Parametric study on state-of-the-art analytical models for fan broadband interaction noise predictions. (8th December 2021)

• Record Type: Journal Article
• Authors: Lewis, Danny; de Laborderie, Jérôme; Sanjosé, Marlène; Moreau, Stéphane; Jacob, Marc C.; Masson, Vianney
• Is Part Of: Journal of sound and vibration. Volume 514(2022)
• Publication Date: 2021-12-08
• Subjects: Aeroacoustics -- Turbomachinery -- Rotor–stator interaction -- Broadband noise -- Fan noise -- Analytical modeling; Sound -- Periodicals; Vibration -- Periodicals; Son -- Périodiques; Vibration -- Périodiques; Sound; Vibration; Periodicals; Electronic journals; 620.205
• DOI: 10.1016/j.jsv.2021.116423
• Languages: English
• ISSNs: 0022-460X

Abstract

Abstract: Rotor–Stator Interaction (RSI) noise predictions using analytical models have become a first hand approach to perform quick and efficient noise assessments. However, a large range of models using different assumptions is currently available and it is of paramount interest to investigate what their impact on the predictions are and how representative of the real configuration they are. The present work proposes to study the impact of such assumptions on predictions using analytical models informed with RANS flow simulations. It focuses on four models which represent the state of the art of RSI noise prediction for fan-Outlet Guide Vane (OGV) stages: Ventres' model, Hanson's model, Posson's model and Masson's model. These models are tested on the NASA Source Diagnostic Test (SDT) baseline case at approach condition. A sensitivity study is carried out to identify and explain the discrepancies introduced by the assumptions made in the models. It is shown that the geometry definition (staggered flat plate), the definition of the impinging flow (2D or 3D, modeling of turbulence), the computation of the acoustic sources (2D or 3D response, isolated or cascade response), as well as the sound radiation method (free-field or in-duct propagation with mean axial or swirling flow) have a substantial impact on the predictions. The impact of the differences in the input quantities extracted from two different RANS simulations is also assessed, showing an important effect of background and wake turbulence intensities as well as of the turbulence integral length scale. The present work also demonstrates that the use of anisotropic turbulence models, which may become more common in the future due to the evolution of engine architectures, must be done carefully because of the sensitivity of the models to the anisotropic parameters which are difficult to assess from RANS simulations.