Aircraft Multidisciplinary Design Optimization Under Both Model and Design Variables Uncertainty

Abstract

Low-fidelity analytical models are often used at the conceptual aircraft design stage. Because of uncertainties on these models and their corresponding input variables, deterministic optimization may achieve under-design or over-design. Therefore it is important to already consider these uncertainties at the conceptual design stage in order to avoid inefficient design and then costly time over runs due to re-design. This paper presents a procedure for reliable and robust optimization of an aircraft at the conceptual design phase. Uncertainties on model and design variables are taken into account in a probabilistic setting. More precisely, at each point of the optimization process uncertainties are modeled by an adaptive normal law strategy in order to fit the historical aircraft database. The statistical parameters are adjusted depending on the available information at the current point of the optimization process. To improve computational cost, response surface approximations are constructed to represent reliability constraints. The developed methodology is applied to the conceptual design of a short range aircraft. Compared to standard deterministic optimization without design margins, the result shows a modest increase on weight, which allows however to ensure a desired reliability and robustness of the design compared to the unreliable and sensitive deterministic optimum.