G. De Tommasi2),
(EX/P1-2) New Dynamic-Model Approach for Simultaneous Control of Distributed Magnetic and Kinetic Parameters in the ITER-like JET Plasmas
1) Euratom-CEA Association, Cadarache, France
2) Euratom-ENEA-CREATE Association, Napoli, Italy
3) Ecole Polytechnique, Palaiseau, France
4) Euratom-VR Association, EES, KTH, Sweden
5) Euratom-UKAEA Association, Culham, Abingdon, U. K.
6) Euratom-TEKES Association, VTT, Finland
Abstract. Real-time control of radially distributed parameters was achieved for the first time on JET in 2004, in view of developing integrated control of steady state advanced tokamak scenarios in ITER. The controller was based on the static plasma response only. It was successful in achieving various targets but was found too sensitive to some rapid plasma events. A technique for the experimental identification of a dynamic plasma model has now been developed, taking into account the physical structure and couplings of the transport equations, but making no quantitative assumptions on the transport coefficients or on their dependences. The approach newly implemented aims to use the combination of heating, current drive and poloidal field systems in an optimal way to achieve a set of simultaneous tasks.
Theoretical plasma transport analysis has led to the choice of the relevant state variables, and of a set of constraints to be imposed on the corresponding state-space model in order to best reproduce the dynamic response of the plasma profiles to heating power and loop voltage changes. To cope with the high dimensionality of the state space and the large ratio between the various time scales involved, our model identification procedure and controller design make use of a multiple-time-scale approximation and of the theory of singularly perturbed systems. Conventional optimal control is recovered in the limiting case where the ratio of the thermal confinement time to the resistive diffusion time vanishes.
Closed-loop simulations of the new controller have been performed in preparation for experiments in 2006, using either the full-order dynamic plasma response or the two-time-scale approximate response. Comparisons show that the reduced-order controller can perform almost as well as the full-order optimal one. Feedforward compensation of some disturbances such as density changes is also included. Finally, the possibility of controlling the plasma boundary flux together with the plasma shape is a new feature to be exploited on JET together with the experimentally identified physics-based plasma model. Simultaneous control of the plasma shape, the magnetic and kinetic plasma profiles and the boundary flux can thus be attempted and will be needed to achieve non-inductively driven advanced tokamak discharges in ITER. The most recent experimental progress on JET is reported.
Full paper available (PDF)