Comparative analysis of model-reduction methods

Abstract

The paper describes model-reduction methods, including static model reduction, dynamic model reduction and modal model reduction. The theoretical analysis is provided for each model reduction methods. An example is used to analyse and compare reduction accuracy of these model reduction methods. The accuracy of a model-reduction criterion is given. Model frequency, mode shape, MAC values and dynamic response should be considered in the accuracy analysis of a reduced model. The static model reduction is limited in application and causes huge error in a small master degree-of-freedom case. Dynamic reduction significantly improves the reduction accuracy compared with static reduction. Iterative dynamic reduction greatly improves the accuracy based on dynamic reduction. Modal model reduction severely restricts node choice, and is very limited in application. The paper discusses the choice of master degree-of-freedom and its influence on the accuracy of a reduced model.