Real Time Control of advanced scenarios for steady-state tokamak operation

12010 ITER Summer School, Austin, USA, 31 May - 04 June Xavier Litaudon

AssociationEuratom-cea

i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Real Time Control of advanced scenarios

for steady-state tokamak operation

Xavier LITAUDONCEA-IRFM, Institute for magnetic fusion research (IRFM)

CEA Cadarache

F-13108 St Paul Lez Durance. France

e-mail: [email protected]

Courtesy: J.F Artaud, A. Bécoulet, S. Brémond, D. Campbell, J. Ferron, G. Giruzzi, C. Gormezano, E. Joffrin, S. H Kim, D. Mazon, D. Moreau, P. Politzer, T. Suzuki, T. Tala



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

OUTLINE

Challenges for continuous operation– continuous tokamak reactor operation

– real time control requirement Real time control of kinetic & magnetic energy

– optimal profile for steady-state & MHD stable profiles

– approaches to profiles control Real time fusion D-T burn control

– burn control with dominant bootstrap and -heating ? Control of core performance with the plasma facing

components constrains– wall scenario compatibility issues

– simultaneous control of core & edge

TOWARDS REAL TIME PROFILE CONTROL ?



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Optimisation of tokamak concept

IC

BT

BP

IP

IP = Iinductive + INon-Inductive

Non-Inductive Current Drive – Externally driven, e.g. waves injection

• To drive 15MA on ITER requires 150MW

• 150MW coupled power requires ~ 1GW fusion

– Internally driven Pression: bootstrap effect

Efficient reactor at high Q =Pfus/Padd relies on the optimisation of bootstrap current

Long-Pulse Operation

→ IP = INon-Inductive

[e.g. Kikuchi M Nucl. Fusion 1990, Gormezano C ITER physics basis Nuc Fus 2007]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Challenges of continuous tokamak operation

Fully non-inductive regime High confinement & bootstrap current Real time control of kinetic & magnetic

configuration close to operational limits with a large fraction self -heating & bootstrap

Technology of Long Pulse Operation

– Coils, Plasma Facing Components, Structure Materials, Heating &Current Drive systems, Diagnostics, data acquisition, fuel cycle…

Worldwide research activity: physics, modelling, technology



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Towards a continuous tokamak reactor

self-heating

bootstrap

A scientific and technical challenge

Existence and control of a self-organised plasma state for continuous tokamak operation ?

i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Pfusion/Padd DD

~400s

0%

20%

Tore Supra

Q ~ 1

2s

10%

20%

JET

Q ~ 10

400-3600s

70%

10-50%

ITER

Q ~ 30

Continuous

80 to 90%

60-80%

DEMO

duration

JT60-SA

DD

~100s

0%

>60%



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Basic standard tokamak parameters

Safety factor q– number of toroidal turns for

one poloidal turn Confinement

– H=/scaling H~1 – scaling = I R2 P-2/3

Stability– q >1 – N = t/(Ip/aB) ≤ 3

Toroidal– t = P/pBtor

– pBtor=B2tor/20

Btor

Poloidal – p= P/pBpol

– pBpol=B2pol/20

Bpol



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

STEADY-STATE REACTOR : Optimisation of QDT & Bootstrap current

QDT t B2

Iboot/Ip ½ p= -½ q NBootstrap optimisation

• fusion power + bootstrap high N , , Bsincetp (1+2)

N

• Optimise shaping

• Stability -q & pressure -wall stabilisation

• Confinement

Fus

ion

gain

opt

imis

atio

n

t [%

]

p

10

8

6

4

2

03210

=2

=3

=4

=5lo

w q

95 li

mit

stability limit

Indu

ctiv

e q 95

~3SS q 95

~4-5



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Enhanced performance for non-inductive regimes

1

Pla

sma p

ress

ure

0 radius

L-mode

H-mode + internal transport barrier

H-mode

Pedestal



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Towards Long Pulse Operation on ITER

Inductive operation – Q 10 Ip~15MA 400s

‘Intermediate’ – Q ~5-10 Ip~12MA 1000s

fully non-inductive – Q~5 Ip~9MA 3000s– Active research activity– Integration of physics &

technology

Inon-inductive/Ip Ibootstrap/Ip

50%

7%

20%

100%

50%

20%



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

ITER STEADY-STATE OPERATION

Steady-State operation at Q ~ 5 (P~Padd)

with full non-inductive current drive + optimized current & pressure profiles

Iboot/Ip 50%

N~3, H98(y,2) ~1.5

nl~7x1019m-3

Ti/Te ~ 1

D~3000s

0 1

1

2

5q95~5 (9MA) at high ,

qmin ~ 1.5-2.5

qo-qmin ~0.5

[Gormezano Nuc Fus 2007, Campell Pop (2001), Green et al PPCF 2003 & ITPA steady-state group]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

nHe/ne=6%

N=2

N=3.5

*He/ E =5

Gyro-Bohm scaling

CONTROL OF CORE CONFINEMENT IN STEADY-STATE ITER OPERATION

Control of high confinement for steady state: H/H~20% Q/Q~50% (at Q~5)

favourable effect of density peaking while nHe/ne< 6%

CRONOS-0DITER-AT: Ptot=73MW, Vloop=0

-0.9

0

[Litaudon PPCF 2006]

ITER SS target



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Bringing Fusion to its “Reactor Era” requires an innovative programme of “discharge mastering”, combining:– real time control of the magnetic/kinetic

configuration (non-linear and time effects)– real time control of component integrity– high-level algorithms and control schemes– a consistent set of simulation tools:

• first principles (“PFlops”)• integrated modelling (“CPU hours”) • fast simulators (“~ 10 ms”)

The decisive role of real time control

[A. Becoulet & G.T. Hoang PPCF 2008 and Joffrin et al PPCF 2003 ]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Control requirement: plasma scenario

Fusion Power

Plasma Current

Inductive Flux

D-T Fuelling

Plasma Density

-particle Fraction

Additional Power

PF

Res

etA

nd

Rec

oo

l

Po

l. F

ield

Mag

net

izat

ion

Bre

akd

ow

n

Cu

rren

tR

amp

Up

Hea

tin

g

Bu

rnte

rmin

atio

nC

urr

ent

Ram

p D

ow

n

Time(s)

CurrentFlat Top



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Example of scenario: JET plasma

JET #67687



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Real Time plasma profile reconstruction(EQUINOX code)

[Joffrin et al PPCF 2003, Joffrin et al PPCF 2007 ]

#57336



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

S.H. Kim et al, PPCF 2009

Plasma scenario : 15MA H-mode ITER scenario

Tokamak simulation: free-boundary equilibrium (DINA-CH) & transport evolution

(CRONOS)

ITER



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

MHD Stability issues & real time control

Edge Localized Modes Damage to Plasma Facing Components

Neo-classical tearing modes Limiting pressure, risk of disruption

Resistive wall modes Limiting pressure

Disruptions Device safety

Fast particle modes Limiting -heating, CD

N

qmin

1 2 30

no wall limit

with wall limit

2/1 NTMs

Tearin

gSn

akes

Fish

bones

Saw

teeth

Hybrid scenarios

Hybrid scenarios

Advanced scenariosAdvanced scenarios

[R. Buttery, EFPW 05]

Necessity for Real Timefeedback control

& localized CD



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

PROFILE CONTROL REQUIREMENTS FOR STEADY-STATE OPERATION

[Shimada et al NF 2004, Polevoi et al IAEA 2002]

ITER SS operation above the no-wall limit

at ITER wall position, the marginal is sensitive to details in q and pressure profiles

qmin~2.4, N <3.85

qmin~2.1, N<2.6

qP

ITER



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

ITER Heating & Current Drive actuators: FLEXIBILITY for profile Control

NNBI ICRH ECRH LHCD

Heating - electrons

- broad deposition

-70% ions

-central heating

-electrons

-localised

-start-up

-electrons

-localised

-off axis

CD - yes

- broad deposition

-no global CD

- Central (MHD)

-yes

-localised

(MHD)

-yes

-off-axis >0.7

Torque yes no no no

Fuelling small no no no

170 GHz1MeV/D- 5 GHz

20MW/CW20MWCW20MW/CW33MW/CW

40-55 MHz

ITER



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Control of a self-organised state ?

[Politzer et al NF 05]

two time scale: fast (blue) & Slow (red) -heating dominant in burning reactor

Actuators

Plasma



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Magnetic/Kinetic configuration RT control



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Oscillation in confinement observed in steady-state tokamak plasmas

limiting the fusion perfomance

Tore Supra* & DIII-D**(i) non-linear coupling between j & T jLH(j,T), jboot (j,T), (j,T)

(ii) non-linear interplay of heating, CD & MHD s<0, double tearing,

ideal MHD limits …

PROFILE CONTROL REQUIREMENTS

ITER SS extra coupling via -heating(i) non-linear coupling between j & T jboot(j,T), (j,T), P(T)

(ii) non-linear interplay of heating, momentum, CD & MHD P(T), limits,

TAE (-particles), … *Giruzzi et al PRL 03**Politzer et al NF 05



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

oscillations of core electron temperature – non-linear interplay

between q-profile, transport and heat sources (and MHD)

Vloop=0 for 6 min, 1 GJ

Neutron (x1010/s)

Zeff ~2

Ti(0) =1.6 keVdensity (x1019m-2)

LH Power (MW)

Transformer flux (Wb)

Te(0) = 4.8 keV

2003

Te(R,t)

Non-linear behaviour in non-inductive regime

[Giruzzi PRL 03; Imbeaux PRL 06; Maget Nuc Fus 06]

TORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Oscillation in bootstrap-dominated regime

[Politzer et al NF 2005]

Time (s)

DIII-D Vl=0 (open

transformer circuit)

1 2 3 4 5

W/W~50% Iboot/Ip~85%N~p~3.3

#119767 #119770

0.5

0.7

0.6

Ip (MA)q95~10



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Non-optimal : narrow profiles and steep gradients

Optimal ITB : broad profiles with moderate gradients- MHD stability - broad Jboot and Jtot

- broad ne for reduced impurity accumulation

control of ITB radius, strength & width

ACCESS TO HIGH N OPTIMAL PROFILES ?

p=cstITB width

10

8

6

4

2

0

Pre

ssu

re (

a.u

.)

1.00.80.60.40.20.0normalised radius

ITB radius



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

High N requires broad pressure profiles

[Sips IAEA 2004 , Litaudon et al PPCF 2004]

ITPA database

1

2

3

4 N

2 4 6p0/<p>

HybridReversed shearQDB

Stability limit improves with ITB radius and width* control of confinement & q-profile ?

*Lao et al APS 99



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

REAL TIME CONTROL OF KINETIC ENERGY

Operation close to the no-wall stability limit while avoiding disruptions

Real time control

of neutral beam heating to match a neutron yield production

Similar results obtained on DIII-D, JT-60U

[G. Huysmans et al., Nucl. Fusion, 1999, C. Gormezano et al FST 2008 ]

JET

t=



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Control of electron temperature gradient

[Mazon, Litaudon, Moreau et al PPCF 02]

PLHCD to slow down q(r,t)

PNBI RT controlled by neutron

PICRH RT controlled by s/LTe where LT= T/T

proportional-integral



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

RT control of magnetic energy

feedback control for non-inductive operation:

1. Primary voltage Vloop-Vloop, ref

2. PLHCD Ip ref - Ip

3. n//-LHCD Liref- Li

with Li /

(a)

More recently*

n//-LHCD Hard X Ray width representative of LHCD absorbed & J profile

TORE SUPRA

[Wijnands Nuc Fus 1997,Litaudon PPCF 1998, *Joffrin Nuc Fus 2007]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Feedback control of qO or qmin during the plasma current ramp-up phase

Change of plasma conductivity through electron heating – ECRH or NBI

RT q-profile using MSE data

[J. Ferron et al Nuc Fus 2006]

RT control of minimum q, qmin

DIII-D



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

RT control of minimum q, qmin

High-p ELMy H-mode

– Ip=0.8MA, Bt=2.5T,

q95=5.8, ne=1.8x1019m-3,

p=1.2-1.5.

Off-axis LHCD control:

– dPLHCD/dt=(qmin,ref - qmin)

– =2MW/s Without control

– qmin down to 1.3

Interaction j & Te

– Requirement for T & J control

1.31.7

interlocksby arcingN//=1.7

JT-60U

[Suzuki et al NF 2008]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Multi-Input-Multi-Output (MIMO) profile control

sensors plasma actuator

q(s)=K(s) P(s)

All transfer functions are in matrix form

actuatorssensors

K(s)

P(s)q(s) qtarget

G(s)

Controller

Controller

P(s)= G(s) q(s)

disturbance

noise

First approach: control based on pseudo-inverse of the steady-state gain matrix

G(s)= gc[1 + 1/(is)] K(0)inv

[D. Moreau et al Nucl Fus 2003, D. Moreau et al Nucl Fus 2008]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

control in the prelude phase

"Model Based" control on 5 q-values

PLHCD is controlled to minimise (q-qtarget) in the least square sense

Access to various q-profiles

[ D. Mazon et al PPCF 2003, D. Moreau et al Nucl Fus 2003]

RT q-profile control with off-axis LHCD in low -phase



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

RT q-profile control in high -phase

Multi-Input, multi-output control 3 actuators: LHCD, ICRH, NBI

Model based SVD control: steady-state gain matrix deduced from open loop experiments

undershoot

end of control

start of control

setpoints

[D. Moreau et al Nucl Fus 2003]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Control of kinetic & magnetic profiles

bootstrap pressure

pressure Internal Transport

Barrier

q-profile

Jbootstrap

Jext.

pressureJplasma

• non-linearly coupled profiles • two time scales

• res ~10s-100s• E ~1-5s



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

3

2

1

5

4

3

2

1

838281

737271

636261

535251

434241

333231

232221

131211

*

*

*

.

G

G

G

G

G

G

G

G

P

P

P

KKK

KKK

KKK

KKK

KKK

KKK

KKK

KKK

T

T

T

NBI

LH

ICRHmodulations

around a reference

steady-state:

Galerkin’s projection of 1/q and ρ*T

ICRH

LHCD

NBI

0 1x

ρ*T

0 1x

q or =1/q


[D. Moreau et al Nuc Fus 2008, T. Tala et al Nuc. Fus 2005]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

q profile and *Te control


[Laborde et al., PPCF (2005),

Moreau et al. Nuc Fus 2008]

The controller minimizes

quadratic error:



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h


[Laborde et al., PPCF (2005), Tala et al Nuc. Fus 2005, Moreau et al Nuc Fus 2008]

*Te

1/q

q

MeasuredReference

Limitation of steady-state gain matrix approach: controller response slow during fast events

Development of an optimal control (time dependent model)



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Two time-scale modelsslow: magnetic profiles, fast: kinetic profiles

Uslow(t)

Slow model: resistive time scale

ddt

A slow (t)Bslow Uslow(t)

V (t)

Ti(t)

slow

Cslow (t)D slow Uslow(t)

Slow controller :

opt. feedback

Fast model: momentum / thermal confinement time scale, = t <<1

d

dV( )

Ti( )

fast

A fast V( )

Ti( )

fast

B fast U fast ()

Fast controller :

V(t)

Ti(t)

V(t)

Ti(t)

slow

V( t)Ti( t)

fast

U( t) Uslow(t)Ufast( t)

opt. feedback

[Moreau et al Nuc Fus 2008]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Identification of a dynamical model Future: closed loop experiments

Generic approach: can be applied to any tokamak with any set of actuators and real-time measurements

Model identified on JET, JT-60U and DIII-D (on-going)

[Moreau et al Proc. 48th IEEE Conf. on Decision and Control China 2008]

Slow model for 1/q(x)

N°70318

JT-60U

1/q(x) & V(x)

x=0.1

x=0.2

x=0.3

x=0.4

x=0.5

x=0.6

x=0.7

x=0.8

x=0.9

x=1.0

x =0.5

x =0.6

x =0.7

x =0.8

x =0.9

x =0.4

x =0.5

x =0.6

x =0.7



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Modelling of real time control of Te and q

ITER model based profile control (CRONOS) : 12MA hybrid mode

S.H. Kim et al, EPFL thesis No 4500

sub to PPCF

ITER

q control @

300s

Te control @

400s

4 actuators

transport

model based

on global

confinement

scaling law

real-time

update of the

static models



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

RT control in dominated bootstrap & -heating regimes : open issues

existence of a stable and unique state with self-consistent pressure and current ?

control at high Iboot with PPadd ?

– rely mainly on q-profile control with minimum external CD?

– pressure control requirements should be minimized

model based control?– strong requirements in terms of integrated transport modelling

‘simulate’ in present day experiment -heating with additional electron heating source – Experiments performed on JET & JT-60U to mimick -heating

in standard ELMy H mode regimes: how to extend to non-inductive operation ?



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

SIMULATION OF ALPHA PARTICLE PLASMA SELF-HEATING USING ICRH UNDER REAL-TIME CONTROL

ICRH applied in response to real-time measured plasma parameters (e.g. neutron rate) simulating the self-heating effect

part of the external heating plays the role of auxiliary heating

Demonstrate stable control of the simulated burn?

[T. Jones, EPS 2001]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

[Takenaga et N Fus 2008]

SIMULATION OF ALPHA PARTICLE PLASMA SELF-HEATING USING NBI UNDER REAL-TIME CONTROL



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Mimick the self-alpha heating and self-driven current in present day non-inductive experiments

Integrated fusion burn control experiment to prepare Long Pulse Operation on ITER & DEMO

– ICRH/ECRH ‘mimic’ the -power → P and Pfus

– ECCD/LHCD ‘mimic’ bootstrap → fBoot > 50%

– Remaining powers for control → Pcontrol

→ Qeff = Pfus /Pcontrol ~ 5-20

Could be tested on long pulse tokamaks : Tore Supra, JT-60SA, EAST etc …

“Proof of principle” through modelling using a simplified version of CRONOS, METIS

Combination of H&CD powers & density actuators are required for burn control:

– Powers : fast and precise control

– Density : slow and coarse control



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Fusion burn simulation at high bootstrap fraction

P

bootbootboot I

Ig+=f

1 alpha

ICRHECRHECRHLHICRH

fus

Pr+P+P+P

P=Q

1

high bootstrap burning plasma– -heating mimicked using ICRH/ECRH

– bootstrap mimicked using LHCD

Fuelling, Heating/CD feedback control

Equivalent fboot and Q

P ICRHα =g fus RDD iDDD TfnR 2 α

ICRHECRHαECRH Pr=P

PLHboot=η LH I LH

bootI LHboot=gboot I boot

controlICRH

αICRHICRH PP=P

controlLH

bootLHLH PP=P

offsetECRH

bootLH

αICRHECRHECRH PPPr=P

αICRHECRHfus Pr+P 15

Control

nn=n__

0

_

Slow time scale



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

METIS : A tool for (burn) control simulation

METIS work-flow organisation

Mixed 0D and 1D equations

Coupled to “Simulink” for real time control design

Fast dynamic simulation

– ~ 1minutes for 300 time slices

– 2s per time slice when coupled to Simulink

Included in the CRONOS suite to prepare integrated modelling

[Artaud, Litaudon et al EPS 2008]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Modelling of Burn control with an L-mode confinement

Control phase

TORE SUPRA

[Artaud, Litaudon et al EPS 2008]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Modelling of Burn control with an L-mode confinement

Target: Q = 15

Target = 80 %

Burn control @ 10s

control feasible for range of P (gfus

):

Q = 15 obtained

target is missed

system runs away

Add nD control (slow):

widens the operational space available for burn control while minimizing the external heating sources.

Control phase

TORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Modelling of Burn control with ITB: Q~15 at fboot~70% at Vloop=0

Control phaseTORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Modelling of Burn control with ITB: Q~15 at fboot~70% at Vloop=0

Control phase

TORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

ITER plasma facing components– Be wall– Divertor: W-baffle + CFC – CFC/W changeout during

shutdown preceding D and D-T phase

All components actively cooled!

ITER

Wall Scenarios Compatibility:• maximum performance• minimum T-retention• minimum erosion• maximum life-time of PFC

Plasma Facing Components: Wall scenario compatibility

ITER



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed hguard/ICRHlimiter

aux.limiter

hor.plate

lower PSL

roofbaffle

W-coating2006/2007

W-coating until 2003/2004

W-coating2004/2005

W-coating2005/2006

ASDEX

Upgrade

Plasma Facing Components: Wall scenario compatibility, R&D in EU

Effort in EU tokamaks to investigate PFC-scenario issues• Tore Supra: long pulse operation with actively cooled CFC components• ASDEX Upgrade: conversion to all tungsten PFCs complete • JET: installation of beryllium wall and tungsten divertor in 2010

TORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Steady-state scenario and Wall compatibility issue

Control of power & particle exhaust

– Control transient & stationary power loads

– Control of fast particle losses

– Control of particle exhaust & recycling according to fuelling requirements, capacity of Tritium extraction plant & necessary Helium removal

[Litaudon et al EPS PPCF 2007]

simultaneously with

Control steady-state fusion performance

– control of kinetic & magnetic energy (confinement & stability) including impurities

– control fully non-inductive operation

boundary conditions core

plasma

Core conditions



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

simultaneous control of transient (ELMs) and stationary power load

[P Lang et al Nucl. Fusion 2005]

Exhaust power controlled by impurity injection:

• noble gases usually chosen• limit heat flux & divertor

temperature to minimize erosion

Feedback control of gas flow:

• radiated power to be actively adjusted

• heat flux to target adjusted in response to variations in loss power (fusion power)

ASDEX Upgrade



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Simultaneous real time control of core confinement and heat exhaust

Control of confinement by acting on D2 flux

– Highest density at a given confinement

Control of prad/ptot by acting on Argon flux– reduce divertor heat

load Control matrix from

open loop exp.

[Dumortier P. et al EPS 2004]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Simultaneous control of – Density by gas puffing

near the top of the vessel– Divertor radiation by gas

puffing in divertor – Energy content by NBI

power

Non-diagonal matrix control between actuators & sensors deduced from open loop experiments

Simultaneous real time control of core confinement and heat exhaust

[Fukuda et al FST 2002]

JT-60U



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

TORE SUPRA

ICRH antenna IR view

Antenna septum

Current profile n// and PLH

PLH , PICRH Surface Temp.

« Search optimisation » algorithm

Target: broadest current profile

PLH

n//

Start

Optimum found

Simultaneous Profile and Heat load control

[Joffrin Nuc Fus 2007, Barana, Fus Eng Des 07]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

CONCLUSION & FUTURE DIRECTION

New & active field of research that needs a wide range of knowledge from plasma physics to control engineering, experiments & modelling

Major & recent experimental progress to tackle real time control issues for steady-state tokamak operation

Challenging issues for future research direction Integrated modelling towards tokamak simulator ?

– Develop generic methods, modelling of RT diagnostics, control loops, plasma physics, tokamak control system etc

integration and compatibility of the control schemes ?– integrate control of fusion performance & stability with control

of power and particle exhaust during the whole plasma operation demonstration of the controllability of bootstrap-

dominated regime with dominant -heating ?– experiments & modelling



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Basic tokamak concept is not stationary

Inductive operation

toroidal coils

shaping coils

magnetic circuit

primary transformer circuit

plasma current (secondary)



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Courant autogénéré (bootstrap)

Bt

ne

• courant autogénéré par le plasma

• produit par le gradient de pression (densité et température)

J(r) n(r+) - n(r-) >0

v//

J<0J>0

piégés

théorie transport collisionnel gradient des particules piégés: courant // "amplifié"

par : transfert du moment // aux électrons circulants par

collisions entre électrons piégés et circulants équilibré par le transfert de moment des électrons aux

ions circulants



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

CONTROL OF KINETIC ENERGY [1/2]

Density gradients

– Large ne for ITG stabilisation but in contradiction with broad pressure & low impurity accumulation requirements

Ti/Te ratio

– ITER Ti/Te~1 contrary to present day experiments

Plasma impurities injection– ITB expansion observed on DIII-D & JET but core impurity

accumulation should be avoided



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

ExB flow shear – Local modification of v with NBI but weak control of v on ITER*

– Control of the diamagnetic term & poloidal rotation with pressure but unfavourable size scaling with * as ExB/ITG *

– positive feedback loops between ExB and pressure

-stabilisation– local control of q (increasing q) and pressure

– positive feedback loop between pressure &

– No size effect (* ) since is the crucial parameters

q-profile– Local reduction of magnetic shear with off-axis non-inductive CD

– Positive feedback with the bootstrap current (no size effects)

– Optimum q-profile for stability & confinement ?

q-control has good prospect for burning plasmas

CONTROL OF KINETIC ENERGY [2/2]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

ITB

ITB: Transport bifurcation

20

10

0

Ti (keV)

3.83.63.43.23.0

R (m)

Review paper : Connors et al NF 2004

Gormezano et al 1996 IAEA

ITB: A layer plasma turbulence reduction

Plasma radius

Litaudon et al 1999 PPCF

Garbet 2004 PPCF

Contour lines of electric potential.



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Shear flow stabilisation

Differential plasma rotation: – Break or reduce the radial extend

of the plasma eddies criterion for stabilisation

v

v

r

r

linE

E drdV



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Effect of magnetic shear on transport

q & magnetic shear: s = r/q dq/dr s>0

s<0

5

1234

q profile

shear

5

1234

q profile

shearH-mode scenario

Advanced

tokamak scenario 0

-1-2

Normalised radius

Low or negative magnetic shear -twist plasma eddies-Rarefaction of rational surfaces close to s~0-Reduction of the growth rate



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Compatibility of ITB with H-mode ?

Combination of Internal & Edge TB lead to: improved confinement improved MHD stability (broad pressure) higher off-axis Jboot

Strong H-mode detrimental to ITB ?:

Large ELMs degrade ITB

JET

[Bécoulet M et al PPCF 2003]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Solutions for a successful ‘wedding’ ?

Radiation

[Litaudon et al PPCF 2007]

3

2

1

0

HIPB98(y,2)

15

10

5

0

3.0

2.0

1.0

0.0

PNBI [MW]

IEFCC [kA]

86420

[keV]

Teo

Tio

4

2

0[10

19 m

-3]

9.08.07.06.05.04.0Time [s]

neo

ne ped

D

JET Pulse N°68973

Magnetic conf. Ergodisation

1.5

1.0

0.5

0.0108642

Time [s]

p

D

20

15

10

5

0

PNBI

PICRH

[MW]

PLHCD

4

3

2

1

0

Wdia[MJ]

N

JET Pulse N° 68802

6

4

2

0

[1019

m-3

]

108642Time [s]

neo

D

20151050

PNBI

PICRH

Prad

[MW]

3210

PLHCD [MW]

Ip [MA]

10

5

0

[keV]

Teo Tio

JET Pulse N° 67869

JET



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Real Time Control of plasma states

[Imbeaux et al EPS 2009]

Te0

PLH

PICRH

ObtainedHot Core requested

MHD (5) detected

TORE SUPRA

Physicist defines the various plasma states (q-profile)

the controller checks whether the plasma is in the reference state and, if not, modifies the LHCD power in the relevant direction:

– state < ref. state increase LHCD

– state > ref. state decrease LHCD



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h #33982

Steady-state plasma as a two-cycles non linear oscillator

#33986

Transition from small to giant oscillations due to MHD event when qmin~2

TORE SUPRA



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Transport Modelling: Bootstrap dominated JET regimes with power upgrade

[Bizarro, Litaudon et Tala NF 2007 & IEEE on plasma science 2008]

Oscillations relaxation



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Control of ion temperature gradient

5

10

10

20R/LTi

NBI [MW]

No ITB reference

Pulse 58179 (3.0T; 1.8MA, =0.2, n/nG=0.4)

Time [s]

ICRH [MW]LH [MW]

0.01

*Te

0.02

0.030.04

2 4 6 8 10 120

1 H [V]

2 n [m-3] x1019

PNBI controlled by LTi

proportional-integral (PI)

[Joffrin et al PPCF 03]



i

a ca

r mf

c ar ed h

i

a ca

r mf

c ar ed h

Progress in Active NTM Control

ASDEX Upgrade

• Controlled ECCD Deposition in ASDEX Upgrade, DIII-D, JT60-U

[Zohm PPCF 2007]

[Humphreys, Nuc Fus 2007]

Documents

Real Time Control of advanced scenarios for steady-state tokamak operation