The development of accurate and efficient numerical models for jointed structures is an important and challenging problem. Due to nonlinearities in the joints, notably dry friction, contact stiffness, and impact, joint models are often complicated and computer-intensive. To create practical models, engineers typically combine “lumped joint models” with reduced-order models for the structural members that they connect. However, the model reduction often distorts how the joint behaves and sometimes destroys important qualitative traits. Simple modal truncation is often inadequate to produce reduced-order models because nonlinearities within the joints such as impact and dry friction can depend critically on the high-frequency characteristics of the mating structural elements. This paper examines the issues surrounding the development of accurate, reduced-order models for nonlinear, jointed structures. The concept of “balanced realizations” from control theory are used to create reduced-order models that best capture the input-output characterization of the linear substructures with the smallest model order. The balanced-realizations are seen to produce very favorable results when compared with standard modal truncation techniques.

This content is only available via PDF.
You do not currently have access to this content.