Modelling Flexible Nuclear Generation in Low-Carbon Power Systems: A Stochastic Dual Dynamic Programming approach

This paper presents a novel approach for modeling nuclear flexibility in a stochastic environment. The optimization of load-following operations, considered a fixed stock, is addressed using a multi-stage stochastic dynamic programming framework and resolved by applying the Stochastic Dual Dynamic Programming (SDDP) algorithm to address computational challenges. This approach enables the incorporation of uncertainty in renewable energy generation and facilitates extensive testing through multiple simulations. As an application, we analyze the provision of flexible nuclear generation in the French energy system by the year 2035. The findings demonstrate that nuclear flexibility can be increased while complying with international safety standards, resulting in a substantial reduction in energy curtailment from renewable sources and overall system costs. More specifically, solar panels exhibit the most significant benefits from enhanced nuclear flexibility, with nearly doubled payoffs observed in the most extreme scenario compared to the benchmark case. Furthermore, the study reveals that nuclear power plant financial profits plateau for flexibility levels ranging between current practices and twice that value, thereby underscoring the potential for operating them in a load-following mode more frequently. Overall, this research sheds light on the importance and feasibility of optimizing nuclear flexibility in evolving power systems characterized by increased renewable energy integration.