A survey of the Tokamak fusion test reactor (TFTR) feedback control design

The TFTR is an experimental Tokamak fusion reactor being built by Princeton\´s Plasma Physics Laboratory for the Department of Energy. Its primary function is to demonstrate that the power that can be produced in a Deuterium Tritium reaction will exceed the power required to sustain the reaction. This will be the first Tokamak device that uses Tritium and has a large enough power density to achieve a break even in energy production. To achieve the power density requires high temperatures which are achieved by causing the current in the plasma to very rapidly increase. The high currents cause heating by the atomic losses. This ohmic heating is not sufficient, however, to achieve the required temperatures. Thus a source of high energy neutral deuterium atoms is "injected" into the plasma to cause further heating. This neutral beam heating is further supplemented with a rapid motion of the plasma (on the order of 50 m/s) to cause compression and adiabatic heating. Each of these heating effects, combined with the problem of initiating a plasma at the start of the reaction, imposes severe control problems because plasma current, position, and temperature are interrelated quantitites. The Plasma Current and Plasma Position feedback control were developed using linear optimal control theory. This paper surveys the results that guarantee when the control system is implemented on the actual machine, the control will be satisfactory. To this end, simulation and design models with eddy currents, plasma current diffusion, and plasma heating have been developed. The resulting control designs are shown to be very insensitive to modeling uncertainties.