A new method of reanalysis: multi-sample compression algorithm for the elastoplastic FEM

摘要

Reanalysis is an efficient method to reduce the computational costs. For the nonlinear FEM, reanalysis based on single convergent solutions is still developing. A new reanalysis method, a multi-sample compression algorithm for elastoplastic FEM, is presented. This method consists of two strategies. First, based on the solved-sample, the approximate displacement of a solving-sample could be estimated by the method of displacement predicted, and serve as the initial value for the iterative computation. Second, the iterative stiffness of the solved-sample is applied in the solution of the solving-sample, which avoids the time-consuming decomposition of the stiffness. Examples of the thick-walled cylinder subjected to inner pressure are illustrated here. The results show that the new method is applicable when variables vary greatly, and significantly reduces the computational costs. This method is useful to obtain the convergent solutions for the elastoplastic nonlinear FEM for multi-sample conditions.