The value of nuclear power plants’ flexibility: A multistage stochastic dynamic programming approach

In future power systems dominated by intermittent renewable generation, nuclear power plants will increasingly need to follow uncertain loads. However, regulatory and technical constraints limit the frequency of load-following operations, making their efficient allocation crucial. This paper explores the economic value of nuclear flexibility by modeling it as a stock constraint within a stochastic dynamic programming framework. We show how non-convex constraints at the reactor-scale translate to the fleet level through a linearized approximation, allowing for the analysis of multiple reactors operating under flexibility constraints. Applying this model to the French electricity system in 2035 using the Stochastic Dual Dynamic Programming (SDDP) algorithm, we estimate a marginal value of EUR 100/MW for the current flexibility level of the French nuclear fleet. Our results show that increased nuclear flexibility enhances system-wide cost efficiency and improves the integration of renewables, with solar generation seeing the largest benefits. However, our findings suggest a potential misalignment with the profit-maximizing goals of individual nuclear operators, which may deter them from increasing flexibility despite its necessity for the system.