(FT/P2-20) Technology Progress in Real-Time Control of MHD Modes at RFX-mod

A. Luchetta1), G. Marchiori, G. Manduchi, L. Marrelli, L. Novello, A. Soppelsa, C. Taliercio, P. Zanca1)
 
1) Consorzio RFX, Associazione Euratom-Enea sulla Fusione, Padova, Italy

Abstract.  The RFX-mod reverse field pinch (RFP) experiment applies active control on the MHD instabilities by acting on the magnetic field configuration at the boundary. These studies aim at acquiring expertise on the control of MHD modes that is also relevant for high beta discharges in tokamaks. To control the MHD activity, 192 saddle coils are installed, covering the entire torus surface. A large number of field and current measurements are acquired and processed in real-time (3x192=576 channels). Excellent results have been achieved by exploiting the real-time MHD mode control system, especially on MHD mode suppression at the boundary, toroidal loop voltage reduction, particle and energy time increase and duration of the plasma discharge. The paper discusses the technological enhancement developed in the last two years. Two new techniques have been introduced recently. The first one is referred to as Clean Mode Control (CMC). It consists in the correction of the aliasing, due to the sidebands produced by the discrete grid of coils, which determines a systematic error on the Fourier analysis of the measurements. As the CMC correction requires the simultaneous availability in the same real-time station of all acquired channels, a significant upgrade of the architecture of the real-time feedback system was necessary. The second technique consists in the introduction of a Multiple Input Multiple Output (MIMO) approach to substitute the Single Input Single Output model used so far, whose dynamic behaviour is not satisfactory. The MIMO approach relies on a full electromagnetic model of the coil and sensor system in the presence of passive structures. The MIMO controller has been developed following the concept of model decoupling. To compute the new algorithms without exceeding the real-time constraints, faster CPUs have been introduced and the data throughput in the real-time network has been optimized.

Full paper available (PDF)