1. of current ramp-up & ramp-down
Transport simulation
of current ramp-up & ramp-down
by F. Imbeaux* presented by X. Litaudon* *
Association Euratom-cea
F. Imbeaux, F. Köchl, V. Basiuk, J. Fereira, J. Hobirk, D. Hogeweij, X. Litaudon, J. Lönnroth, V. Parail, G. Pereverzev, Y. Peysson, G. Saibene, M. Schneider, G. Sips, G. Tardini, I. Voitsekhovitch
On behalf of : JET-EFDA contributors, Tore Supra work programme, ITER Scenario Modelling group (ITM-TF)
Appreciate to have the opportunity
In-vessel view into JET … no more time for promoting fusion…

2. Modelling of current ramp
Aim of the working group: model current ramp-up (and down) in ITER
Implications on PF system design,
H&CD methods for current profile shaping
flux consumption
Main issues are related to the transport model
 try to validate a model against present experiments
Validation: li , Vloop, Te, Ti test against several JET, Tore Supra AUG, experiments (ohmic, NBI, LHCD, ECCD)
Up to now, energy and current diffusion are modelled
L mode edge plasmas 2

3. Consideration on current ramp transport modelling
Choice of the transport model : scaling-based, empirical, 1st principles
Li prediction and Flux consumption strongly depend on the Te at r >0.5
Model has to predict Te up to r = 1 with L-mode edge
Scaling-based transport model are
a priori less sensitive to the assumptions on the boundary conditions (stiffness issue, drift wave models not accurate close to the edge)
Have hopefully a correct dependence on Ip and machine size for extrapolation
May miss several physical effects  try to validate on extensive range of machines / heating schemes
empirical models: Bohm/gyro-Bohm and Coppi-Tang have been tried as well
Validation on existing experiments is essential
Excellent opportunity for code-to-code benchmarking :
Jetto and Cronos used in comparison with experimental data
Astra, Jetto, Cronos used in ITER predictions 3

4. Database from JET, Tore Supra, AUG with various H&CD mix
Current ramp-up for base-line (q95~3) & AT scenario (q95~5) :
Pulse
H&CD scheme
JET 72823
LHCD+ ~1MW NBI for MSE&CXS
JET 72818
~1MW NBI for MSE &CXS
JET 71828
Ohmic
JET 71827
JET 70497 TS
ECCD TS
LH+ECCD
AUG (ohmic) just added last week
TORE SUPRA }
To be transferred to ITPA database
Current ramp-down from base-line scenario (2.4T/2.7MA, q95~3):
Pulse ramp-down additional heating
JET slow none
JET fast none
JET fast NBI heating until ramp-down
JET fast NBI heating after ramp down (in low Ip phase) 4

5. Scaling-based model
Scaling-based model: energy content of ohmic or heated Ip ramps with L- mode edge correctly modelled by either
H mode scaling with H98 = 0.4 – 0.5
L mode scaling with H97 = 0.6
ci = ce, renormalised so that
Fixed c(r) shape : power balance chi’s during ramp-up tend to be rather flat, then strong increase towards the plasma edge : c(r,t) = A(t)(1+6 r r20)
Boundary Te (r = 1) taken from experiment (guessed from ECE)
Ne profile taken from experiment (JET: inversion of interferometry data)
Flat Zeff assumed, <Zeff> taken from experiment (Bremsstrahlung) 5

6. Calibration on JET shot 70497 (constant q95 ~3 ohmic ramp-up)
Model ci = ce, renormalised to IPB98 scaling, H98 = 0.5 (mimics L mode), radial shape 1+6 r r20 (adjusted to fit experimental Te profile peaking).
Te is correctly reproduced.
CRONOS simulation
red - : Simulation
blue * : ECE
Purple * : Thomson scattering
2.6T/2.6MA, q95~3
t = 0.5 s
t = 2 s
t = 4 s
t = 5 s 6

7. Apply same model on another ohmic ramp for AT scenario (JET #72818)
Model ci = ce, renormalised to IPB98 scaling, H98 = 0.5 (mimics L mode), radial shape 1+6 r r20 (adjusted to fit experimental Te profile peaking).
Te is well reproduced for r > 0.6. Even if larger deviations occur inside r = 0.6, they almost do not affect the li evolution.
Li slightly overestimated, Dli ~ 0.08, ~ measurement accuracy.
Vloop good, slightly underestimated
2.7T/1.8MA, q95~5
t = 6 s
Simulation
ECE
Thomson scattering
time (s)
time (s) 3 4 5 6
3.5 4
4.5 5
5.5 7
ITPA meeting Milan, October 2008, F. Imbeaux

8. Ohmic Ip Ramp-down (JET 72202)
JETTO
2.4T/2.7MA down to 1.0MA
Experimental
Bohm/gyro-Bohm*
Scaling, H98 = 0.5
Both models follow equally well li and the volume averaged Te
time (s)
*without non-local Bohm multiplier 14 16 18 20 22 24 8

9. Ohmic Ip Ramp-down (JET 72202)
Comparison of the two models : some difference in Te profile peaking, that Bohm/gyro-Bohm provides a better agreement (for JET)
A possible approach consists in combining: profile dependence as B/gB + scaling renormalisation for multi-machine capability ? r
JETTO
Experimental
Bohm/gyro-Bohm
Scaling, H98 = 0.5 r 9

10. LHCD assisted ramp-up in AT scenario (JET 72823)
Scaling base model, H98 = 0.4
LHCD calculated during interpretative run
Li slightly overestimated, Dli ~ 0.1
2.7T/1.8MA, q95~5 2 3 4 5 6
time (s)
time (s) 3 4 5 6 3 4 5 6

11. LHCD assisted ramp-up for AT scenario (JET 72823)
Excellent fit of the volume averaged temperatures, both electron and ion
Simulation
ECE
Thomson Scattering
2.7T/1.8MA, q95~5
t = 3 s
t = 4 s
t = 5.5 s

12. LHCD assisted ramp-up for AT scenario (JET 72823)
NBI blips (MSE & CXS) during current rise :
CXS & MSE measurements
2.7T/1.8MA, q95~5
Ion temperatures from CXS TI TI
t = 4 s
t = 5 s
Simulation
* CXS

13. LHCD assisted ramp-up for AT scenario (JET 72823)
NBI blips (MSE & CXS) during current rise :
CXS & MSE measurements
2.7T/1.8MA, q95~5
t = 4.5 s
t = 5.5 s
Comparison to MSE q- profile (EFTM)

14. First attempts to test GLF23 in JET ramp-up phase
When applied from the edge at high q (JET 71828, ohmic, 1 s after breakdown, 5 < q < 15 )
Very low transport predicted near the edge (r = 1)  barrier forms and non monotonic Te profiles appear
(2.6T/2.6MA, q95~3 at flat top ) Te Ti
JET 77251
5.5s
JET t=1s
After breakdown
Exp
Normalised radius
When trying to patch the edge (impose c = 10 m2/s from r = 1 to r = 0.75), the same problem appears at r = 0.75
Possibility to use GLF23 ? : on the whole radius ? At high q ?

15. Tore Supra ramp-up experiments
Fast current ramp (0.7 MA/s), plateau Ip = 0.9 MA reached at t = 1 s
off-axis co-ECCD (.7 MW) and/or LHCD (.8 MW) during ramp at t = 0.25s
Same li evolution, but different Te evolution & time of first sawtooth
Blue : usual scaling-based model H98 = 0.5
Green / red: two experimental measurements of li
TS40676 : ECCD
TS40679 : LH+ECCD

16. 1st set of simulation: ITER ramp with constant boundary shape
L-mode transport model: (r) validated on JET & TS with H98 = 0.5.
The edge temperature set to Tb=20*Ip[MA] eV, although with the chosen transport model Tb assumption is not that important.
Use a formula for Zeff (used by the ITER team):
Zeff = ( x(0.5/ne)2.6)), Carbon is main impurity
The current ramp follows the ITER reference one with 15MA at 100s:
1.5MA t=4s, 3.0MA t=8s, 7.5MA t=30s, 13MA t=75s, and 15MA t=100s.
The simulations start at 3MA (t=8s), with an initial condition for the q- profile (li=1) and a temperature profile.
ITER divertor shape from Ip3MA: full bore plasma.
densities <ne>/nGW = 0.15; 0.25; 0.4 (low to medium density).
Use a variation of heating: Ohmic, 10MW and 20MW of ECRH with power deposition at mid-radius (heating only, no current drive). 16
ISM Working group

17. Modelling of ITER ramp with various heating & densities
<ne>/nGW = 0.25
<ne>/nGW = 0.4
<ne>/nGW = 0.15
10 MW ECRH
<ne>/nGW = 0.25
20 MW ECRH
<ne>/nGW = 0.25 17
Working group
D. Hogeweij et al, EPS 2008

18. Modelling of ITER ramp with various heating at <ne>/nGW = 0.15
Te profile evolution. Hollow profiles achieved transiently with off-axis ECRH, still present at the end of the ramp
Te ITB may be obtained with such hollow q-profiles (not in the model)
D. Hogeweij et al, EPS 2008 18
Working group

19. 2nd set of simulation: ITER ramp with evolving boundary shape
ASTRA
CRONOS
JETTO
JET like
Tore Supra like 19
ISM Working group

20. 2nd series: ITER with prescribed evolving boundary shape
Ohmic ITER ramp-up, time-dependent plasma boundary (preset), using the same model as before, H98 = 0.5
Code comparison: rather close agreement between CRONOS and JETTO, in spite of the differences in the treatment of equilibrium, transport coefficient renormalisation, …
Detailed code comparison starting from basic simulations going on between ASTRA and JETTO 20
Working group

21. Perspectives
Up to now, mainly the scaling-based model was used, work being extended to empirical models like Bohm/gyro-Bohm and Coppi-Tang
Test on multiple machine data : JET, Tore Supra and Asdex Upgrade
Rather good agreement with experimental data (on Te, li and Vloop prediction) has been found in several cases. None of these models can be ruled out yet.
Analysis still in progress, more detailed trends / observations to be given when the full database analysis will be completed
Multiple transport codes working on the same dataset allow detecting bugs / sensitivity to unexpected parameters / assumptions
Coppi-Tang implementation : still doubts on a few details (definitions of some quantities)  need to have the same as in TSC
Extend data base to other heating schemes, other machines (ITPA) should contribute to further validate the models
Continue testing other models :
introduce more sophisticated radial dependence in scaling-based ? Try further GLF23, through first attempt was quite unsuccessful
Include free-boundary equilibrium calculations : try for the end of 2009

22. JET ohmic ramp-up 71828
Electron energy content well fitted by H98 = 0.4 or H97 = 0.6
Also by “old Bohm/gyro-Bohm model”, i.e. without the edge “H-mode” factor. However, can we expect that it would work so well on other tokamaks ?
Coppi-Tang model not accurate
Black dots : experimental data from LIDAR (Thomson scattering)
G. Sips et al, EPS 2008

23. First attempts to test GLF23 in JET ramp-up phase
When applied from the edge at high q (JET 71828, ohmic, 1 s after breakdown, 5 < q < 15 )
Very low transport predicted near the edge (r = 1)  barrier forms and non monotonic Te profiles appear
(2.6T/2.6MA, q95~3 at flat top ) Te Ti
Normalised radius
When trying to patch the edge (impose c = 10 m2/s from r = 1 to r = 0.75), the same problem appears at r = 0.75
Possibility to use GLF23 ? : on the whole radius ? At high q ?