Model predictive profile control and actuator management in tokamaks

Operating tokamaks (a nuclear fusion device) requires a control system to ensure that physics goals are met while satisfying operational limits. The control of two key parameters using MPC, as well as managing actuators between multiple control tasks is considered.

PhD Candidate: ir. Bert Maljaars
Supervisor:  dr. ir. F. Felici
Promotor: prof. dr. M.R. de Baar, prof. dr. ir. M. Steinbuch
Project Financing: TU/e, Eurofusion
Project Period: November 2012 – December 2016

The tokamak is a nuclear fusion device in which a plasma is confined using magnetic fields and heated to achieve fusion conditions. Tokamaks require a plasma control system (PCS) to help ensure that physics goals are met, while remaining within operational and machine limits. Such a PCS comprises several control tasks that need to be executed simultaneously with a limited set of actuators.

In the first part of this project, the control task of achieving and maintaining a current and pressure distribution inside the plasma is considered. Due to the presence of  important plasma physics and hardware limits, a model predictive control methodology is chosen that deals with those limits while minimizing a performance criterion. Closed-loop simulations for the ITER, ASDEX Upgrade and TCV tokamaks and experiments at the TCV tokamak are part of this project.

The second part of this project deals with the problem of actuator management when multiple controllers need to use the same set of actuators. This part involves demonstrating simultaneous control of multiple quantities in the ITER tokamak, analysis of possible schemes for integrated control in tokamaks using actuator management, as well as the development of an efficient algorithm to solve this actuator allocation problem.