A Three Dimensional Heterogeneous Coarse Mesh Transport Method for Reactor Calculations

Transport theory;Coarse mesh methods;Monte Carlo method;Benchmark problems

Forget, Benoit ; Mechanical Engineering

University:Georgia Institute of Technology

Department:Mechanical Engineering

关键词: Transport theory;    Coarse mesh methods;    Monte Carlo method;    Benchmark problems;   

Others  :  https://smartech.gatech.edu/bitstream/1853/16257/1/forget_benoit_y_200608_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Current advancements in nuclear reactor core design are pushing reactor cores towards greater heterogeneity in an attempt to make nuclear power more sustainable in terms of fuel utilization and long-term disposal needs.These new designs are now being limited by the accuracy of the core simulators/methods.Increasing attention has been given to full core transport as the flux module in future core simulators.However, the current transport methods, due to their significant memory and computational time requirements, are not practical for whole core calculations. While most researchers are working on developing new acceleration and phase space parallelization techniques for the current fine mesh transport methods, this dissertation focuses on the development of a practical heterogeneous coarse mesh transport method.In this thesis, a heterogeneous coarse mesh transport method is extended from two to three dimensions in Cartesian geometry and new techniques are developed to reduce the strain on computational resources.The high efficiency of the method is achieved by decoupling the problem into a series of fixed source calculations in smaller sub-volume elements (e.g. coarse meshes). This decoupling lead to shifting the computation time to a priori calculations of response functions in unique sub-volumes in the system. Therefore, the method is well suited for large problems with repeated geometry such as those found in nuclear reactor cores.Even though the response functions can be generated with any available existing fine-mesh (deterministic or stochastic) code, a stochastic method was selected in this dissertation.Previous work in two dimensions used discrete polynomial expansions that are better suited for treating discrete variables found in pure deterministic transport methods.The amount of data needed to represent very heterogeneous problems accurately became quite large making the three dimensional extension impractical.The deterministic method was thus replaced by a stochastic response function generator making the transition to continuous variables fairly simple. This choice also improves the geometry handling capability of the coarse mesh method.