Physics issues are discussed for compact stellarator configurations which achieve good confinement by the fact that the magnetic field modulus |B| in magnetic co-ordinates is dominated by poloidally symmetric components. Two distinct configuration types are considered: (1) those which achieve their drift optimization and rotational transform at low β and low bootstrap current by appropriate plasma shaping; and (2) those which have a greater reliance on plasma β and bootstrap currents for supplying the transform and obtaining quasi-poloidal symmetry. Stability analysis of the latter group of devices against ballooning, kink and vertical displacement modes has indicated that stable β values on the order of 15% are possible. The first class of devices is being considered for a low β near term experiment that could explore some of the confinement features of the high β configurations.

Externally applied magnetic fields are used on the Texas Experimental Tokamak (TEXT) to study the possibility of controlling the particle, impurity and heat fluxes at the plasma edge. Fields with toroidal mode number n = 2 or 3 and multiple poloidal mode numbers m (dominantly m = 7) are used, with a poloidally and toroidally averaged ratio of radial to toroidal field components 〈|br/Bø〉 ≅0. 1%. Calculations show that it is possible to produce mixed islands and stochastic regions at the plasma edge (r/a ≥ 0.8) without affecting the interior. The expected magnetic field structure is described and experimental evidence of the existence of this structure is presented. The edge electron temperature decreases with increasing 〈|br/Bø〉, while interior values are not significantly affected. The implied increase in edge electron thermal diffusivity is compared with theoretical expectations and is shown to agree with applicable theories to within a factor of three.

Experimental devices for the study of the physics of high beta (β ² 4%), low aspect ratio (A º 4.5) stellarator plasmas require coils that will produce plasmas satisfying a set of physics goals, provide experimental flexibility and be practical to construct. In the course of designing a flexible coil set for the National Compact Stellarator Experiment, several innovations have been made that may be useful in future stellarator design efforts. These include: the use of singular value decomposition methods for obtaining families of smooth current potentials on distant coil winding surfaces from which low current density solutions may be identified; the use of a control matrix method for identifying which few of the many detailed elements of a stellarator boundary must be targeted if a coil set is to provide fields to control the essential physics of the plasma; the use of a genetic algorithm for choosing an optimal set of discrete coils from a continuum of potential contours; the evaluation of alternate coil topologies for balancing the trade-off between physics objectives and engineering constraints; the development of a new coil optimization code for designing modular coils and the identification of a 'natural' basis for describing current sheet distributions.

Key physics issues in the design of a high-β quasi-axisymmetric stellarator configuration are discussed. The goal of the design study is a compact stellarator configuration with aspect ratio comparable to that of tokamaks and good transport and stability properties. Quasiaxisymmetry has been used to provide good drift trajectories. Ballooning stabilization has been accomplished by strong axisymmetric shaping, yielding a stellarator configuration whose core is in the second stability regime for ballooning modes. A combination of externally generated shear and non-axisymmetric corrugation of the plasma boundary provides stability to external kink modes even in the absence of a conducting wall. The resulting configuration is also found to be robustly stable to vertical modes, increasing the freedom to perform axisymmetric shaping. Stability to neoclassical tearing modes is conferred by a monotonically increasing ι profile. A gyrokinetic δf code has been used to confirm the adequacy of the neoclassical confinement. Neutral beam losses have been evaluated with Monte Carlo codes.