Recent work on the control of MHD instabilities at ASDEX Upgrade

Post on 15-Jan-2016

26 views 0 download

Tags:

description

Recent work on the control of MHD instabilities at ASDEX Upgrade. S. Günter, J. Hobirk, P. Lang, P. Merkel, A. M ück, G. Pereverzev, ASDEX Upgrade Team Max-Planck-Institut f ür Plasmaphysik Garching, Germany. Sawtooth control by ECCD ELM control by plasma shaping and pellets - PowerPoint PPT Presentation

Transcript of Recent work on the control of MHD instabilities at ASDEX Upgrade

Recent work on the control of MHD instabilities at ASDEX Upgrade

S. Günter, J. Hobirk, P. Lang, P. Merkel, A. Mück, G. Pereverzev, ASDEX Upgrade Team

Max-Planck-Institut für Plasmaphysik Garching, Germany

• Sawtooth control by ECCD• ELM control by plasma shaping and pellets• Current profile control by off-axis NBI?• RWM physics on ASDEX Upgrade?• NTM control, see next talk

Sawtooth behaviour depends on NBI sources

one beam only

Sawtooth behaviour for different NBI sources

Off-axis heating only, leads to density peaking j’ decreased (increased off-axis BS current) diagmagnetic stabilization (* increased)

one beam only

Sawtooth behaviour for different NBI sources

two off-axis beams

2.0 3.0 4.0 5.0 6.0

t [s]

10000

20000

15000

12000

f [kHz]

Sawteeth/fishbones

(q=1)=0.2 (q=1)=0.2

(q=1)=0.1

two q=1 surfaces

• no sawteeth, but continous (1,1) activity

• two q=1 surfaces in the plasma (off-axis NBI-CD)

Experiments with slow Bt-ramp, 0.8 MW co-ECCD and 5.1 MW NBI

Sawtooth tailoring by co- ECCD

Influencing (1,1) mode activity by co-ECCD

• co-ECCD at pol = 0.4

• no sawteeth, only fishbones

• FB amplitude also decreases(SXR amplitude reduced by factor of 3)

Sawtooth tailoring by ctr-ECCD

Destabilisation of (1,1) activity by on-axis ctr-ECCD

• For ctr-ECCD deposition close to plasma center (here pol = 0.1) reversed q-profile destabilization of (1,1) mode

• No sawteeth or fishbones, but continous (1,1) activity

NTM control by sawtooth mitigation (off-axis-ECCD)

Co –ECCD:• no sawteeth as expected• Reduced fishbone amplitude• NTM triggered after ECCD (by ST)

Counter-ECCD:• NTM triggered by FB during ECCD

ELM mitigation: type II ELMs

Inner divertor

Outer divertor

power density

#15865 #15863

Consider two discharges with different plasma shape

Type I Type II

0.0 0.2 0.4 0.6 0.8 1.0 1.20

2

4

6

8

10

12

Ele

ctro

n de

nsity

[1019

m-3]

poloidal

15863 15865

0.0 0.2 0.4 0.6 0.8 1.0 1.20.0

0.5

1.0

1.5

2.0

2.5

Ele

ctro

n te

mpe

ratu

re [k

eV]

poloidal

15863 15865

but with similar edge temperature and density profiles

Influence of closeness to Double Null

n=8 peeling mode

• Ballooning Stability unchanged• Low-n modes become more stable (broad mode structure) • Stability of medium-n modes unchanged, but eigenfunctions more localised at plasma edge

Operational regime for type II ELMs

• closeness to DN/high • q95 > 3.5 • high n/nGW

Why do we need high density/high q95?

JET, ELM precursorslow n modes only for high density(Perez, Koslowski et al., IAEA 2002)

Hypothesis: type II ELMs only if low-n modes are stable

Influence of edge collisionality

jBS since * jBS since n

Theory: low mode number MHD activity destabilised by current gradient

Is type II ELM regime accessible for ITER?

• Higher density increases edge collisionality BS current density reduced

reduced drive for low-n modes

If not n/nGW, but collisionality counts, type II ELMs would not occur in ITER Other means for ELM control?

ELM mitigation: by pellets

Control of ELM frequency possible (each pellet triggers an ELM)

Control of ELM frequency by pellets

• small pellets (2 … 3x1019 D atoms, not strong fuelling)• Confinement degradation ~ f-0.16

(less than for frequency change by, e.g., heating power, density puff) ~ f-0.6

Mitigation of ELM size possible

• same plasma parameters• natural ELM frequency 52 Hz

Mitigation of ELM size possible

Energy loss per pellet triggered ELM as for type I ELMsat same frequency

Current profile control by off-axis NBI?

Redirected NBI box provides off axis deposition of 93 keV ions:

• NB driven current clearly seen by reduced OH flux consumption

• current profile changes seem much smaller than expected

Two off-axis beams, an example (#14513)

S3+S5 S6+S7

Plasma current

ne,0

pol dia

NBI

li

gas

Raus

One off-axis beam (Te change compensated by ICRH)

S3 S6

ICRH

Plasma current

ne,0

pol dia

NBI

li

gas

Raus

#18091

Strong change in li only for one-beam discharge

Change in li for one-beam casein agreement with ASTRA code

ASTRA

experiment

on- off-axis beams one-beam discharge

Very small change in li,much smaller than predicted(ASTRA li shifted up)

two-beam dischargeASTRA

experiment

q-profile for two-beam discharge (q=1 surface)

ASTRA predicts observable changeof q-profile, but no change measured(MSE, q=1 radius)

But: in the plasma centre (tor < 0.15) q-profile changes as two q=1surfaces at tor < 0.10 and tor = 0.2 observed

2.0 3.0 4.0 5.0 6.0

t [s]

10000

20000

15000

12000

f [kHz]

Sawteeth/fishbones

(q=1)=0.2 (q=1)=0.2

(q=1)=0.1

two q=1 surfaces

Current profile modifications due to one off-axis beam

Change in radius of q=1 surface is significant and agrees with ASTRA predictions

Current profile modifications due to one off-axis beam

Current profile modifications mainly caused by off-axis beam (ASTRA)

Comparison to MSE measurements

ASTRA predictions

Non-stiff electron temperature profiles for one-beam discharge

one beam (without additional heating)two beams (#14513)

Non-stiff ion temperature profile for one beam case

Ion temperature during off-axis NBI modeled by MMM95, agreement with measured pressures

To explain unchanged current profile one needs a particle pinch!

Anomalous particle pinches are well-known in theory (density peaking)

Simple picture: strong turbulence of background plasma redistributes particles while

maintaining the two adiabatic invariants and

with the density follows from

const.

Does theory predict such a particle pinch?

Need: full non-linear turbulence simulation with marker particles, in progress (B. Scott)

So far: quasi-linear GS2-calculations (G. Tardini, A. Peeters)

First results: particle pinch exists, but too smallTo be done: realistic density profile of fast particles, parameter scan

G. Tardini

Simulations for realistic wall structures (as planned for AUG)

low triang.

high triang.

Plasmaseparatrix+ 3 cm in midplane

Wall structures only relevant onlow field side (ballooning mode structure)

Realistic model for AUG wall structures

3D MHD code with 3d wall structures

• 3d MHD code CAS3D extended for

• 3d ideally conducting walls

• MHD eigenfunctions fully self-consistent

• Benchmark with 2d MHD code CASTOR successful

Simulation results for realistic wall structures

Efficiency of realistic wallcompared to closed wall

Without wall: ßmarg = 1 %

<ß> = 4.5 %

closed wall

rw/rpl

1.2 1.4 1.6 1.8 2.0

A

0.08

0.06

0.04

0.02

0.0

Wall resistivity causes mode growth on wall time (RWMs)

Further plans: - Resistive 3d walls (already started) - Feedback system (active coils to stabilise RWM)

Summary

• Sawtooth mitigation by localized ECCD demonstrated• Seed island control allows to control NTM onset

• type II-ELMs achieved by plasma shaping compatible with required plasma parameters: N, q95, H, n/nGW

• open question: does collisionality count? (BS current)

• ELM mitigation by pellets demonstrated• smaller pellets at higher frequency needed

• off-axis NBI current for current profile control only for non-stiff ion temperature profiles?

• RWM physics: 3D ideal MHD code with 3D ideally conducting wall structures, finite wall resistivity being implemented

Influence of edge density (BS current): ballooning modes

Experiment

Ideal ballooning limit:

ne = 9 1019 m-3

ne = 1.1 1020 m-3

Second stable regime low density

• Higher density increases edge collisionality BS current density reduced

Increased magnetic shear prevents access to second stable regime

Non-stiff ion temperature profiles for one-off-axis beam

electron temperature and density constant, but diamagneticpressure decreases hint to non-stiff ion temperature profiles

one beam discharge

two beam discharge

diamagnetic pressure nearly constant,pol increases for off axis beams(fast increases, mainly ||)

Non-stiff rotation profiles? (mode frequency also dependent on diamagnetic drift)

Strong reduction in (1,1) mode frequency for one-beam discharge

2.0 3.0 4.0 5.0 6.0

t [s]

10000

20000

15000

12000

f [kHz]

t [s]

2.0 3.0 4.0 5.0

1000

2000

5000

10000

20000two beams (#14513) one beam (#18091)

on-axis beams

off-axisbeams

on-axis beams

off-axisbeams

Good match of electron temperature profiles ...

… by additional central ICRH for the one-beam discharge to adjust Inductive current profiles

Two-beam discharges: so far non-symmetric beam deposition

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.0 0.2 0.4 0.6 0.8 1.0

Q1 /+60 /z=+9.5Q2 /+70 /z=+9.5

Q3 /–70 /z=+9.5Q4 /–60 /z=+9.5

Q5, Q6, Q7, Q8

A. Stäbler

Future two-beam experiments: try to match symmetricdeposition (closeness to DN)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.0 0.2 0.4 0.6 0.8 1.0

Q1 /+60 /z=+9.5Q2 /+70 /z=+9.5

Q3 /–70 /z=+9.5Q4 /–60 /z=+9.5

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.0 0.2 0.4 0.6 0.8 1.0

Q1 /+69.5 /z=±0

Q2 /+79.5 /z=±0Q3 /–60.5 /z=±0

Q4 /–50.5 /z=±0

Z = 0 z = 9.5 cm

Q5, Q6, Q7, Q8 A. Stäbler

CASTOR with antenna: calculate torque

Re P

jant B cos ~tor

Torque on the plasma due to external error fields:

1/

~

Maximum torque

An example: Interaction of NTMs with perturbation fields

No simultaneous large NTMs of different helicities observed in experiments

Analytic theory: • for NTMs stabilising effect of additional helical field can be proven for

small values of ||

• effect vanishes for ||

Is there an effect remaining for realistic values of || ?

If so: new stabilisation method for NTMs can be propsed:stabilisation by external helical perturbation fields

An example: Interaction of NTMs with perturbation fields

Many other problems, but: so far no non-linear MHD code can deal

with realistic ||

Proposal for a solution in non-aligned coordinate system

( )1//11//102

0 2

1−−⊥⊥ ∇⋅+∇⋅−=∇−

∂qbqbTT

t

rrχ

2//10//11//012

1 qbqbqbTTt

∇⋅−∇⋅−∇⋅−=∇−∂∂

−⊥⊥

rrrχ

3//11//12//022

2 qbqbqbTTt

∇⋅−∇⋅−∇⋅−=∇−∂∂

−⊥⊥

rrrχ

( )1111//0// 2

1−− ∇⋅+∇⋅−= TbTbq

rrχ

( )211001//1// TbTbTbq ∇⋅+∇⋅+∇⋅−= −

rrrχ

( )312011//2// TbTbTbq ∇⋅+∇⋅+∇⋅−= −

rrrχ

In the following, for simplicity (not in the code): Cartesian coordinates with one perturbation field component

Heat conduction equation for different Fourier components of temperature:

BBb ii

rrr=

… …

//2 qbTT

t∇⋅−=∇−

∂∂

⊥⊥ ( )Tbq ∇⋅−= //// χ

To close the equations one should not truncate the Fourier series in T, but in q heat flux along perturbed magnetic field line remains finite (nearly vanishing temperature gradients)

Fourier decomposition for perturbation

2//10//11//012

1 qbqbqbTTt

∇⋅−∇⋅−∇⋅−=∇−∂∂

−⊥⊥

rrrχ

3//11//12//022

2 qbqbqbTTt

∇⋅−∇⋅−∇⋅−=∇−∂∂

−⊥⊥

rrrχ

( )1111//0// 2

1−− ∇⋅+∇⋅−= TbTbq

rrχ

( )211001//1// TbTbTbq ∇⋅+∇⋅+∇⋅−= −

rrrχ

( )312011//2// TbTbTbq ∇⋅+∇⋅+∇⋅−= −

rrrχ

In the following, for simplicity (not in the code): Cartesian coordinates with one perturbation field component

Heat conduction equation for different Fourier components of temperature:

BBb ii

rrr=

//2 qbTT

t∇⋅−=∇−

∂∂

⊥⊥ ( )Tbq ∇⋅−= //// χ

To lowest order (for explanation): include only terms up to first order in q

T2 adjusts itself such that q||1 becomes small

( )1//11//102

0 2

1−−⊥⊥ ∇⋅+∇⋅−=∇−

∂qbqbTT

t

rrχ

What about the radial derivatives?

rrebbrr

11≈ bkib ⋅=∇⋅r

perturbation field:

1102

0 qr

bTTt r ∂

∂−=∇−

∂∂

−⊥

10112

1 qbkiTTt

⋅−=∇−∂∂

( )r

qqb

ii

r Δ

−−≈

−+

)12/()12/(

111

iqbki1

01 ⋅−≈r

1122

2 qr

bTTt r ∂

∂−=∇−

∂∂

⊥ ( )r

qqb

ii

r Δ

−−≈

−+ )12/()12/(

111

( )2

)2/1()2/1(

0111

−+ +⋅−≈

ii qqbk

Introduces an additional error or order (r)2 , but equations for each grid point ensure vanishing temperature gradients along perturbed field lines

simplest discretisationat i’s grid point

new scheme

Convergence properties: single magnetic island

||= 108

Still convergence only (r)2

But: absolute error reduced by factor of 10Improvement increases for larger ||

Magnetic islands with two helicities

Magnetic islands with two helicities

||= 1010

Magnetic islands seen in temperature contours, but still strong gradient in ergodic region

Magnetic islands with two helicities

||= 1012

Temperature gradient vanishes in ergodic region due to increased radial transport along magnetic field lines

An example: Interaction of NTMs with perturbation fields

Is there an effect remaining for realistic values of || ?

If so: new stabilisation method for NTMs can be propsed:stabilisation by external helical perturbation fields (next talk)

YES!

Improved Diagnostics at the edge

Comparison with theory possible ...

Plasma shape is important for ELM losses

• higher upper triangularity leads to bigger ELM losses

• can be explained by wider ELM affected region at higher triangularity

Our understanding of transition type I type II ELMs

Strong plasma shaping (high , closeness to DN)- stabilises low-n modes- reduces width of medium-n eigenfunctions

High edge density (reduced BS current density)- reduces drive for low-n modes- closes access to second stable regime for ballooning modes (limits achievable pressure gradient)

Can we expect type II ELMs in ITER ? (low collisionality!)