Eliminating boundary nodes in substructuring of large structures

Substructuring techniques have been widely used in model reduction of large structures. In these methods a large structure is partitioned into several components and reduced components are built. Boundary degrees-of-freedom (DoF) at the interfaces between components are used to assemble the reduced components and to form a reduced model of the original structure. In the current substructuring methods the boundary DoF or a transformation of these DoF remain in the reduced model. In this paper a methodology is suggested that could eliminate the boundary DoF from the reduced model which in turn leads to having even a smaller reduced model. This method which uses a different partitioning of the DoF of the structure is illustrated for a two-component structure. An example on a simple structure shows how the method can be implemented. The results show that the same level of accuracy compared to a standard substructuring can be obtained with fewer number of DoF in the reduced model.