Periodic Reporting for period 2 - SCONE (Simulations of CrOr and fan broadband NoisE with reduced order modelling)
Reporting period: 2019-05-01 to 2020-12-31
The general objective of the SCONE project is to support the development of dedicated CROR/UHBR Fan noise prediction tools that can help engineers to evaluate the best acoustic designs. These tools are based on numerical simulations. They include different levels of fidelity from fast analytical models (a few seconds of computation) to high-fidelity three dimensional simulations (several days of computation). This will make it possible to build quieter aircraft.
During this project, a lot of efforts were made to analytical modeling of airfoil trailing-edge noise source and propagation. The spectrum of hydrodynamic wall-pressure fluctuations under a turbulent boundary layer is scattered as acoustic waves at the trailing edge of an airfoil. These two aspects, turbulent pressure fluctuations and acoustic scattering, are treated independently in the frame of the aeroacoustic analogy. Concerning the wall-pressure spectrum, an analytical model has been developed and validated with the help of state-of-the-art direct numerical simulations and wind-tunnel measurements on a controlled-diffusion airfoil with adverse pressure gradient. The scattering of an incident pressure gust at the trailing-edge of a swept airfoil has been tackled by approximating the edge to that of a half plane extending infinitively upstream. Moreover, the extension of Amiet’s airfoil trailing-edge noise theory to the case in which the mean flow speed has a non-zero component along the edge due to sweep has been done.
A large numerical database (on two dimensional airfoils) has been built to provide reliable inputs for the analytical model construction. Thanks to a PRACE grant (16,5 millions of CPU hours), a large set of configurations has been computed. Relevant quantities for aerodynamic are diffiuclt to obtain experimentally since it is challenging to measure pressure on the surface without affecting the flow field. The present computational approach with high resolution of the turbulent structures, allow a very detailed measure of all instantaneous quantities. The results compare very well with available data from an other CleanSKY project (CRORTET), in which two airfoils have been characterized experimentally, and therefore give confidence in the study. These data have been used to improved analytical models and are now available to the scientific community (https://gitlab.isae-supaero.fr/daep/scone/-/wikis/home).
The high-fidelity simulation of a complete rotating machine (i. e. over 360°) is not possible in industry due to CPU cost, so a reduction of the computational domain to a single sector has been made possible via specific boundary conditions. The well known phase-lagged approach has been coupled to Large-Eddy Simulations (LES) to simulate a complete CROR and a complete FAN stage configuration. Three certification conditions (approach, cutback and sideline) were simulated for different geometries. Comparison between several turbulence modeling was also done (RANS, URANS, phase-lagged LES). The noise emission is computed using a hybrid approach: the sources comes from the CFD simulation and they are propagated using an acoustic analogy. The simulations compare well to the experimental data available.
A large numerical database (on two dimensional airfoils) has been built to provide reliable inputs for the analytical model construction. Thanks to a PRACE grant (16,5 millions of CPU hours), a large set of configurations has been computed. Relevant quantities for aerodynamic are diffiuclt to obtain experimentally since it is challenging to measure pressure on the surface without affecting the flow field. The present computational approach with high resolution of the turbulent structures, allow a very detailed measure of all instantaneous quantities. The results compare very well with available data from an other CleanSKY project (CRORTET), in which two airfoils have been characterized experimentally, and therefore give confidence in the study. These data have been used to improved analytical models and are now available to the scientific community (https://gitlab.isae-supaero.fr/daep/scone/-/wikis/home).
The high-fidelity simulation of a complete rotating machine (i. e. over 360°) is not possible in industry due to CPU cost, so a reduction of the computational domain to a single sector has been made possible via specific boundary conditions. The well known phase-lagged approach has been coupled to Large-Eddy Simulations (LES) to simulate a complete CROR and a complete FAN stage configuration. Three certification conditions (approach, cutback and sideline) were simulated for different geometries. Comparison between several turbulence modeling was also done (RANS, URANS, phase-lagged LES). The noise emission is computed using a hybrid approach: the sources comes from the CFD simulation and they are propagated using an acoustic analogy. The simulations compare well to the experimental data available.
This work has refined and generalized a class of analytical methods for airframe noise prediction, providing a better understanding of the physical mechanisms of sound generation and propagation due to boundary-layer turbulence scattering at the edge of an aerodynamic profile. These methods are matured to be embedded in the early stages of the design cycle, often automated with optimisation algorithms, providing an accurate and low-computational-cost noise estimate.
In addition, this work provide an unprecedented database (available online) to study trailing edge noise sources with refined resolution of the near field turbulent flow. A detailed characterization of the surface pressure fluctuations provides an accurate input for possible analytical approaches, thus improve the far field predictions of noise creates by airfoils relevant to a CROR blade. The flow is computed for multiple values of Mach number, Reynolds number and angles of attack. In these simulations, a large set of probes is located around the geometry and these data will be proposed to the aeroacoustic/aerodynamic community.
Moreover, advanced methods have been developed to allow the use of Large Eddy Simulation in the context of reduced computational domain (through the use of phase-lagged boundary condition). The phase-lagged LES approach has been applied successfully to two complex configurations (CROR and FAN/OGV) showing that this approach can be used to predict acoustic data. It is a clear avantage to allow the use of high-fidelity CFD Methods for rotating machine in industry (The CPU cost is greatly reduced, it depends on the number of blades so we can expect a reduction of CPU time between 10 and 20).
In addition, this work provide an unprecedented database (available online) to study trailing edge noise sources with refined resolution of the near field turbulent flow. A detailed characterization of the surface pressure fluctuations provides an accurate input for possible analytical approaches, thus improve the far field predictions of noise creates by airfoils relevant to a CROR blade. The flow is computed for multiple values of Mach number, Reynolds number and angles of attack. In these simulations, a large set of probes is located around the geometry and these data will be proposed to the aeroacoustic/aerodynamic community.
Moreover, advanced methods have been developed to allow the use of Large Eddy Simulation in the context of reduced computational domain (through the use of phase-lagged boundary condition). The phase-lagged LES approach has been applied successfully to two complex configurations (CROR and FAN/OGV) showing that this approach can be used to predict acoustic data. It is a clear avantage to allow the use of high-fidelity CFD Methods for rotating machine in industry (The CPU cost is greatly reduced, it depends on the number of blades so we can expect a reduction of CPU time between 10 and 20).