Investigation On GPU-Parallel Accelerated Neutron Transport Method Of Characteristics For Hexagonal Cores And Multiple Nulcides Resonance Interference Effect

For designing or analyzing cores with rectangle assemblies,there exist many code systems with high accuracy and complete functions. However,for cores with hexagonal assemblies,indigeneously developed codes are still much in need. The first part of this thesis focuses on developing a whole core transport calculation computing code for hexagonal assemblies and investigating acceleration means via both mathematical methods and techniques of using hardware devices. Core physics calculation needs the support from nuclear data library, but the treatment for the resonance nuclear data has been a difficult problem. The second part of the thesis investigates a convenient method to treat the resonance data with high accuracy and computing efficiency.To develop a neutron transport code for cores with hexagonal assemblies, this thesis chooses the Method of Characteristics(MOC) because of its high accuracy and good geometry capability. In our MOC code, Modular Ray Tracing is used to trace the tracks passing through cores. Modular Ray Tracing can accommodate the generic geometry inside an assembly. It saves memory and the geometry processing time when applied to a core with assemblies of the same shape and size, thus enhancing the ability of modeling large scaled cores. In this thesis, linear source approximation is used when tracing the characteristic tracks as it has been proved to be more accurate in previous studies. To verify the accuracy and capability of our code,a hexagonal mini-core benchmark problem is established with its reference solution given. The benchmark verification shows that the accuracy of our code is very satisfying.As neutron transport calculation is very demanding and time consuming, Coarse Mesh Finite Difference(CMFD) has been widely adopted as an effective way to accelerate the source iteration in transport calculation. However in a core with hexagonal assemblies there are non-hexagonal meshes around the edges of assemblies,causing a problem for CMFD if the CMFD equations are still to be solved via tri-diagonal matrix inversion by simply scanning the whole core meshes in different directions. To solve this problem, we propose an unequal mesh CMFD formulation that combines the non-hexagonal cells on the boundary of neighboring assemblies into non-regular hexagonal cells.We also investigated the alternative hardware acceleration of using graphics processing units(GPU) with graphics card in a personal computer. The tool CUDA is employed, which is a parallel computing platform and programming model invented by the company NVIDIA for harnessing the power of GPU. The hexagonal mini-core benchmark problem is used again to assess the effectiveness of CMFD and GPU parallel acceleration. Both CMFD and GPU acceleration give good speedup performance individually. And when two acceleration methods are used simultaneously,they provide a speed gain of hundreds of times.An improvement for the application of Dancoff factor is developed that combines Stamm’ler’s two-term method for resonance integral calculation with the neutron current method for Dancoff factor calculation. Stamm’ler’s formulation, which is originally derived for the infinite lattice geometry, can be easily revised to contain the Dancoff factor explicitly. While the neutron current method can easily calculate the Dancoff factor for the general irregular assembly geometry. For the resonance interference effects, the resonance interference factor(RIF) table is built in pairs of nuclides, only for the interference between U-238 and other resonance nuclides, spanning over a range of background cross-section and number density ratio of the pairing nuclides. A series of verification calculations have been carried out for problems of infinite lattice and single assembly geometry, with two or multiple resonance absorbers. For these verification calculations, the proposed combination of Dancoff factor with the Stamm’ler formulation plus the use of RIF for resonance interference correction give very good results, with quite acceptable errors for practical applications.