Managing Margins Under Uncertainties Surrogate Modelling and Uncertainty Quantification

Abstract

Engineers account for uncertainties by using Design Margins, typically by introducing them implicitly. Although methods for quantifying uncertainty are well established at discipline level, they are not applied systematically. A hierarchy of assumed independent design margins makes it difficult to deduce architect level margins. Further, the implicit application of margins and reserve factors also obfuscates the mutualisation of the margins. Therefore, uncertainty aggregation is difficult, in the quantitative sense, leading to potential over design. Quantifying these uncertainties, making them explicit and aggregating them correctly enable the discovery of appropriate margins to manage the identified risks and opportunities.

We show how uncertainties, arising at different levels in a design process, may be aggregated to provide variations in the performance metrics of a complex system. The approach is illustrated through a case study based on a notional aircraft; the performances metric chosen for this example is the cruise fuel consumption.