Skip to Main content Skip to Navigation

Robust design of acoustic treatment for nacelle noise reduction using computational aeroacoustics and uncertainty quantification

Abstract : In modern turbofan engines, fan noise is one of the main noise sources due to the constant increasing of engines bypass ratio for fuel burn reduction purposes. As fan noise is characterized by broadband and tonal components, acoustic liners are introduced for their effectiveness in mitigating both components through dissipation effects that are tunable by modifying the liner geometry. Simulations issued from prediction numerical tools are thus extensively used for their tuning, since experiments cannot be considered for obvious costs reasons. As the design of liner systems is frozen in early stages of an aircraft development, it exists a non-negligible variability on its operating environment. This variability directly impacts the design of liners by inducing large discrepencies on the quantities used for its numerical design. Moreover, each time these quantities are updated due to the increased maturity of the aircraft program, lined surfaces are to be reoptimized. The updating phase thus represents important costs in terms of computational time, which could be avoided by accounting for such a variability in preliminary phases of the liner design. This is the main problematic of the present work. As this external variability directly impacts the liner environment, the computational modeling of the liner acoustic performance is uncertain. In order to quantify and account for such an uncertain nature, a robust design of the liner is carried out, by quantifying the overall uncertainty that lies within the liner design computational process. The state-of-the-art computational aeroacoustic model of nacelle liners performance is an industrial numerical code, Actran/TM, which therefore has to be extensively studied so as to exhibit the principle components that are subject to the overall uncertainty. A stochastic modeling of uncertainties is then introduced and grafted on the computational model. It allows for simulating the previously mentioned external variability, by accounting for the uncertainty that lies within the model (modeling errors and model parameters errors), through parametric and nonparametric probabilistic approaches of uncertainties. Then, the propagation of uncertainties in the system is analyzed using the computational model and the Monte Carlo stochastic solver. The acoustic response is then random and the quantification of uncertainties consists in estimating statistics, such as confidence regions associated with a certain confidence level of quantities of interest. From these statistical information, the robustness of a given liner design towards a simulated variability on its performance model can be defined, in addition to the state-of-the-art liner acoustic performance. This information then allows for knowing the propensity of a given liner design to maintain its nominal performance when its environment is changing in a predefined range of variation accounted for by the level of uncertainty imposed on the stochastic model. Then, making a compromise between performance and robustness, the best liner design can be chosen
Complete list of metadatas
Contributor : Abes Star :  Contact
Submitted on : Tuesday, January 19, 2021 - 7:36:08 PM
Last modification on : Friday, February 5, 2021 - 3:32:04 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03115688, version 1



Vincent Dangla. Robust design of acoustic treatment for nacelle noise reduction using computational aeroacoustics and uncertainty quantification. Mechanical engineering [physics.class-ph]. Université Paris-Est, 2020. English. ⟨NNT : 2020PESC2024⟩. ⟨tel-03115688⟩



Record views


Files downloads