Substructure recursive approach of eigenvalue analysis of large random structures

A novel numerical approach, which combines component modal synthesis method with recursive stochastic finite element method (RSFEM), is employed to study eigenvalue problems of large random structures. This method is called as substructure recursive stochastic finite element method (SRSFEM) in this paper. Different from traditional the first or second order of perturbation stochastic substructure methods, this approach adopts non-orthogonal polynomial expansions to represent the unknown random eigenvalues of large structures. Afterwards, perturbation technique is used to determine a series of deterministic coefficient terms of the unknown polynomial expansions. By one numerical example of random variable cross section beam, it is demonstrated that compared with perturbation stochastic substructure method based on the first or second order Taylor expansion, the results of SRSFEM match well with those from Monte-Carlo simulation method in wide range of random fluctuation.