Uncertainty quantification using interval modeling with performance sensitivity

In this paper an interval modeling approach for uncertainty quantification of a structure with significant parameter variation is presented. Model uncertainty can be categorized as dominant uncertainty due to structural variation, such as joint uncertainty and temperature change, and minor uncertainty associated with other factors. In this paper, a singular value decomposition (SVD) technique is used to decompose parameter variations into principal components that are weighted based on the sensitivity of the performance metric to parameter variations. From this process, parameter bounds in the form of an interval model are generated and each interval corresponds to one identified bounded uncertainty parameter with its associated principal direction. The proposed approach can be used to differentiate between dominant and minor uncertainties. A beam structure with an attached subsystem proposed by Sandia National Laboratories is used to demonstrate this approach.