Plasma Models for Real-Time Control of Advanced Tokamak Scenarios

Full text: An integrated plasma profile control strategy, ARTAEMIS, is being developed for extrapolating present-day advanced tokamak (AT) scenarios to steady state operation. The approach is based on semi-empirical modeling. It was initially explored on JET, for current profile control only. The present paper deals with the generalization of this strategy to simultaneous magnetic and kinetic control. The methodology is generic and can be applied to any device, with different sets of heating and current drive actuators, controlled variables and profiles. In AT discharges, the multiple parameter profiles which define the plasma state (safety factor, plasma rotation and pressure, etc...) are known to be strongly coupled, and a limited number of heating and current drive (HCD) actuators are available for plasma control. The system identification algorithms take advantage of the large ratio between the magnetic and thermal diffusion time scales and have been recently applied, in their full version, on both JT-60U and DIII-D data. On JT-60U, an existing series of high-bootstrap-current (70%), 0.9 MA non-inductive AT discharges were used. The actuators consisted of four groups of neutral beam injectors corresponding to on-axis perpendicular NBI, off-axis perpendicular NBI, on-axis co-current tangential NBI and off-axis co-current tangential NBI. On DIII-D, actuator modulation experiments were carried out in the loop voltage control mode to avoid feedback in the response data from the primary circuit. The plasma current varied between 0.7 and 1.2 MA. In addition to the loop voltage, the four HCD actuators were co-current, counter-current and balanced NBI, and electron cyclotron current drive. The reference plasma state was that of an AT scenario which had been optimized to combine non-inductive current fractions near unity with 3.5 <βN < 3.9, bootstrap current fractions of > 65%, and H98(y,2) =1.5. The determination of the device-specific, control-oriented models that are needed to compute optimal controller matrices is discussed. The response of the relevant parameter profiles to variations of specific actuators can be satisfactorily identified from a small set of dedicated experiments. This provides, for control purposes, a readily available alternative to first-principle plasma modeling. (author)