Component Mode Synthesis for Substructures with Non-Matching Interfaces

When performing vibration analysis of complex vehicle structures, it is often important to be able to evaluate the effects of design changes in one or more substructures (e.g., for design optimization). It may also be convenient to allow different components to be modeled independently by different groups or organizations. For both cases, it is inevitable that some substructures will have non-matching finite element meshes at the interface where they are physically connected. Thus, a key challenge is to be able to handle the dynamic assembly of components with non-matching meshes and the subsequent global vibration analysis in a systematic and efficient manner. To tackle this problem, the enhancement of component mode synthesis methods for handling finite element models partitioned into non-matching substructures is considered in this paper. Some existing methods are reviewed, and new methods are developed. Three main approaches are considered: interface interpolation, interface reduction, and component synthesis. First, among various interpolation schemes for non-matching interfaces, the moving least squares interpolation and assumed interface shapes are employed. Second, for interface reduction, the interface coordinates are projected onto a local and a global interface modal basis. Third, for component synthesis, a conventional synthesis method and a mixed formulation with Lagrange multipliers are used. Appropriate combinations of these approaches yield a suite of reduced-order vibration modeling techniques that are well suited for distributed simulation environments and/or design studies. Numerical results from the proposed methods are compared using a finite element model of a plate that is partitioned into two components with non-matching interfaces. In evaluating the proposed methods, special attention is paid to their efficiency for vibration power flow calculations under design changes and their applicability to distributed simulation.