Global Stability Issues for a Next Step Burning Plasma Experiment UFA Burning Plasma Workshop Austin, Texas December 11, 2000 S. C. Jardin with input from C.Kessel, J.Manickam, D.Meade, P.Rutherford
Workshop charge boils down to two questions in each area: Are we ready to design a burning plasma experiment with confidence that it will succeed ? What will we learn from it if we do build it? Note the trap we can fall into if the answer to either of these is too positive: the key is the right balance
I A (MA) B R 5/ Let us consider FIRE, as it is being proposed as a next step burning plasma experiment JET, JET-U FIRE AIRES designs A major step in the study of alpha- heating dominated plasmas, and in simultaneous ( *, *) values Provides critical data point in a new parameter regime for benchmarking of advanced MHD+ -particle simulation codes Will demonstrate self-organization in core and edge in a way that cannot be totally predicted
FIRE operating modes I P (MA)B T T(s) N f BS Standard operating mode (LF) High-field (shorter pulse mode) Advanced Tokamak 1 st stability Reversed Shear Wall stabilized
Guidelines for Predicting Plasma Performance Confinement (Elmy H-mode) ITER98(y,2): E = I 0.93 R 1.39 a 0.58 n B 0.15 A i 0.19 0.78 P heat H(y,2) Density Limit: n 20 < 0.75 n GW = 0.75 I P / a 2 H-Mode Power Threshold: P th > (2.84/A i ) n B 0.82 R a 0.81
High Field: H = 1.0 (12 T, 7.7 MA)Low Field: H = 1.2 (10 T, 6.5 MA) Time (sec) Q > 10 for 9 secQ > 10 for 18 sec -heating ICRF total
S = (1+ 2 )/2 = a/R
High Field Low Field i /2 q 95
Physics Question: Role of the m=1 mode Ideal MHD theory predicts m=1,n=1 mode unstable at design for q 0 < 1 High-n ballooning modes also predicted to be unstable in the vicinity of and interior to the q=1 surface Proper physics description must take into account: energetic particle drive, kinetic stabilization, 2-fluid effects, and non-linear saturation mechanism This should be [and is] one of the major thrusts of the 3D macroscopic simulations communities FIRE will provide critical data point for code benchmarking and hence for extrapolations
Low Field: 10 T, 6.5 MA time (sec) surface number axis edge q = 1 q = 2 q = 3 PEST unstable eigenfunction at t=12.5 sec Balloon and Mercier stability
High Field: 12 T, 7.7 MA time (sec) surface number edge q = 1 q = 2 q = 3 PEST unstable eigenfunction at t=12.5 sec axis Balloon and Mercier stability
Comparison of unstable Eigenvalues Low Field 2 = High Field 2 =
UNSTABLE STABLE q'….(edge shear) I 90 …(edge current) Manickam FIRE nominal operating point is stable to kink modes. Relation of stability boundary and ELMs being studied = 3.3% N = 2.61 Stability boundary for plasmas with the FIRE , and A, and with q 95 =3.1
(From LaHaye, Butter, Guenter, Huysmans, Marashek, and Wilson) Physics question: NTM neoclassical tearing mode sets limits in many long-pulse discharges scaling of this to new devices largely result of empirical fitting of quasi- linear formula this is another major thrust of 3D macroscopic modeling effort active feedback looks feasible FIRE will provide critical data point
conventional operating modes the effect of H-mode profiles on MHD stability (Manickam, Chu,…) relation to ELMS, n ~ 5-10 peeling modes, bootstrap currents error fields and locked modes (LaHaye, et al) need to assess disruption effects reversed shear operating modes characterization of no-wall advanced mode for entire discharge (Ramos) wall stabilized advanced modes (GA/PPPL/Columbia experiments on DIII) other advanced modes off axis CD to raise q 0 (Kessel) edge current drive to improve stability (?) Other Physics Issues for FIRE
Example of Perturbation Study that can be done on FIRE: ICRF heating power increased by 5 or 10MW for 6 sec Other suggestions for XPs welcome !
Summary Overall, MHD stability looks favorable. Primary uncertainty is in non-catastrophic areas. MHD activity associated with q=1 surface edge currents due to H-mode pedestals (ELMs) neoclassical tearing modes. Active feedback requirements error fields and locked modes What will we learn? How does core self-organize with ’s and m=1 mode? How does edge self-organize with bootstrap and ELMs How does interior self-organize with NTM, at new ( *, *) How well can our codes predict these nonlinear events ?