An improved approximate method for computing eigenvector derivatives

Abstract An improved approximate method for computing eigenvector derivatives has been developed. In this formulation an eigenvector derivative is assumed to be spanned by a set of truncated normal modes augmented by a residual static mode. The coefficients in the expansion are computed by a Bubnov-Galerkin method. The formulation has been implemented as a set of Direct Matrix Abstraction Programming alters for MSC/Nastran . Numerical examples show the method provides sufficient accuracy and reduced computation time when compared to the exact solution.