| It's well known that the current generation core physics analysis method was established at the end of 1970s. Due to the hardware computing capabilities of that time, the current method with deficiency in the theory is hard to handle the increasingly aggressive core design. Consequently, the Next Generation Method (NGM, the 3rd generation) for numerical analysis of LWR core is under extensive development worldwide. The approach to NGM generally adopted is to completely abandon the currently used methods, which are based on transport calculation, assembly homogenization and 3D nodal diffusion calculation, and replace it with full core 3D Pin-by-Pin calculation. Therefore, the resources invested and the experience accumulated over the years in the methods and computational codes for LWR cores physics analysis would be abandoned as well. For this reason, an innovative idea was proposed by Reactor Physics Group of Shanghai Jiao Tong University, whose objective is to meet the future requirements by improving current methods and considering more practical engineering application issues. Our approach is to embed advanced assembly homogenization method via 2D transport calculation inside the 3D nodal diffusion core simulator. The major works in this paper focuse on the development of an MOC code and the investigation on advanced homogenization method.Firstly, an MOC transport solver named PEACH as the basic tool for NGM is developed in this thesis. It employs assembly-based modular ray tracing (AMRT) and several efficient acceleration methods. Besides owning the ability of handling quite arbitrary geometry in the assembly, the AMRT method is very efficient and memory saving. Flat source based step characteristics (SC) scheme, diamond difference (DD) scheme and linear source (LS) approximation scheme are respectively used in PEACH as independent physical computational model. The challenges regarding negative angular flux and negative source caused by DD and LS are also fixed separately. The numerical results of OECD/NEA C5G7-MOX 2D benchmark demonstrate that DD and LS are quite accurate compared with SC under the same large mesh divisions. In order to meet the speed requirement, multi-group coarse mesh finite difference (CMFD), two-level CMFD and the interpolation table of exponential function are employed as major acceleration methods, whose acceleration effects are proved to be very good as well. Moreover, parallel computation is designed for PEACH based on Massage Passing Interface (MPI) and the obtained parallel efficiency is quite high.Secondly, the albedo-based re-homogenization method, improved conventional homogenization method and colorset homogenization method are investigated respectively based on the fundamental theory of Equivalent homogenization method in this thesis. According to our original idea of NGM, the values of albedo can only be computed from last core diffusion calculation. Numerical results are shown that the accurate equivalent homogenization parameters can't be gained based on albedo, since these values are anglular independent. After appropriate spectrum leakage correction, the accuracy of the colorset homogenization method is comparable to that of full core heterogeneous transport method, even if 2G energy structure is chosen.Finally, the NGM testing system involving MOC solver, nodal diffusion solver, pin power reconstruction module, depletion module etc. is also established in this thesis. Under considering more practical engineering application, our NGM code testing system employing a full set of methods proposed in this thesis is precise and efficient after verifying several benchmark problems. The issues of efficient transport tool and advanced homogenization methods are the key problems to our NGM. They have significant academic value and application value as well.
|
PDF Full Text Request