The Research Of Reducing Rod Cusping Effect

Modeling a control rod insertion is one of the most difficult part in the Nodal Methodespecially when the position of the rod tips does not match node boundaries. The nodalequations consider uniform cross sections for the whole node and each node has only one flux,how to homogenize the control rod contribution inside the node may greatly affect the result.With a simple volume weighting method the keffbehavior versus insertion showsdiscontinuities. This numerical effect is known as rod cusping effect.For a better research of the rod cusping effect, a Nodal Green’s Function Method(NGFM) was first developed. After applying discontinuity factor to this method andconsidering all the influencing factors when changing NGFM to second boundary condition，this program may have the same accuracy with Simulate3with which the difference of powerdistribution within1%and the difference of keffwithin1‰. Besides, a new model was addedinto this program to get the adjoint neutron flux which can be used to reduce the rod cuspingeffect, or in the perturbation theory to calculate the control rod contribution directly.This modified program was then used as a tool to analysis the rod cusping effect. Thevolume-approximate flux weighting method, the volime-approximate flux-importanceweighting method and the adaptive mesh method were used to reduce the rod cusping effect.All the methods have been tested in several configurations and each set of results showed thatthe rod cusping was reduced. As far as the computer run time and stability is concerned, thevolume-approximate flux weighting method was implemented in the cross section treatmentmodule of NGFM.At last, considering that if the control rod insertion can be accurately described, the rodcusping effect would be ultimately eliminated. The Finite Element Method (FEM) possessesflexible meshing capabilities that have a potential to accurately describe the whole core. Inthis paper, a theory for solving the two-group neutron diffusion equation has been developedusing FEM. Based on this theory a calculation program is compiled in FORTRAN.Comparison the present code with CITATION and theoretical solutions were made againstboth2-dimensional and3-dimensional problems. Results showed that the present code has agood precision.