The ecology of urban areas is constantly deteriorating by noise generated by rotating machines (RM): helicopters (big and small civil helicopters for urgent medical help, for controlling traffic, for police...),future aerial taxis, unmanned aerial vehicles (UAV), wind turbines, etc. The installation of wind turbine is facing the problem of noise emission even far from towns. This will amplify with irruption of new RM in smart cities of near future. Noise mitigation can be obtained by optimizing RM shapes and design characteristics.
In particular, the aerodynamic properties of new rotating machines and especially the acoustic radiation generated by the growing army of the single-rotor machines and multi-rotor systems must be simulated accurately, but this is a difficult numerical challenge.
The partners join their efforts in developing new high-accuracy and parallel-implicit algorithms for predicting RM noise with scale-resolving turbulent flow simulation and far field acoustics. The challenge is to combine into novel algorithms three main methods, namely hybrid turbulence modeling (HTM), highly accurate low dispersion schemes, and immersed boundary methods (IBM) and to adapt them to aeroacoustic analysis in various RMs.
A 2018 review (in [1]) of the combination of IBM and scale resolving models mentions only 2 works dealing with compressible HTM [2,3]. These works rely on second-order approximations and involve neither moving geometry nor acoustics propagation. The combination which we propose will be the results of novelties for each ingredient and for the effort in combining them. High-order is mandatory for aeroacoustics since sound propagation can be accurately computed only with schemes of low dissipation and low dispersion. High-order accurate methods like ENO and Discontinuous Galerkin are very cpu-time consuming while not always showing good dispersive properties. The proposing French and Russian partners have introduced a family of edge-based high-order schemes that are perfectly well adapted to accurate acoustics [4]. This family of schemes has been continuously improved and extended to applications [5]. HTM is a stimulating domain of investigation and both partners has contributed and in particular for smart LES ingredients like the Dynamic Variational Multiscale (DVMS) model [6] which is specially adapted to acoustics due to its very low intrinsic dissipation. The adaptation of these methods to IBM is a difficult open problem since all their particular qualities need be saved when combined with IBM.
[1] P.E. Weiss and S. Deck, On the coupling of a zonal body-fitted/immersed boundary method with ZDES: application to the interaction on a realistic space launcher afterbody flow, Computer and Fluids, 176(2018)338-352.
[2] L. Mochel, P.E. Weiss and S. Deck, Zonal immersed boundary conditions: applications to a high Reynolds number afterbody flow, AIAA J. 2014; 52(12):2782-94.
[3] J. Tiacke, M. Mahak and P. Tucker, Large-scale multifidelity, multiphysics, hybrid Reynolds averaged Navier-Stokes/Large-Eddy Simulation of an installed aeroengine, J. Propul Power 2016:1-12 doi:10.2514/1.B35947.
[4] I. Abalakin, A. Dervieux, T. Kozubskaya, On accuracy of noise direct calculation based on Euler model, International Journal of Aeroacoustics, Vol. 3, N 2, 2004, pp. 157-180.
[5] Bakhvalov Pavel, Abalakin Ilya, Kozubskaya Tatiana, Edge-based reconstruction schemes for unstructured tetrahedral meshes, Int. J. Numer. Methods Fluids. 81(6) (2016) 331–356.
[6] C. Moussaed, S. Wornom, M.V. Salvetti, B. Koobus and A. Dervieux, Impact of dynamic subgrid-scale modeling in variational multiscale large-eddy simulation of bluff body flows, Acta Mechanica, 12 , 3309-3323, 2014.
All developments and computations carried out in this research project are done with the parallel in-house research CFD code AIRONUM^{(*)} (registered at APP) on the side of the French partners, and with the parallel in-house research CFD&CAA code NOISEtte for the Russian partner.
^{(*)}: The development of the code AIRONUM of INRIA and Montpellier university
started in 2004.
AIRONUM was derived from the AERO software developed in a collaboration between University of Colorado at Boulder and INRIA.
The main novelties concerned the focus on turbulence LES modeling and the choice of F95 programming language.
AIRONUM was mainly used for research on turbulence modeling,
see e.g. [1-6,8-9].
This code was also used for research in numerical methods. Experiments were made with Tapenade for the Automatic Differentiation
of AIRONUM. A particular emphasis were put on relatively massive parallel computing, with a research of scaling relying on the combination of
Restrictive Additive Schwarz and deflation [7]. A novel multirate formulation was also developed in AIRONUM and is described in [10].
Updated december 21, 2022.