Detailed Fission Reactor Neutronics Modeling Method Based On Parameter Visualization

The large-scale of geometries and the complicated hierarchical relations of fission reactor became the big obstacle that made Monte Carlor (MC) modeling by manual time-consuming and error-prone. Meanwhile, the large-scale of geometries beyond the capacity of traditional CAD-based modeling method because of the high costing of hardware resources, long conversion time and the high-redundancy of generated models. In this study, a new parameterized-based modeling method was introduced, which could generated high-fidelity full reactor models easily and got the higher calculation efficiency based on the visualization. This study based on the framework of SuperMC and the main research works of this thesis are as following:(1) A new hierarchical modeling method based on parameter visualization was studied. The approach could create or gather the geometries and physics parameters of fission reactor as well as the relations of different components of the models into a dedicated data structure, which called as Geometric Hierarchy Tree (GH-tree). Furthermore, the proposed method that was more intuitive and convenient obtained significant hardware reduction in the process of creating detailed fission reactor models and parapering the primary data for generating the high-efficiency MC calculation models.(2) A rapid conversion method based on the self-optimizing GH-tree was introduced. The method could optimize the GH-tree by the judgement of optimality, which could ensure the data redundancy and the reasonable structure of GH-tree. Furthermore, based on the meta-geometries and relations in GH-tree, a new BREP and CSG conversion method was studied. The new method only converted the meta-geometries while other geometries could generate by the corresponding relation in batches. The method was verified by the BN600 model proved that its conversion speed was promoted enormously and the generated models got the faster calculation speed than the conventional method.(3) The multi-level visualization was supported by specifying corresponding segments, which were subdivided by three conditions. The wanted models could be filtered by the properties among levels, components and materials, then the fluency of interaction was improved enormously and the real-time visualization was supported. Furtermore, a rapid two-dimension (2D) visualization method was improved by take the advantage of Gh-tree, which was applied for accelerating the algorithm.The efficiency and accuracy of methods were proved by constructing the BN600 models, the Hoogenboom benchmark, the detailed PWR BEAVRS model and the conceptual reactor models of the Accelerator Driven subcritical System (ADS). These new methods were greatly shorten the modeling period and had much better user interactivity compared with the manual texted-based modeling method and traditional CAD-based modeling method. Moreover, the accuracy of the new methodswas proved by the numerical calculation results by comparing the generated SuperMC models and MCNP reference models.