Uncertainty Quantification for Robust Design

Simulation-based performance prediction has been an active research topic for many years and is now applied for industrial test-cases in all fields of computational mechanics. However, classical methods developed in Computational Fluid Dynamics (CFD) or Computational Structural Mechanics (CSM) usually assume a perfect knowledge of the parameters of the system studied, such as geometry, operational conditions, etc. Under this asumption, efficient numerical methods have been developed, yielding an accurate performance prediction for optimization purpose. However, everyday life is subject to uncertainty and the parameters of every systems are subject to random fluctuations. For instance, the flight conditions of an aircraft can fluctuate according to the weather, the wing design can vary because of manufacturing tolerances, etc. These uncertainties modify the response of the system and in some cases significantly degrade its performance. This is particularly true if the system has been optimized for some precise operational conditions.