cross-machine ntm physics studies and implications for iter

20th IAEA Fusion Energy - Vilamoura Cross-machine NTM physics - Buttery

Cross-machine NTM physics studies and implications for

ITERPresented by R. J. Buttery1,

On behalf of: P. Belo2, D. P. Brennan3, S. Coda4, L.-G. Eriksson5, J. Graves4, S. Günter6, C. Hegna7, T. C. Hender1, D. F. Howell1, H. R. Koslowski8, R. J. La Haye3, M. Maraschek6, M. L. Mayoral1,

A. Mück4, M. F. F. Nave2, O. Sauter4, E. Westerhof9, C. Windsor1, and the ASDEX Upgrade6 and DIII-D3 teams, and JET-EFDA contributors*.

1EURATOM/UKAEA Fusion Association2Association EURATOM/IST3General Atomics, San Diego4CRPP, Association EURATOM-Confédération Suisse5Association EURATOM-CEA, CEA-Cadarache

6MPI für Plasmaphysik, EURATOM Association7Dept. of Engineering Physics, University of Wisconsin8Association EURATOM-FZ Jülich IPP.9FOM-Rijnhuizen, Ass. EURATOM-FOM.*see annex 1, Pamela et al., Nuc Fus 43 (2003) 1540.


2

1

03210

ITER?ITER?

AUG

JET

[M. Maraschek et al., EPS03]

3/2 NTMs

~i*0.9

Are NTMs a problem for ITER?•NTM physics is expected to scale with *

–often observed in local onset scalings

low threshold in ITER?


0

1

2

3

4

4 6 8 10 12rho-i* x1000

be

tan

ASDEX-UDIII-DJ ET

N

i* x10-3

ITER

?IT

ER

?

?? ??

3/2 NTMs

[R. Buttery et al., NF04]

DIII-D AUGJET

Are NTMs a problem for ITER?•NTM physics is expected to scale with *

•But analyses in global suggests another possibility

–often observed in local onset scalings

low threshold in ITER?

•Key aspect in resolving the onset is the seeding process...

– which is it?


Time (s)

40

0Neutron rate/a.u.

Magneticsspectrogram/kHz

4/3 NTM

5/4

7/5?

n=1 sawtooth precursor

-30%

-13%

Which modes are a concern? •2/1 NTMs terminate performance & unacceptable in ITER

•3/2 NTMs significant effect– typically 15-20% on confinement

~ -30% in fusion power

– trace Tritium experiments showconsistent with ~50% fall ininward pinch in vicinity of island

•Higher m/n NTMs also impact fusion performance at low q95

JET: 3.7MA, 2.9T, q95=2.7:– up to 13% effect on confinement– up to 30% effect on neutrons


•2/1 NTMs terminate performance & unacceptable in ITER

•3/2 NTMs significant effect– typically 15-20% on confinement

~ -30% in fusion power

– trace Tritium experiments showconsistent with ~50% fall ininward pinch in vicinity of island

Which modes are a concern?

•Higher m/n NTMs also impact fusion performance at low q95

JET: 3.7MA, 2.9T, q95=2.7:– up to 13% effect on confinement– up to 30% effect on neutrons

AUG: 4/3 NTMs at q95=3.7:

– up to 20% effect on stored energy Time (s)

AUG


Content •NTM * scalings:

–Onset criteria for NTMs

–How do the scalings do?

•Role of the seed: the Sawtooth

– Influence on thresholds

–Sawtooth control

–Advances in sawtooth prediction

•The seeding process

–Sawtooth coupling mechanisms

–Other trigger mechanisms and effects

•Implications for ITER


metastable threshold

*

NTM onset criteria •NTMs driven by hole in bootstrap

–but onset criteria depend on small island stabilisation effects

– require a seed island to reachpositive growth

•Onset highly sensitive to seed size

– scaling of seeding processmay be the critical thing

–uncertainties both in seed needed and seed obtained

seed

positivegrowth

saturation

Island size

– these introduce a * dependencein the metastable threshold


How do the * scalings do?Pretty good in terms of underlying NTM physics and metastable threshold...

0

1.5

0 0.3i *

p

e (r

s/L

p)

JETDIII-DASDEX UITER

ITER scenario 2operation point

Regression fitagainst i

* alone:

Pe=5.5i*1.08

–power ramp-down experiments measure at which 3/2 NTM self-stabilises

– ITER baseline operation point deeply into metastable region

small triggers can excite mode

mode removal requires driving island down to small sizes


How do the * scalings do? •But they are not predictive of NTM onset and time on JET...

– stays close to NTM onset scaling prediction once H-mode reached

for both local and global parameter fits

JET Pulse No: 47282

0

0.2

0.4

0.6

0.8

1

1.2

-7 -6 -5 -4 -3 -2 -1 0

time relative to NT M onset (s)

exp

erim

enta

l

/

- f

it p

red

icti

on

Local fit

G lobal fit

rises 50%

while ratio close to 1

Time prior to NTM onset (s)

/

(NTM

fit

pre

dic

tion)

JET


How do the * scalings do? •But they are not predictive of NTM onset and time

– stays close to NTM onset scaling prediction once H-mode reached in JET


–similarly on AUG:

proximity to scaling is a necessary-but-not-sufficient condition for NTMs

–there must be an extra control parameter...

AUG

scaling

experiment


What is the hidden control parameter?

•Employ neural network to look for pattern in data...–automatic optimisation from

choice of 27 input parameters

– train to predict onset time

•Network successful –predicts decreasing time to

NTM as onset approached unlike * scaling!

–best network uses just N, * and sawtooth period

period even more significant than * 0

0.5

1

1.5

2

0 0.5 1 1.5 2

Actual time to NTM (s)

Predicted time to NTM (s)

Data forms two clumps: Fast evolution during rise Slow evolution at high

JET


Role of the sawtooth

•Long sawteeth can lead to many low modes (blue)

•Sawtooth control can offer substantial mitigation (red)– here achieved by:

ICRH phasing to avoid core pinch establishing sawtoothing heated

L-mode to avoid peaked profiles

– avoids all modes, even with much more heating power

•Sawtooth period plays key role in NTM onset

Time (s)

ICRH/MW

NBI/MW

termination

Magnetics: #58884 only

Expanded intime: 15-18s

2/1

3/2

4/3

0

15kHz

long sawtooth

SXR/a.u.

JET

20th IAEA Fusion Energy - Vilamoura Cross-machine NTM physics - Buttery [*Porcelli et al, NF44, 362]

Sawtooth control in ITER ITER has two possible strategies:

–early production to stabilise sawteeth - 50s may be possible* extend further with modified

start up and current drive but still limited and not steady state

Destabilisation of fast particle stabilised sawteeth now achieved:

– core ICRH stabilises sawteeth

Further progress with ECCD on AUG see Maraschek talk today

differ from resistive sawteeth in most present devices

– ICCD destabilises as inversion radius is approached

JET

– current drive destabilisation is this possible for fast

particle stabilised (ideal) sawteeth?...


Sawtooth prediction is key •Good progress in the theory...

–kinetic effects stabilise sawteeth at high rotation

– important in reconciling data from present devices

Experiment: Theory:

[Gra

ves

et a

l, PP

CF42

, 104

9]

eg: Rotation dependence on JET...

JET


Sawteeth with NNBI

[Kramer et al, NF40, 1383]

•JT60U also finds fast particles from energetic negative ion beams stabilising…

–350keV NNBI gives sawteeth of 300ms

– cf PNBI: 130ms

JT60U


Pure coPure counter

Bal

ance

din

ject

ion

Co-deposit

ion

insid

e q=1

stabilis

es

sawte

eth

Injection direction

Sawteeth with NNBI

Deposition

outside q=1destabilising in co-direction

[Kramer et al, NF40, 1383][Graves et al, PRL92, 185003]

•JT60U also finds fast particles from energetic negative ion beams stabilising…

•Explained by Graves: –finite ion orbit effects

change free energy

–350keV NNBI gives sawteeth of 300ms

– cf PNBI: 130ms

–depends on deposition location...

•Possible mechanisms for sawtooth control in ITER?


How is initial seed made?

–Magnetic coupling? NTM often too slow for

toroidal coupling to n=2

•Sawteeth often trigger 3/2 NTMs before the crash...

[*Hegna, Bull.Am.Phys.Soc.48, 280]

3 wave seeding possible: – bicoherence analysis shows phase lock

between driving (11+43) and 32 fields

but frequencies are not always consistent...

0 1 0 2 0 3 0 4 0 5 0- 5 0

- 4 0

- 3 0

- 2 0

- 1 0

0

1 0

2 0

1

( k H z )

2 (k

Hz

)

0 . 0 10 . 0 40 . 0 70 . 1 00 . 1 30 . 1 60 . 1 90 . 2 20 . 2 50 . 2 80 . 3 00 . 3 30 . 3 60 . 3 90 . 4 20 . 4 50 . 4 80 . 5 10 . 5 40 . 5 70 . 6 0

6 2 . 6 4 - 6 2 . 6 5 2 s

J E T # 5 1 9 9 52 / k H z

b i c o h e r e n c e o fn = 1 , 4 / 3 a n d 3 / 2

1 / k H z

SS pp ee cc tt rr oo gg rr aa mm :: ## 55 11 99 99 55

n = 1 p r e c u r s o r3 / 2

4 / 3

T i m e ( s )

n = 2

6 2 . 6 2 6 2 . 6 8

JET

– Ion polarisation effects? MHD can change island rotation*

potential to lower/reverse ionpolarisation effects enabling seeding

avoids need for frequency locking


Forced reconnection at crash •At low , long sawteeth trigger NTMs directly at the crash–excite multiple NTMs & 2/1 much more likely concern for

ITER

– codes such as NFTC and NIMROD now able to 3D model such processes in detail...

n=1 perturbed pressure

n=2 perturbed current

Te

•Example: forced reconnection inducing a 3/2 in DIII-D–NIMROD simulation now includes rotation shear:

– island is still destabilisedby forced reconnection

–but as island grows itsstructure becomes distorted by rotation

•Shows viability of mechanism for NTM seeding


Fishbone triggers - at higher N?

•Fishbones also trigger NTMs–3/2 NTM thresholds on AUG

generally higher than for sawtooth

AUG


Fishbone triggers - at higher N?

•Fishbones also trigger NTMs–3/2 NTM thresholds on AUG

generally higher than for sawtooth

•Fishbones recently observed to also trigger 2/1 NTMs:–at N=2.5 on JET

–although on JET these do not extend to low

unlike cases with fast particlestabilised sawteeth

i

*JET

6.6 Time (s)

Spectrogram:Magnetics /kHz0

7.4

254/3 NTM

fishbones

2/1 NTM

JET


Ideal triggers at high N? •At high N often see weak/no seeding…

–modes often near ideal limit

–a particular issue for hybrid scenario and 2/1 NTMs

4li

[Bre

nnan

et a

l, PP

10, 1

643]

•Separate studies show 2/1 NTM threshold lowered by error fields

– possibly an ion polarisation effect...

DIII-D

•Modelling of DIII-D case showspoles in classical tearing stability:

0

2

0 12Bpen

N /

[6.4

n0

.86 /

B0

.92]

Rotating onset Locked onset

DIII-D


Conclusions for ITER on NTMs •ITER deeply metastable to NTMs, but tractable?

–benign scalings for some NTM onset mechanisms– control of seeds possible for others

•Baseline scenario - key issues are fast particles & sawtooth– further triggers at higher N may remain at high N

•Hybrid scenario - main concern is 2/1 NTM (3/2 fairly benign) –does 2/1 onset threshold fall with *? - mitigate with high qmin?

•However, caution required for ITER...–adverse NTM physics scalings and high fast particle

populations

–need to confirm scalings of high N modes, especially 2/1 NTMs

–need to integrate control techniques into scenarios to develop ready to use tools (not lengthy research programmes) for ITER

Nevertheless, we now see the principal physics ingredients assembled, a new generation of codes

identifying the effects, and good progress in control and predictive

capability. Ongoing work is important to provide solutions for

ITER


–higher e-dissipation raises w0 ~(De)0.5

Transient transport events can seed NTMs

• Does not require frequency matching between MHD modes and the island

• May explain error field effects

•Ion polarisation effects depend on island rotation - apol

~

Ion polarisation destabilising

w

e*

i*

w0

–naturally leads to small islands via ion polarisation effects

Rotation from balance of ion and electron dissipation:

MHD event: - ergodises the edge- increases e- transport, De

- rotation more in e- direction

Islands produce 3-D structure in |B|

Neoclassical ion viscosity

e- transport(De)

[Hegna, Bull.Am.Phys.Soc.48, 280]








–best network uses just N, * and sawtooth period

period even more significant than *

•Sawtooth period plays key role in NTM onset

100% NBI

100% ICRH

JET


Preemptive current drive on DIII-D •Use real time MSE tracking to put ECCD on NTM resonant surface, raising NTM thresholds

DIII-D


Mode removal in ITER

– Island evolutions show scale length of small island term, wd, does not change much with *

–Mode removal in ITER will require driving islands down to similar size to those required in present devices

0

1

2

3

4

0.06 0.16i *

Fit

ted

wd

(cm

)

i*


Use of correct local parameters

•Studies on AUG show that NTMs track correctly calculated bootstrap parameter, better than p

2/1 NTM

3/2 NTM

AUG

AUG


How do the * scalings do? •But they are not predictive of NTM onset and time on JET...

– stays close to NTM onset scaling prediction once H-mode reached


– JET NTM onsets align well with natural discharge evolution

(clue: ICRH phasings)0

1

2

3

4

0 0.009

BetaN

-90 RF

no RF

+90 RF

Onset fit

i*

N

discharge histories

NTM onset

JET


* scalings sometimes work for NTM onset

•AUG discharges sometimes approach scalings from below and get NTM when the scaling is reached

AUG


Formalism - origin of * scaling •Evolution of island size w governed by modified

Rutherford: wr

dt

dw

rr

44222222 2.028

35.065.0

b

pol

d

GGJ

bdbsP

ww

wa

ww

a

ww

w

ww

war

•Example: ion polarisation term, apol f( g(,) i

classical tearing finite

island transport

ion banana width

field curvature

ion polarisation currents

bootstrap

small island terms stabilising - lead to minimum size and for growth

•seed get > seed need

•uncertainties in both

gwa

awr

seedpol

polseedisonsetp .

])/(1[

/..'

2*

wseed

seed

positivegrowth

onset p rises rapidly as w

falls

saturation p

ons

et

wseed


ITER possible figure? •Possibly how it looks...

q0

N

00.5

2

1

1

3

sawtooth: forced

reconnection 21/32/43

3wave/Hegna32

fishbone32/21

Poles in delta-prime 32/21

ELMs? 21?






0

0.5

1

1.5

2

0 0.5 1 1.5 2

Actual time to NTM (s)

Predicted time to NTM (s)

Data forms two clumps: Fast evolution during rise Slow evolution at high



–best network parameters:Parameters: Residual† Errors sawtooth i

* 34.3 17% sawtooth 34.4 20% i

* 35.7 26% 35.9 31%

i* 37.5 29%

†(predicted actual time to NTM)2

–Sawtooth period more useful than i* !


Future work•Continue good progress on sawtooth models

•Demonstrate sawtooth control with strong FP populations at high beta

•Explore NTM triggering mechanisms and ways to control them

•Find beta limit in hybrid scenario and how it scales

•Resolve NTM small island physics and its scaling -particularly for 2/1 modes

cross-machine ntm physics studies and implications for iter

Documents