Efficient visualization of high-resolution virtual nuclear reactor

High-resolution virtual reactors will enable scientists to simulate and analyze almost every aspect of the performance of existing nuclear reactor designs in unprecedented detail. However, high-quality visualizations of full-core nuclear reactor simulation data with high (pin-by-pin) spatial resolution require both time- and memory-consuming processing of large-scale combinatorial solid geometries with complicated constructions. This paper presents an efficient approach for performing this task using a novel solid-model-oriented rendering pipeline. The proposed method depends on the repeated structures within the reactor model, homogeneous material properties of the solid entity, and visibility of that entity in the current view, coupled to an efficient multiscale-driven visualization framework. Virtual reactor datasets are maintained as solid representations of the simulation models until the rendering stage. The proposed method can guide the production and reduction of mesh geometry directly according to statistics gathered during rendering. In combination with a level-of-detail representation of the solid geometry set, the reactor visualization can be accelerated because of the effective runtime reduction of large-scale cells. The results and performance of the method are demonstrated for the large-scale data visualization of the full-core pin-mode virtual reactor of the Chinese Daya Bay nuclear power station that comprises 757,839 solid units. The example is rendered successfully within a common user’s resource-limited computing environment, respecting the limited budget for CPU and memory.