Experimental Assessment of Actuator Management Strategies in DIII-D

Recent experiments in DIII-D have tested the capabilities of an actuator management algorithm based on nonlinear, real-time optimization. A multitude of control tasks will need to be carried out by a finite set of shared actuators in reactor-grade tokamaks such as ITER. This motivates the development and testing of actuator managers in present devices with the ultimate goal of extrapolating these solutions to future fusion reactors. Such actuator managers must calculate, in real time, the commands of the tokamak actuators that fulfill the necessary control requirements despite changing plasma conditions and actuator availability. In this work, the actuator management problem is posed as a nonlinear optimization problem that is solved in real time in a computationally efficient manner. The proposed approach does not depend on particular control objectives or actuators, facilitating the integration of the actuator manager with independently designed controllers. Initial DIII-D results in the steady-state high-q_min scenario have demonstrated the capabilities of the scheme to perform both simultaneous-multiple-mission and repurposing actuator sharing, which will be required in ITER and future fusion reactors.