Speaker
Dr
Federico Felici
(Eindhoven University of Technology)
Description
To maintain a high-performance, long-duration tokamak plasma scenario, it is necessary to maintain desired profiles while respecting operational limits. This requires real-time estimation of the state of the profiles, monitoring of their evolution with respect to predictions and known limits, and their active control within a desired envelope. Model-based techniques are particularly suitable to tackle such problems due to the nonlinear nature of the processes and the tight coupling among the various physical variables. Control-oriented models required for such techniques have been recently developed for particle, temperature and current density profiles. These are implemented in the RAPTOR control-oriented physics-based tokamak simulation code. RAPTOR is capable of tokamak discharge plasma profile evolution in real-time. It includes the main nonlinear dependencies in the profiles, effects of sawteeth and NTMs, as well as all relevant sources, sinks, heating and current drive actuators.
We present various new applications of control-oriented profile evolution models for TCV, ASDEX-Upgrade and ITER. Real-time estimation of profiles is performed on TCV and ASDEX-Upgrade by combining diagnostic data with real-time model-based predictions in a dynamic state observer. By monitoring the differences between measurements and predictions, unexpected signals can be determined that may come from diagnostic faults, actuator failures or unexpected evolution of the plasma itself. These strategies for model-based real-time scenario monitoring are tested on TCV and ASDEX-Upgrade and can provide a first line of defense against easily predictable disruptions that have a clear origin, minimizing the use of last-resort disruption mitigation measures.
To maintain the plasma in the desired state for the duration of the discharge, profile controllers are designed that can deal with varying constraints in the plasma and actuators as well as provide ahead-of-time warning of constraint violations. At the same time, a real-time actuator allocation algorithm performs a prioritization of control objectives and assigns resources to each controller. Simulations of combined plasma q-profile, temperature, and NTM control for an ITER scenario are shown, representing a first example of a coupled, resource-aware control scheme that will benefit future tokamak operation.
| Country or International Organization | The Netherlands |
|---|---|
| Paper Number | EX/P8-33 |
Primary author
Dr
Federico Felici
(Eindhoven University of Technology)
Co-authors
Ms
Anna Teplukhina
(EPFL-SPC)
Mr
Bert Maljaars
(Eindhoven University of Technology)
Dr
Chiara Piron
(Consorzio RFX)
Dr
Chris Rapson
(IPP Garching)
Dr
Cristian Galperti
(EPFL-SPC)
Dr
Dick Hogeweij
(DIFFER)
Hugo Van den Brand
(DIFFER)
Dr
Jean-Marc Moret
(EPFL-SPC)
Dr
Jonathan Citrin
(DIFFER)
Dr
Louis Giannone
(IPP Garching)
Prof.
Maarten Steinbuch
(Eindhoven University of Technology)
Prof.
Marco De Baar
(DIFFER)
Mr
Mark Scheffer
(Eindhoven University of Technology)
Dr
Matthias Willensdorfer
(IPP garching)
Dr
Olivier Sauter
(EPFL-SPC)
Mr
Ondrej Kudlacek
(Consorzio RFX)
Dr
Paolo Piovesan
(Consorzio RFX)
Dr
Remy Nouailletas
(CEA-IRFM)
Dr
Stefano Coda
(EPFL-SPC)
Mr
Thomas Blanken
(Eindhoven University of Technology)
Dr
Trang Vu
(CEA-IRFM)
Dr
Wolfgang Treutterer
(IPP Garching)
