Algebraic substructuring based eigenvalue reanalysis method for large-scale finite element models

Abstract

In this paper, we propose an efficient eigenpair reanalysis method capable to solve large-scale finite element (FE) models. By applying the algebraic substructuring algorithm to the Craig-Bampton (CB) method, the resulting formulation of the reduced matrices is provided in a form of very tiny submatrix computation. And, to control the reduced matrices to more compact size while maintaining accuracy, the interface subspace reduction is applied subsequently. An automatic updating strategy is proposed, and it can give a recognition of which submatrices are revised in the new FE model. Thus, by using this algorithm, a reduced model corresponding to the new FE model is constructed very effectively. This is the core idea of the proposed method that can efficiently compute new eigenpairs, with small computation time. Several real engineering problems are considered to verify the performance of the proposed method in terms of computation time and computer memory.