Model order reduction based on successively local linearizations for flexible multibody dynamics

An efficient method of model order reduction is proposed for the dynamic computation of a flexible multibody system undergoing both large overall motions and large deformations. The system is initially modeled by using the nonlinear finite elements of absolute nodal coordinate formulation and then locally linearized at a series of quasi-static equilibrium configurations according to the given accuracy in dynamic computation. By using the Craig-Bampton method, the reduced model is established by projecting the incremental displacements of the locally linearized system onto a set of local modal bases at the quasi-static equilibrium configuration accordingly. Afterwards, the initial conditions for the dynamic computation for the reduced model via the generalized-alpha integrator can be determined from the modal bases. The analysis of computation complexity is also performed. Hence, the proposed method gives time-varying and dimension-varying modal bases to elaborate the efficient model reduction. Finally, three examples are presented to validate the accuracy and efficiency of the proposed method.