- Home
- Uncertainty Quantification in Aeroacoustic Robust Design Applications

Analysing the effect of uncertainties on aerodynamic noise sources and mechanisms. Uncertainty is introduced to the noise source region and to the noise propagation region. The effect of geometrical and operational uncertainties on generated noise is assessed, along with simultaneous study of the impact of uncertainties on aerodynamic and aeroacoustic properties. The outcome provides a clear perspective of uncertainties during the prediction of aerodynamic noise, thus defining the required background for inclusion of acoustic performance in robust design of rotating machines.

The first major objective of the present study is to quantify the effect of uncertainty introduced during the source identification stage of the hybrid CAA approach. Uncertainty is introduced in the inputs of the unsteady CFD solution, by considering the input as non-deterministic. The variables may be geometric or operational. A non-intrusive approach is applied to this case in order to quantify the effect of non-deterministic inputs (flow inputs) on the output of the CFD solver, from which the noise sources are calculated. Aim of this study is to determine the impact of uncertainty on the noise sources, initially, and subsequently on the predicted noise, for the first time. The impact of uncertainties on tone noise and broadband noise mechanisms is investigated individually.

The second major objective of this proposal focuses on UQ in the acoustic propagation stage of the hybrid CAA approach, for a given flow field.

A UQ study in the propagation part of the hybrid CAA approach is realized, for a given flow field, calculated by LES. Uncertainty is then be introduced in the input of the acoustic analogy solver (acoustic inputs). UQ is initially carried out by a non-intrusive approach, while an intrusive approach is also implemented on the acoustic codes. The behaviour of the non-intrusive and intrusive methods in the propagation region is assessed. The effect of uncertainty on tone noise and broadband noise mechanisms is investigated individually, in the same manner as described previously.

Application of UQ and RDM in aeroacoustic problems deals with a large number of uncertainties. Therefore, a further challenge consists of tackling large number of uncertainties.

Reduced order methods is used in combination with Non-Intrusive PC (NIPC), in order to counter the curse of dimensionality. A generalized spectral decomposition method, developed at VUB, is implemented. Additionally, adaptive sampling techniques and regression is utilized to further improve its efficiency.

Modelling of geometric uncertainties is a demanding task which also deals with a large number of uncertainties. Geometrical uncertainties are modelled either by parameterizing the geometry with a finite number of variables or by approximating them by a KL decomposition. The efficiency of the two methods is compared, in order to specify the most preferable for realistic applications, since geometric uncertainties are of great interest to the industry.

The uttermost objective of this project is to investigate the effect of uncertainties on aeroacoustics, in order to realize the first step towards aeroacoustic robust design in rotating machines. Uncertainty quantification of geometrical and operational parameters on aeroacoustic properties is also accompanied by a study of aerodynamic performance, in order to provide the required theoretical background for complete robust design in rotating machines. Future goal of the research group is to couple the outcome of the present project with multi-objective optimization methodologies currently explored by VUB within the EUFORIA project SBO 140068, in order to apply robust design in realistic applications.