Multi-objective pressurized water nuclear reactor operation optimization in the context of the French energy transition using evolutionary algorithms

In the context of the introduction of renewable energies in France, Nuclear Power Plant Operations are a key component for the compensation of the intermittent production of solar and wind power. In this work, we focus on the optimization of the operation cost and stability of power of a real-life power transient, while maintaining safety standards on a 1300 MW Pressurized Water Reactor, by changing the control rods management parameters. We first develop a model of the reactor targeted for load follow operations. We then propose a method based on fitness landscape analysis in order to determine the objectives that are best suited for the problem, adapt the search space and tune the optimization algorithm. We first use that method to minimize the volume of effluents and xenon oscillations at the beginning of the reactor exploitation cycle, with an asynchronous parallel algorithm based on MOEA/D for a classic day-night power transient. We then expand that study to the complete exploitation cycle, in order to find rod parameters that are efficient for the complete reactor cycle. We develop for that study an asynchronous parallel algorithm based on MOEA/D assisted by a surrogate model. The analysis of the solutions shows the need to increase the maneuvering band of the temperature regulation rods, while adjusting the overlaps of the power shimming rods. (author)