Actuator Management and Allocation

Mapping many competing control demands onto a limited set of shared, constrained actuators without violating hardware limits.

The shared-resource problem

Shape, position, and current controllers all demand action from the same poloidal-field coils and power supplies; heating and current drive share beam and wave sources; fueling shares gas and pellet systems. Actuator management is the layer that turns the sum of these demands into physically realizable commands, respecting current, voltage, rate, and thermal limits.

Allocation as optimization

When demands exceed capacity, something must give. Allocation is posed as a constrained optimization each cycle: find the actuator commands that best satisfy the weighted control objectives subject to hard limits. Priorities are explicit - vertical stability and machine protection outrank shape fidelity, which outranks a secondary profile goal - so the system degrades gracefully under saturation.

Handling saturation

An actuator that hits its limit can no longer help its loop, and naive controllers keep integrating error against a wall (integrator windup). Actuator management prevents this by informing the loops of remaining headroom and by anti-windup logic, so that when a coil saturates the controller does not accumulate a demand it can never deliver.

Redundancy and reconfiguration

Over-actuated systems have spare degrees of freedom that can be reallocated if an actuator fails. If a coil or supply drops out mid-discharge, allocation redistributes its role across the remaining hardware where the geometry allows, and tells the scenario layer what capability was lost so the plan can adapt.

The interface it enforces

By owning the actuators, this layer lets the higher layers speak in terms of goals rather than raw commands, and guarantees that no loop can drive the hardware past its limits. It is the translator between what control wants and what the machine can actually do.