Experimental results from the COMPASS-C tokamak reveal a sharp threshold in amplitude above which externally applied static resonant magnetic perturbations (RMPs) induce stationary magnetic islands. Such islands (in particular, m = 2, n = 1 islands) give rise to a significant degradation in energy and particle confinement, suppression of the sawtooth oscillation and a large change in the impurity ion toroidal velocity. The observed threshold for inducing stationary (2,l) islands is consistent with a phenomenological resistive MHD model which takes into account plasma rotation (including poloidal flow damping) and externally applied resonant fields. Broadly similar results are found for applied fields other than m = 2, n = 1. Other results from RMP experiments are also discussed, such as the stabilization of rotating MHD activity, stimulated disruptions and extensions to the disruptive density limit. Finally, the likely effect of field errors on large tokamaks is briefly examined in the light of the COMPASS-C results.

Oscillating MHD modes in JET are often observed to slow down as they grow and generally stop rotating (lock) when the amplitude exceeds a critical value, then continue to grow to large amplitudes (b̃r/Bθ ∼ 1%). The mode can grow early in the current rise or after perturbations, such as a pellet injection or a large sawtooth collapse, and maintain a large amplitude throughout the remainder of the discharge. Such large amplitude quasistationary MHD modes can apparently have profound effects on the plasma, including stopping central ion plasma rotation, reducing the amplitude and changing the shape of sawteeth, flattening the temperature profile around resonant q surfaces and reducing the stored energy. Perhaps most important, large amplitude locked modes are precursors to most disruptions. Some large amplitude modes can be avoided by proper programming of the q evolution. The apparent reasons for the mode locking in a particular location are discussed and a comparison with theory is made.

A non-rotating helical magnetic field perturbation grows in the JET plasma when the electron density falls below a minimum value. The growth of this perturbation usually leads to a disruption. The helical phase of the perturbation is reproducible for a given configuration of the applied poloidal and toroidal magnetic fields. Small asymmetries in the applied magnetic fields are shown to cause the growth of the perturbation. Various sources of such field errors in JET have been identified. The existence of a minimum density for a stable plasma can place a severe constraint on the operation of JET. The magnitude of the minimum density depends upon various plasma parameters. Neutral beam injection can be used to reduce the minimum density.

In this paper we present results from the first use of neural networks for real-time control of the high-temperature plasma in a tokamak fusion experiment. The tokamak is currently the principal experimental device for research into the magnetic confinement approach to controlled fusion. In an effort to improve the energy confinement properties of the high-temperature plasma inside tokamaks, recent experiments have focused on the use of noncircular cross-sectional plasma shapes. However, the accurate generation of such plasmas represents a demanding problem involving simultaneous control of several parameters on a time scale as short as a few tens of microseconds. Application of neural networks to this problem requires fast hardware, for which we have developed a fully parallel custom implementation of a multilayer perceptron, based on a hybrid of digital and analogue techniques.