Dunne L., Dunne J.

Owing to manufacturing processes, and the influence of environmental factors, real structures depart in service from nominal design conditions. Example cases where this departure can have a marked impact on the dynamic response over the entire frequency range, include the NVH response of automobile and aerospace structures, as well as the dynamics of power-train components. With particular reference to vibro-acoustic analysis, this paper discusses methods available to include the effects of model parameter uncertainty in the low frequency response analysis process. The emphasis in the paper is on the treatment of uncertain model parameters. Fast re-analysis methods are first reviewed. The accuracy of a modal correction method in which the effects of structural modifications are estimated by adjustment and re-solution of the modal equations is also reviewed in the context of uncertainty analysis. The accuracy of approximate response statistics is established for a represented vibro-acoustic structure by comparison with “exact” results obtained by Monte Carlo analysis. An approach to stochastic optimization for vibro-acoustic analysis based on a first-order estimate of variance is discussed, along with the effect of uncertainty on the threshold of predictability for low frequency analysis. An application of stochastic optimization to a representative vibro-acoustic design problem is presented, to illustrate the concept and practical importance of quality-cost relationship.