نوع مقاله : مقاله پژوهشی
نویسندگان
گروه مهندسی هستهای، دانشکده مهندسی انرژی، دانشگاه صنعتی شریف، صندوق پستی: 8639-11365، تهران - ایران
کلیدواژهها
عنوان مقاله English
نویسندگان English
When performing reactor core neutron calculations, the main cost lies in cell calculations to generate macroscopic cross-sections of fuel assemblies (FAs). The development and utilization of a cross-section library based on the reactor type is the primary procedure to reduce calculation costs. In some cases, when there is no library available for specific analyses, it becomes necessary to conduct cell calculations. Examples of such cases include investigating FAs vibration neutron noise or coolant asymmetric distribution around FAs due to bowing. In this study, an Artificial Neural Network (ANN) was used instead of high-volume cell calculations to significantly reduce the time required for calculations in the VVER-1000 reactor. For criticality calculations at Beginning of Cycle (BOC), the calculation time was reduced from 42 minutes using 32 computational cores to just 1.3 minutes. Additionally, the results achieved were of acceptable accuracy, with the difference in the critical boric acid value obtained using ANN compared to cell calculation as a reference being less than 1%, and the power difference being less than 5%.
کلیدواژهها English
