Finite Element Mesh Considerations for Reduced Integration Elements

Year
2007
Author(s)
Gordon S. Bjorkman, Jr. - U.S. Nuclear Regulatory Commission
Jason M. Piotter - U.S. Nuclear Regulatory Commission
File Attachment
342.pdf168.56 KB
Abstract
Finite element models of spent fuel casks and canisters that are typically used in impact and impulse analyses may contain tens of thousands of nonlinear elements. These models use explicit time integration methods with small time steps. To achieve reasonable run times, fully integrated elements are replaced with under-integrated elements that use reduced integration procedures. When fully integrated these elements produce a linear strain distribution. Reduced integration, however, results in a constant strain distribution, which requires more elements through the thickness of the canister shell to achieve the same accuracy as fully integrated elements. This paper studies the effect of the number of reduced integration elements through the thickness of the canister shell and the ratio of element height to shell thickness on the accuracy of the strains in regions of high through-thickness bending, such as the junction between the shell and base plate. It is concluded that mesh refinement has a significant effect on the maximum plastic strain response in such regions and that a converged solution may not be attainable within practical limits of mesh refinement, if the results are based solely on the maximum plastic strain on a cross section at a structural discontinuity. However, a converged solution is obtained by investigating the response of other elements on the same cross section that are not located on the surface of the stress concentration at the structural discontinuity. Based on the results, a “rule of thumb” is proposed for mesh refinement in a region of severe structural discontinuity wherein reasonably proportioned reduced integration solid elements are used and plastic strains are evaluated