, France 5 Japan Atomic Energy Agency, Naka, Ibaraki, Japan 6 Association EURATOM-ÖAW/ATI, Atominstitut, TU Wien, Vienna, Austria. 7 General Atomics, San Diego, California, USA 8 TECH-X, Boulder, Colorado, USA 9 Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA 10 ITER Organization, Route de Vinon sur Verdon, 13067 St Paul Lez Durance, France, 11 JET, Culham Science Centre, Abingdon, OX14 3DB, United Kingdom 12 EUROfusion, Garching, Germany 13 King Mongkut's University of Technology Thonburi, Bangkok, Thailand Codes (9): ASTRA(6.x, 7.x), CRONOS, FASTRAN, JINTRAC, TASK/TR, TOPICS, TRANSP, FACETS
Particle transport in ITER prediction 2 Ultimate goal: modelling of integrated control of - core density (burn control for DT), - mode of operation and transition (L/H), - ELM pacing, - divertor detachment, - impurity accumulation Responsibility of ITPA TGs: - Core transport model validation (T&C) - Pedestal transport (PED) - Core boundary conditions and edge fuelling (SOL/DIV) - Core transport solvers + integrated modelling (IOS) Core transport solvers: (i/e transport, sources, sinks, bnd.cond., evolution) - Electron/ion transport simulation - Core fueling by pellets (permanent/instant models) - Core fueling by gas puffing - Sink with ELMs (permanent/instant models) - Impact of boundary conditions (from SOL/DIV transport and control) - Effect of volume evolution
3 Particle transport benchmarking in IOS TG Goal: - To verify proper treatment of particle transport among various integrated modelling codes in conditions close to those expected in ITER Next steps: - To predict ITER fusion performance more accurately (density peaking?) - To predict impurity behaviour (strong dependence background density in neoclassical impurity transport) - To address control issues in transients (L-H transition, H-L transition, burn control)
Benchmarking steps Phase I Unification of definitions Phase II: Stationary target plasma: Fixed equilibrium, prescribed target plasma D(x), V(x), and T i (a), T e (a), n e (a) 1) Ion vs electron transport: Fixed particle sources S edge (x), S pel (x) 2) Fuelling models (pellet): Fixed source from puffing, S edge (x): comparison of pellet models; impact of boundary conditions λ pel (n a,t a,q) permanent vs instant pellet models; 3) Fuelling models (edge): Fixed source from pellet, S pel (x): comparison of puffing models, impact of boundary conditions (n a,t a ) 4) Sink with ELMs (permanent vs instant model): Fixed sources 5) Sensitivity scans of transport coefficients: Fixed sources and sink: evaluation of impact of particle transport to fusion performance Phase III: Evolving target plasma: - prescribed evolution of plasma shape, D(x), V(x), S edge (x), S pel (x), T i,e (a), n e (a) - evaluation of impact of particle transport to scenario evolution Phase IV: Expand integrated modelling within the ITER design limitations (+ desirable contribution from ITPA TGs) - physics-based core transport models (+ T&C TG) - boundary conditions + gas puffing from SOL/DIV (+ SOL/DIV TG) - particle source from pellet and sink with ELM (+ T&C+PED TG) 4
Core transport solvers: Phase I unification of definitions 5 TRANSP (PTSOLVER), ASTRA 7.x (default electron transport) CRONOS (default electron transport) TOPICS (default ion transport, can simulate electron transport) JINTRAC (default ion transport) 1 ' 1 ' ( nv i ) + ( V Γi ) = S ' ' i V t V ρ Differences are underlined by red and blue
6 Core transport solvers: Phase I unification of definitions Toroidal metric - Note that in NCLASS metric ( ρ) 2, ρ is included in D,V - Note that GLF23 and TGLF predict particle flux, Γ - Thus, separation to diffusive and convective parts is less trivial for solvers with different weights ( ρ) 2, ρ for these parts Electron vs ion transport solver - ASTRA enables simulations of both ion and electron transport modes - JINTRAC can emulate the electron solver by modifying the pinch term: 2 n ( ) e n D i inp ne ni ρ vi = vinp + vinp, ni = nd + nt ni ni ρ ρ ρ - Difference between ion and electrons solver predictions diverges with increase of plasma contamination by impurities - Predictions of simulations in the same mode (electron/ion transport) practically coincide (ASTRA and JINTRAC) (see slides 8,9 below)
Core transport solvers: Phase II Stationary target plasma 7 Fixed boundary conditions, n e (a) = 4.6x10 19 m -3 ; Prescription: n Be /n e = 2 %, n Ar /n e = 0.05%, n He (x) = 0.95 (1 x 4 ) 2 10 19 m -3
Core transport solvers: Phase II Stationary target plasma Ion vs electron transport solvers 8 - Difference between ion and electrons solver predictions diverges with increase of plasma contamination by impurities, P fus ~ n i 2, P LH ~ n e 0.7 (see also slide 9) (ASTRA & JINTRAC)
Core transport solvers: Phase II Stationary target plasma Ion vs electron transport solvers 9 - Predictions of simulations in the same mode (electron/ion transport) practically coincide (ASTRA & JINTRAC)
Core transport solvers: Phase II Stationary target plasma Initial benchmarking results 10 Note that model settings are not perfectly satisfied yet
Core transport solvers: Phase II Stationary target plasma Fuelling models (pellet) I 11 ITER Pellet Injection System: Maximal Intact pellet speed V pel,max = 300 m/s, fixed geometry S pel = n pel δv pel f pel (n pel 6 10 19 mm -3 ) => 2 variables: - Pellet size: δv pel = 90/50/33/17 mm 3, - Pellet frequency: f pel 32 Hz (2 injectors 16+16 Hz) Next step TBD: - Dependence of penetration depth on pellet model - Dependence of fuelling efficiency on pellet size - Permanent vs instant pellet model
Core transport solvers: Next step TBD: Phase II Stationary target plasma Fuelling models (pellet) II 12 Efficiency vs depth Depth vs model Depth vs size See refs. in Pegorie et al, PPCF 51 (2009) 124023 Depth vs boundary conditions, λ pel (n,t,q) See model in Polevoi et al, PPCF 43 (2001) 1525
13 Issues Encountered So Far in the Benchmarking of Particle Transport (question for October) First Principle Models - FPMs predict Γ i, i.e. D i, V i : is it OK? What about Γ e, D e,v e for electron density solvers? - Update models for GLF23, TGLF, BgB (a few bugs were found in the original release) - How to extend GLF23, TGLF for multi-impurity case? (recommendations how to treat multi-impurity case in present versions (M z,z eff )) - Proper conversion of cylindrical D, V to toroidal geometry - Clarify interpretation of minor radius in non-cylindrical metric in FPMs - What to use for cases where the models predict stability near the axis (x < 0.3) (critical for impurity accumulation) Neoclassics - NCLASS clarify units for torque - Readdress particle transport near the center (n D ~ n T ) - Core impurity transport
Issues Encountered So Far in the Benchmarking of Particle Transport 14 Hollow profiles (question for October) - Are present first-principle models valid for positive fuel ion density gradients? Such profiles are expected at the routine density ramp-up phase in ITER&DEMO and in machines with shallow pellet or NBI fueling and can appear after the L-H transition They may also appear in present-day experiments in stationary plasmas with substantial NBI MAST experiments with shallow pellet injection and JET experiments with the edge NBI fueling are not described properly by FPMs and turbulence codes (see example below)
15 Example: Pellet injection in MAST [L Garzotti et al, PPCF 56 (2014) 035004] n e dv for ψ N < 0.7 [10 20 ] t 0 + t abl l t 0 +10 ms ψ N <0.7 ψ N <0.7 GS2 predicts suppression of turbulence, but density within x < 0.7 continues to grow
Issues for the longer term 16 Nonlinear effects (T&C in collaboration with PEP) - Pellet ablation and drift (instant models): impact on heat and particle transport - ELM triggering by HFS and LFS pellets and its influence on pellet fueling efficiency - Sink with ELMs (instant models): impact on heat and particle transport Evolution of density profiles during L-H-L transition - C-Mod shows a change in the edge density but little or no core change during the L-H transition. There are signs of higher edge density than core density. - JET Experiments show a fast edge density rise and slow core/overall density rise at L-H transition