Structural dynamics modification, optimization, and substructure separation

Complicated structures, such as aircraft and automobiles, exhibit complex dynamic behavior which can now be measured and modeled. These models are typically second order modal representations. The initial models can be generated from either advanced concept analytical methods such as finite element analysis, or prototype tests using experimental modal analysis. This thesis develops methods which use these models to improve structural design by optimizing the dynamics of the structure. The methods include (1) a general treatment of first order structural dynamics modification for evaluating the effects of design modifications, (2) sensitivity analysis to predict the effects of design modification, and (3) optimization algorithms to guide the selection of feasible design modifications. A performance index consisting of a weighted sum of multiple output and multiple input frequency response functions is used to numerically evaluate dynamic performance of a prescribed design. The formulation includes fixed and variable frequency sensitivities to accommodate a broad range of response models. Examples are provided which (1) verify the algorithms, and (2) illustrate the potential of the techniques in solving difficult vibration problems in complicated structures. Included, is a study of the effects of body mount stiffness and damping rates on light truck ride quality. In addition, this thesis examines the concept of scaling an experimental model directly to first order form. This would overcome the inaccuracies of the second order form that has traditionally been the basis of experimentally obtained modal models. The second order model has been shown to be inaccurate in complicated structures which exhibit high non-proportional damping levels. Finally, this thesis examines substructure separation, a new method of system identification. The procedure requires an initial modal model of a complete assembly. Performance indices are developed to measure the effects of discrete elements on the dynamic modal coupling of major substructures. Optimization is then used to identify the stiffness and damping of the coupling elements which must be removed to dynamically separate the initial modal model of the complete assembly into the independent modal models of the component substructures