Reduced-Order-Model Interpolation for use in Global Modal Parameterization

Interpolation of reduced order models (ROM) is an important topic in many areas of model reduction. Recently, a system-level model reduction technique for flexible multibody systems, Global Modal Parameterization (GMP), has been proposed. This method is based on an interpolation of ROMs for different undeformed configurations in order to reduce model equation assembly time. As is shown in this paper, the way this interpolation is performed significantly impacts the accuracy of the simulation. This work compares 3 different interpolation strategies. Firstly two strategies which directly interpolate the reduced mass and stiffness matrix, through linear and quadratic interpolation, are considered. A third novel interpolation strategy, which operates in the eigenspace of the ROMs, is proposed in order to obtain better accuracy for the eigenfrequencies of the interpolated system. Finally the three approaches are compared through a numerical validation on a planar slider-crank mechanism.status: publishe