overview of the tcv digital real time plasma control

Overview of the TCV digital real time plasma control system and its applicationsIAEA 13th Technical Meeting on Plasma Control Systems, Data Management and Remote Experiments in Fusion Research

06.07.2021

Cristian Galperti

Federico Felici, Trang Vu, Olivier Sauter, Francesco Carpanese, Mengdi Kong, Gino Marceca, Antoine Merle, Alessandro Pau, Federico Pesamosca, Marcelo Baquero-Ruiz, Stefano Coda, Joan Decker, Basil Duval, Mateusz Gospodarczyk, Alexander

Karpushov, Blaise Marletaz, Daniele Carnevale, Nicolo’ Ferron, Jesse Koenders, Bob Kool, Gabriele Manduchi, Marc Maraschek, Peter Milne, Andre’ Cabrita Neto, Arthur

Perek, Emanuele Poli, Timo Ravensberger, Matthias Reich, Natale Rispoli

The TCV tokamak

Galp

ert

i C

ristian

2

IAEA 13th Technical Meeting on Plasma Control Systems, Data Management and Remote Experiments in Fusion Research, 5-8 July 2021

+NEW X3 (118GHz)

+NEW fast steerable antenna

+Flexible digital control system

(100% Simulink-programmable)+ NBH (1 and new 2)

Bt = 1.5T, Ip<1MA, a=0.25m R=0.88m 𝛋<2.8

The TCV tokamak

Galp

ert

i C

ristian

3


+NEW X3 (118GHz)

+NEW fast steerable antenna

+Flexible digital control system

(100% Simulink-programmable)+ NBH (1 and new 2)

>390 signals at 10-250 kHz and >10 cameras at 1kHz to be interfaced

Multi actuators experiment:16 PF coils, 2 Ohmic coils, 8 launchers, >3 gasvalves, 6 gyrotrons, 2 NBHs

Demanding constraint on main plasma control cycle time (<100 us, 10kHz)

Plant-wide real-time distributed system

Control research experiment: control system flexible and quick to implementnew ideas but rigorous enough to maintain the good ones.

Bt = 1.5T, Ip<1MA, a=0.25m R=0.88m 𝛋<2.8

> 10 years continuous development -> mature system

Control code mainly written and maintained in Simulink and autogeneratedfor MARTe2 capable systems, all data stored in MDSplus

New: introspective generated code, fully and automatically interfaceable to MDS+ for its run-time configuration

Various types of interconnected control subsystems:

• Low latency systems for hard real time control at fastest rate

• Oversampling systems for fast diagnostics acquisition followed by complex real-time analysis algorithms

• Multi-core computational systems for CPU hungry codes

• Real-time vision nodes for vision in the loop systems

EtherCAT for fast, flexible, distributed and cost effective I/O interconnection

The digital real-time control system

Galp

ert

i C

ristian

4

IAEA13th Technical Meeting on Plasma Control Systems, Data Management and Remote Experiments in Fusion Research, 5-8 July 2021

2021 top view

Galp

ert

i C

ristian

5


RF

M c

ard

Node 1:

Soft X-ray, Te & densityTypical rate 10kHz

cPCI board PC

2.1 GHz CPU

(Intel core 2 duo)

AD

C&

DA

C

Mo

du

les

64 DMPX Soft X-Ray

channels

4 X Te channels

14 FIR (density)

channels

RF

M c

ard19" industrial PC

3.5 GHz CPU

(Intel i7 6 cores)

RF

M c

ard

Node 7:

Multi CPU fast magnetics

analysis nodeTypical CPUs rate 1 kHz

Typical sampling rate 250kHz

19" industrial PC

3.0 GHz CPU

(Intel i7 8 cores)

AD

C

Mo

du

les

72 fast

magnetic

probes +

24 saddle

loopsNode 8:

Multi CPU fast ECE

analysis nodeTypical CPUs rate 1 kHz

Typical sampling rate 250kHz

19" industrial PC

3.0 GHz CPU

(Intel i7 8 cores)

54 ECE

channels

EtherCAT network (fieldbus)

NBH1 ECRH RF

power

AD

C&

DA

C

Mo

du

les

14 ECRH refs

35 coils power supply

cmds

3 gas valves

134 Magnetics

1 density (central FIR)

4 diamagnetics loops

21 coil currents

12 photodiodes

Node 2:

Multi CPU node (release)Typical rate 10 kHz

19" industrial PC

3.3 GHz CPU

(Intel i9 10 cores)

Node 3:

Multi CPU node, (debug)Typical rate 10 kHz

TOP

Launc

her 10

PW

M

mo

dula

tor

19" industrial PC

3.3 GHz CPU

(Intel i9 10 cores) 32 configurable

PWM outputs

32 GPIO outputs

RF

M c

ard

Node 5 and 6:

Multi CPU Computational nodeTypical rate 1 kHz

19" industrial PC

3.5 GHz CPU

(Intel i7 6 cores)

DIN rail emb. PC

1.46 GHz CPU

(Intel Atom 1 core)

BaratronsNode BKM1:

EtherCAT master nodeTypical rate 5 kHz

RF

M n

etw

ork

(ri

ng

)

RF

M c

ard

AD

C

Mo

du

les

RF

M c

ard

RF

M c

ard

RF

M c

ard19" industrial PC

3.1 GHz CPU

(Intel i9 14 cores)

10x

multispect

ral

imaging

system

Node MANTIS:

Realtime vision nodeTypical rate < 800 Hz

TOP

Launc

her 11

L4/L5

polariz

ers

Hardware highlights

Galp

ert

i C

ristian

6

IAEA13th Technical Meeting on Plasma Control Systems, Data Management and Remote Experiments in Fusion Research, 5-8 July 2021

Twin multi-core computer main machine control system

The two computers can runreal-time code in parallel on the same discharge (only 1 controlling it though)

10 kHz control rate on 192 inputs – 64 outputs system

Articulated EtherCAT network for agile low bandwidth – lowI/O systems interconnection

EtherCAT network (fieldbus)

NBH1 ECRH RF

power

AD

C&

DA

C

Mo

du

les

14 ECRH refs

35 coils power supply

cmds

3 gas valves

134 Magnetics



21 coil currents

12 photodiodes

Node 2:

Multi CPU node (release)Typical rate 10 kHz

19" industrial PC

3.3 GHz CPU

(Intel i9 10 cores)

Node 3:

Multi CPU node, (debug)Typical rate 10 kHz

TOP

Launc

her 10

PW

M

mo

dula

tor

19" industrial PC

3.3 GHz CPU

(Intel i9 10 cores) 32 configurable

PWM outputs

32 GPIO outputs

DIN rail emb. PC

1.46 GHz CPU

(Intel Atom 1 core)

BaratronsNode BKM1:

EtherCAT master nodeTypical rate 5 kHz

RF

M c

ard

RF

M c

ard

TOP

Launc

her 11

L4/L5

polariz

ers

Control code handling

Galp

ert

i C

ristian

7


MDS+ parameters and

waveforms DB

Simulink model

Common set of loader

mechanism

MARTe2 control system

Automatic code

generation =

Simulink

simulation

RT execution

Control code handling

Galp

ert

i C

ristian

8


MDS+ parameters and

real shot waveforms DB

Simulink model

Common set of loader

mechanism

MARTe2 control system

Automatic code

generation =

Sim

ulin

k

sim

ula

tio

n

RT executionFunctionally equivalent

MATLAB and C++

loader classes

RT

exe

cu

tio

n

MATLAB/Simulink MDS+ loader classes

Galp

ert

i C

ristian

9


MARTe2 based control architecture

Galp

ert

i C

ristian

10


AD

C&

DA

C

Data

sou

rces

TCV actuators

14 ECRH signals

35 coils power supply cmds

3 gas valves

TCV diagnostics

134 Magnetics



21 coil currents

RF

M B

oar

d

Dat

aso

urc

e

SimulinkWrapperGAM

CORE 1

Inte

rTh

read

Bro

ker

EtherCATBroker

ADC/DACBrokers

RFMBroker RFM network

Eth

erC

AT

Mast

er

Data

sou

rce

EtherCAT networkMDS+ Wavegenbroker

MD

S+

Waveg

en

Data

sou

rce

MD

S+

Para

mete

rs

Obje

cts

co

nta

iner

MDS+ Paramsinterface

MDS+ Wavegenbroker

MD

S+

Waveg

en

Data

sou

rce

MD

S+

Para

mete

rs

Obje

cts

co

nta

iner

MDS+ Paramsinterface

CORE 2

SHOT

CONFIG

MDS+

STORAGE CORE(S)

SimulinkWrapperGAM

MD

SWri

ter

Bro

ker

MDSWriterBroker SHOT

DATA

MDS+

MDSWriter

Datasource

MDSWriter

Datasource

Supervisory actuator manager and off normal event handling

Galp

ert

i C

ristian

11


T. Vu et al. IEEE TNS 2021

Complete configurable plasma supervisory, actuator manager and off normal events handlerto deal with multiple tasks and few actuators Beta real-time control togther with confinement

status (L/H) control via NBH 1

Vision in the loop detachment control

Galp

ert

i C

ristian

12


T. Ravensbergen et al., Nature Communications 12,

1105 (2021)

J.T.W. Koenders et al., 47th EPS Conference on

Plasma Physics (Virtual), P1.1058, June 21-25, 2

Divertor detachment front detected in real-time through vision processingalgorithms

0D detachment signal used to control detachment via SISO controller acting on a fuelling or seeding valve

TCV has a state of the art, powerful and flexible digital control system

Together with the extreme flexibility of the machine, this promotes TCV as a cutting edge facility for plasma control experiments

A number of domestic and international control activities have been done and are ongoing on this topic

Its software architecture is very general yet robust, it has allowed the seamless cohexistence of several control modules (e.g. basic plasma control together with advanced actuator manager)

The system will be expanded in the future by adding additionaldiagnostics nodes (e.g. real-time Thomson scattering) alongside pure processing nodes

Conclusions and outlook

Galp

ert

i C

ristian

13


Thank [email protected]

Galp

ert

i C

ristian

14


N. M. T. Vu et al., “Tokamak-agnostic actuator management for multi-task integrated control with application to TCV and ITER,” Fusion Eng.

Des., vol. 147, p. 111260, Oct. 2019.

F. Felici et al., “Integrated real-time control of MHD instabilities using multi-beam ECRH/ECCD systems on TCV,” Nucl. Fusion, vol. 52, no. 7,

p. 074001, Jul. 2012.

T. C. Blanken et al., “Model-based real-time plasma electron density profile estimation and control on ASDEX Upgrade and TCV,” Fusion Eng.

Des., vol. 147, 2019.

F. Felici et al., “Real-time physics-model-based simulation of the current density profile in tokamak plasmas,” Nucl. Fusion, vol. 51, no. 8,

p. 083052, Aug. 2011.

E. Maljaars et al., “Profile control simulations and experiments on TCV: a controller test environment and results using a model-based

predictive controller,” Nucl. Fusion, vol. 57, no. 12, p. 126063, Dec. 2017.

E. Poli et al., “TORBEAM 2.0, a paraxial beam tracing code for electron-cyclotron beams in fusion plasmas for extended physics applications,”

Comput. Phys. Commun., vol. 225, pp. 36–46, Apr. 2018.

H. Anand et al., “Distributed digital real-time control system for the TCV tokamak and its applications,” Nucl. Fusion, vol. 57, no. 5, p.

056005, May 2017.

T. Vu et al., “Integrated real-time supervisory management for off-normal-event handling and feedback control of tokamak plasmas,” IEEE Trans.

Nucl. Sci., pp. 1–1, 2021.

M. Kong et al., “Control of neoclassical tearing modes and integrated multi-actuator plasma control on TCV,” Nucl. Fusion, vol. 59, no. 7, p.

076035, Jul. 2019.

F. Carpanese, F. Felici, C. Galperti, A. Merle, J. M. Moret, and O. Sauter, “First demonstration of real-time kinetic equilibrium

reconstruction on TCV by coupling LIUQE and RAPTOR,” Nucl. Fusion, vol. 60, no. 6, p. 066020, Jun. 2020.

T. Ravensbergen et al., “Real-time feedback control of the impurity emission front in tokamak divertor plasmas,” Nat. Commun., vol. 12, no. 1,

p. 1105, Dec. 2021.

C. Galperti et al., “Integration of a Real-Time Node for Magnetic Perturbations Signal Analysis in the Distributed Digital Control System of

the TCV Tokamak,” IEEE Trans. Nucl. Sci., vol. 64, no. 6, pp. 1446–1454, Jun. 2017.

U. A. Sheikh et al., “Disruption avoidance through the prevention of NTM destabilization in TCV,” Nucl. Fusion, vol. 58, no. 10, p. 106026,

Oct. 2018.

Marceca G. et al, Proceedings, 47th EPS Conference on Plasma Physics: virtual conference, June 21-25, 2021

Carnevale, D., et al. "Runaway electron beam control." Plasma Physics and Controlled Fusion 61.1 (2018): 014036.

F Pesamosca et al, Model-based frequency decoupled control of plasma shape and position in the TCV tokamak, EPS 2021

overview of the tcv digital real time plasma control

Documents