Reallocation of System Mass and Stiffness for Achieving Target Specifications using a Superelement/Substructuring Methodology - 2003 International Journal of Vehicle Noise and Vibration
P.Avitabile, H.Tsuji, J.O’Callahan, DeClerck, J.P.
The development of a system model involves the assemblage of many important components from a dynamic standpoint. The performance of the system is dependent on the dynamic characteristics of each of the components. In an attempt to improve the overall system characteristics, targets are typically specified at the system level. The determination of the component characteristics (derived from superelement formulations) to achieve these targets is the purpose of this work.
Using these targets, an inverse optimization process can be utilized to determine the adjustment of the mass and stiffness of the system. However, the inverse process will smear the mass and stiffness over all the system degrees of freedom unless all the system equations are utilized in the inverse process or unless some constraints are applied to the set of equations. This smearing tends to confuse the proper mass and stiffness that are needed to achieve the design targets to meet the desired component dynamic characteristics.
Previous work utilized Analytical Model Improvement techniques along with Localization of Model Change procedures to determine the component target characteristics but did not address component superelement descriptions. As an extension of the procedure, the existing topology of the finite element model can be modified to restrict or allow additional elemental connectivities in the optimization process.
The efforts in this work are directed towards the implementation of the technique using superelements as the component description. The topology of the superelement is mapped with an arbitrary array of simplistic elements to determine the distribution of mass and stiffness necessary to achieve the performance goals specified without allowing smearing to adjacent component interconnection regions.
A PDF file will be available for downloading shortly
In the meantime, e-mail Peter_Avitabile@uml.edu for copy of paper