Guillaume DUPRÉ Thesis Defense

Abstract : Most renewable electricity generation units deployed in recent years are inherently intermittent. In the absence of large-scale storage solutions, electricity production and consumption must be constantly balanced to ensure grid stability. This role, traditionally held by fossil fuel-based power plants, is increasingly being assumed by nuclear power plants. Thus, this thesis aims to enhance the flexibility of pressurized water nuclear reactors to meet future electrical grid needs. To achieve this, several reactor core control systems have been designed based on advanced methods in the field of control systems, namely predictive control and sequential gain control. A nonlinear multi-loop reactor model, intended for control law synthesis, had to be developed. Of just sufficient complexity, it is well-suited for short-term prediction purposes. The proposed solution comprises two components: 1) a real-time piloting assistance system (world patent) now part of Framatome's commercial offerings, and 2) a hierarchical piloting solution compatible with existing temperature regulation loops, significantly enhancing performance in terms of flexibility and adherence to operational constraints compared to current piloting methods. This solution relies on the most suitable techniques for nonlinear predictive control implementation.