| Summary: | BackgroundThe small transportable nuclear power system designed with the solid block heat pipe stack as the core has the advantages of good environmental adaptability, system safety and reliability, deployment flexibility, and resistance to external events. In order to control the sharp rise in reactivity under the water flooding accidents, a spectral shift absorber Gd2O3 is added to its core fuel. During the process of power flatten by adjusting fuel enrichment and Gd2O3 mass fraction at different locations in the core, it is necessary to ensure that the core solution meets a series of objectives such as refueling cycle and critical safety in both the conditions of normal operation and the accident. It makes the power flattening problem of small transportable nuclear reactor with high dimensional decision variables, multiple objectives and constraints, and multiple operating conditions.PurposeThis study aims to develop a core power flattening algorithm for small transportable nuclear power supply to reduce the computational cost.MethodsBased on a co-evolutionary algorithm framework and the laws of physics, a power flattening algorithm was developed to reduce the dimensionality through clustering algorithms. Convergence acceleration was achieved through the collaboration between sub-populations. The agent model based on Gaussian process regression (GPR) was used to predict and screen a large number of core schemes, so as to effectively reduce computational costs.ResultsComputation results show that the power peak factor of the initial scheme is decreased from original 1.30 to 1.14 after the core optimization.ConclusionsCompared with the traditional differential evolutionary algorithms and differential evolution algorithms embedded with clustering, the co-evolutionary based power flattening algorithm has significant advantages in terms of optimization quality and efficiency. The use of the clustering method and the surrogate model in the framework of co-evolutionary algorithms can effectively deal with the optimization of the complex reactor core.
|
|---|