Comparison of CMS, Krylov and balanced truncation based model reduction from a mechanical application engineer's perspective

Abstract

Component Mode Synthesis (CMS) is a well known and established method for order reduction of Finite Element (FE) models. One advantage of CMS is a clear physical interpretability and another, more practical one, is the availability in common FE packages. In the last years a lot of research has been done, in order to adapt reduction methods, which are based on Krylov subspaces and balanced truncation for FE models. Several recent publications denote mode based reduction methods, like CMS, as out-dated while the latter ones are so called 'modern methods'. For a mechanical application engineer the question arises, whether these methods are really so advantageous, that the reliable CMS should be exchanged against one of the two other methods. This paper is devoted to a numerical and qualitative comparison of these three methods with respect to each other. The contribution starts with an introduction, where the 'mechanical application engineer's perspective' is explained in terms of requirements and boundary conditions of the reduction process. Next, all three approaches will be briefly outlined and representative literature will be cited. In the next chapter, all three methods will be demonstrated by a simple three mass example, so that some basic characteristics can easily be seen and first conclusions can be drawn. In the subsequent section the different methods will be applied to simple FE structures and the quality of the reduced models will be examined. Finally a conclusion will be drawn whether one of the three methods is clear better (or worse) with respect to the other ones. © The Society for Experimental Mechanics, Inc. 2012.