Prediction and uncertainty propagation of correlated time-varying quantities using surrogate models

Abstract

The identification of correlated quantities is of particular interest in several fields of engineering and physics, for example in the development of reliable structural designs. When ‘time-varying’ quantities are analysed, pairs of correlated interesting quantities (IQs), e.g. bending moments, torques, etc., can be displayed by plotting them against each other, and the critical conditions determined by the extreme values of the envelope (convex hull). In this paper, a reduced order singular value-based modelling technique is developed that enables a fast computation of the correlated loads envelope for systems where the effect of variation of design parameters needs to be considered. The approach is extended to efficiently quantify the effects of uncertainty in the system parameters. The effectiveness of the method is demonstrated by consideration of the gust loads occurring from the aeroelastic numerical model of a civil jet airliner.