Researchers at Framatome describe a new operator-assistance tool called the OAPS system. OAPS is a real-time predictive adviser for pressurized water reactors (PWRs). It gives operators step-by-step recommendations — for example how fast to change turbine power, how much fresh water or borated water to inject, and how to act on control rods — and it updates those recommendations as plant measurements come in.

The paper explains how the team built OAPS and tested strategies in simulators. The conventional axial offset (AO) control strategy — AO meaning the distribution of power along the reactor height — was previously validated on two full-scope simulators. In the work reported here, the authors showcase three new strategies on an intermediate-complexity PWR simulator developed by Framatome: finding the fastest feasible power-change rates, speeding up cancellation of axial power oscillations, and minimizing water and boron effluents during transients.

OAPS uses a control method called model predictive control (MPC). MPC runs a mathematical model of the plant forward in time under many possible control choices. It then solves a constrained optimization problem to pick the best sequence of actions given limits such as plant safety bounds (limiting conditions of operation) and practical limits on pumps and valves. The system continually re-plans using the latest sensor readings, so recommendations can be updated every minute or transmitted within seconds after a change in the turbine schedule.

This kind of tool matters because electricity grids are becoming harder to predict. Wind and solar power can swing quickly. Nuclear plants that once ran steadily now face more short-notice requests to reduce or raise output. Human operators currently prepare scenarios hours in advance and then must monitor and tweak them during the transient. OAPS aims to reduce that workload, shorten preparation time, and help operators carry out safe, optimal transients with fewer manual adjustments.