Actuator Sharing and Allocation
Many control goals compete for the same limited actuators; allocation decides who gets them, when, and how much.
The shared-actuator problem
A fusion device has more control objectives than actuators. Electron cyclotron power might be wanted for sawtooth control, tearing-mode stabilization, current-profile shaping, and heating all at once. These requests cannot all be met simultaneously, so the control system needs an allocation layer that arbitrates among them.
Priorities and pre-emption
Allocation follows a priority order. Machine protection outranks performance; catching a growing tearing mode outranks trimming the current profile. When a high-priority need appears, the allocator can pre-empt a lower-priority use, redirecting a beam or a mirror to where it is most needed. This is decided within the control cycle, not by an operator.
Constraints on allocation
- Total power and current limits per actuator
- Physical aiming ranges and slew rates of steering mirrors
- Minimum on-times and duty-cycle limits
- Coupling, where one actuator affects several controlled quantities
Mathematical framing
Allocation is often posed as a constrained optimization solved each cycle: given the demands from each controller and the actuator limits, find the actuator settings that best satisfy the highest-priority demands. Because it runs in real time, the solver must be fast and bounded, so simple quadratic programs or rule-based schemes are common.
In the Kronos program
The Hyperion breeder shares its electron cyclotron and beam systems among sawtooth, tearing-mode, current-profile, and heating functions, so its supervisor includes an allocator that reassigns them as the discharge phase changes. The burner generators have a different actuator set built around the mirror plug fields, but the same allocation principle applies. Allocation policies are tested in the flight simulator.