Simulation numérique du rayonnement sonore dans des géométries complexes en présence d'écoulements réalistes – AEROSON

The reduction of noise in aeronautics and car industry is currently driving an intensive research in aeroacoustics. In particular, numerical tools of simulation are developed, which should provide a better understanding of the complex phenomena resulting from the interaction between acoustics and flow. Besides the simulation of noise generation by turbulent flows which has been significantly improved in the past few years ' in particular with the progress of the Direct Numerical Simulation - the seemingly simpler problem (linear equations) of acoustic propagation in a known mean flow has been less investigated. Our objective is the design of a general finite element method, able to solve the time harmonic problem in the realistic cases that arise in industry. Such a tool would be very useful for the engineer by providing him intrinsic quantitative data (radiation pattern, resonance frequencies...) helpful for the optimum design of the industrial setup. The absence of such a useful code can be explained by several difficulties inherent in the problem. In particular, the choice of a finite element scheme is not straightforward in this context, the treatment of artificial boundaries of the computational domain is particularly intricate because of the coexistence of acoustic waves and hydrodynamic vortices and the modelization of realistic boundary conditions raise difficult open questions. Recent work of several teams give some possible strategies to build this tool: ' The team Innovation Works of EADS has developed a complete solution for the axisymetric potential case, using a coupling between finite elements and integral equations. ' POEMS and CERFACS teams, in partnership with EADS, have developed an alternative approach to handle with the non potential case: this method relies on a finite elements discretization of a so-called regularized formulation of Galbrun's equation. ' The team Guided Waves of the LAUM applies a multimodal strategy to solve the problem in a non-uniform duct. ' The LAUM and POEMS have studied and compared different numerical models for impedance discontinuities on the walls. Developing a general and efficient method is a challenge that requires both to overcome some limitations of the present methods and to build coupling strategies between them. We will build on the CERFACS experience in the field of high performance computing and in computational fluid dynamics. The EADS links with the aeroacoustic Airbus team will be leveraged to define representative test cases of industrial difficulties in the aeronautic industry. The four teams already have experience collaborating together. They belong to the working group « Modélisation physique et outils numériques : Comprendre et modéliser » of IROQUA (http://www.iroqua.net/). The project AEROSON has been approved by this working group during its meeting at ONERA.