| The development of advanced nuclear reactors has strategic significance for alleviating the energy crisis. The simulation located with Monte Carlo method as the core is obviously advantageous and better accommodating to future needs because of the extreme exploration and complexity of advanced nuclear reactors. However, for reactors simulation, fission source convergence and calculation speed are bottleneck problems of Monte Carlo development and also the international hot research topics. Expansibility of functions and structure of present Monte Carlo codes is insufficient for reactors application. Especially for advanced nuclear reactors, these codes are not easy to apply and fast develop to accommodate to new needs.Focusing on these aspects, the main research and innovation work of this thesis are as following:(1) Based on exponential iteration neutron transport equation, research on the k-effective calculation method using combined estimator of three estimators was performed. A new adaptive overlapping mesh method for Shannon Entropy diagnostic was proposed and varified through one quarter reactor core benchmark Inp24. Compared to traditional method, this method makes the mesh division relate to physics to eliminate difficulties in estimation of whether fission source is sampled efficiently. This method also can automatically locate the meshes in fissile regions to make the code more easy to use for problems of complex geometry and materials distribution. After the analysis of present acceleration methods of source convergence, a method of pre-sampling fissile regions according to the physic characteristics and then taken as the initial fission source was proposed from different perspective. This method was tested with the Inp24benchmark case and the convergence speed was enhanced by almost31.8%.(2) Cloud Computing was introduced to reactor Monte Carlo calculation. A Cloud Computing framework based on C/S framework and Socket network communication technology and collaborative parallel task schedule method based on monitor were designed. This design makes users can intelligently perform collaborative calculation of distributed task via network. A dynamic loading balance method for parallel calculation on particles was proposed. Calculations are performed in elastic Cloud Computing way according to the real situation of resources and tasks. The testing with ITER benchmarking model demonstrated that such loading balance method can enhance the calculation efficiency by42.99%and guarantee the consistency of calculation results.(3) Considering the needs and development trends of advanced nuclear reactors, a functional architecture of coupling with multi-physics and basing on advanced computer technology was designed. The structure of particles trajectory and tallies, simulation flow path of neutron and photon transport, geometry description method and structure were defined and designed to gain better expandability of code. A uniform XML standard data exchange format was defined to integrate geometry and physics automatic modeling, neutron and photon transport calculation, process and results visualization. Such integrated approach can eliminate some difficulties in describing and results analysis of complex advanced nuclear reactors. SuperMC was implemented based on such code design scheme. With SuperMC, starting from original CAD models, the first wall, divertor cassettes, inboard toroidal field coils and equatorial port of fusion reactor ITER benchmark model were analysed. The advantage of SuperMC over MCNP on complex geometry description and dealing method were demonstrated. |
PDF Full Text Request