登入選單
返回Google圖書搜尋
出版Canada Centre for Mineral and Energy Technology, Minerals Research Program, Mining Research Laboratories, Energy, Mines and Resources Canada, 1979
ISBN06601052339780660105239
URLhttp://books.google.com.hk/books?id=8jRRAQAAIAAJ&hl=&source=gbs_api
註釋Stress corrosion cracking can potentially cause fuel failures at circumferential ridges in the sheath. finite element computer codes elestres and feast are used to calculate two-dimensional, axisymmetric, elastic-plastic sheath stresses at and near the circumferential ridges. the level of stresses is given by the fuel design (such as pellet/sheath diameters, pellet length, pellet density) and by the operating conditions (such as power history). the accuracy, convergence, and cost of the stress calculations, however, are determined by the discretization used in the calculations: the number of nodes in the finite element mesh; the shape of the individual finite elements; the pattern into which the finite elements are arranged; and the size of the calculation-steps into which the power history is subdivided. in this study, we determined the discretization that gives a reasonable compromise of accuracy, convergence, and cost for the stress calculations. the highest accuracy is obtained by using triangular finite elements of aspect ratios close to one, and assembling them into hexagonal patterns. to caputure the stress gradients, five nodes are required across the sheath wall at the ridge, but two nodes are sufficient near the midplane of the pellet. the resulting mesh contains about 200 nodes and about 300 finite elements. calculation-steps of 6 kw/m provide a convergent solution, with a factor of two margin against numerical divergence. with the above arrangement, feast needs about 2.4 seconds on a cdc/cyber 175 computer to calculate sheath stresses per calculation-step. six radial nodes at the ridge would double the computing cost. decreasing the size of the calculation-steps means more calculations are needed to analyze a given power history, giving a linear increase in computing cost.