A UNIVERSAL ADJOINT-WEIGHTED ALGORITHM FOR GEOMETRIC SENSITIVITY ANALYSIS OF K-EIGENVALUE BASED ON CONTINUOUS-ENERGY MONTE CARLO METHOD

摘要

There exists a typical problem in Monte Carlo neutron transport: the effective multiplication factor sensitivity to geometric parameter. In several methods attempting to solve it, Monte Carlo adjoint-weighted theory has been proven to be quite effective. The major obstacle of adjoint-weighted theory is calculating derivative of cross section with respect to geometric parameter. In order to fix this problem, Heaviside step function and Dirac delta function are introduced to describe cross section and its derivative. This technique is crucial, and it establishes the foundation of further research. Based on above work, adjoint-weighted method is developed to solve geometric sensitivity. However, this method is limited to surfaces which are uniformly expanded or contracted with respect to its origin, such as vertical movement of plane or expansion of sphere. Rotation and translation are not allowed, while these two transformation types are more common and more important in engineering projects. In this paper, a more universal method, Cell Constraint Condition Perturbation (CCCP) method, is developed and validated. Different from traditional method, CCCP method for the first time explicitly articulates that the perturbed quantity is the parameter of spatial analytic geometry equations that used to describe surface. Thus, the CCCP can treat arbitrary one-parameter geometric perturbation of arbitrary surface as long as this surface can be described by spatial analytic geometry equation. Furthermore, CCCP can treat the perturbation of the whole cell, such as translation, rotation, expansion and constriction. Several examples are calculated to confirm the validity of CCCP method.