Institute of Engineering and Computational Mechanics
Model reduction in flexible multibody dynamics
One important step in the simulation process of flexible multibody systems is the linear model reduction of the elastic degrees of freedom. Thereby, the dynamical behaviour of the original system is approximated by a model of lower dimension. In the past years many modern reduction methods apart of the modal reduction were developed in applied mathematics. The application, development and implementation of structure preserving model order reduction methods in elastic multibody systems are the goals of the research work.
The reduction of the equations of motion is based on linear model order reduction via projection. Therefore, the possible dynamical behavior is described in a subspace of significantly smaller dimension. The reduction methods which are developed in applied mathematics approximate the input-output behavior of the elastic system by generating an error as small as possible. One major requirement for model order reduction of mechanical systems is the preservation of the second order structure because the reduced elastic body is plugged into an elastic multibody system after its reduction.
The most often applied methods for the reduction of mechanical systems are modal methods, e.g. the Craig-Bampton method from the field of Component-Mode-Synthesis. The quality of the reduced models depends on the experience in selecting the right modes. The difference between important and unimportant modes in complex structures is not detected very easily and needs an experienced user. Alternatively, modern reduction techniques are used and classified in two categories: The reduction based on Krylov subspaces and the SVD or Gramian matrix based reduction methods. These new reduction techniques and the modal reductions are implemented in the preprocessor MOREMBS which is developed at the Institute of Engineering and Computational Mechanics.