Post-processing of large-eddy simulations for jet noise predictions

Despite almost 60 years of active research on jet noise, the noise generated by expulsion of hot gases at the exhaust of aircraft engines remains an important part of the total aircraft noise. This is particularly true in the case of supersonic aircrafts, where the high speed of the exhaust gases yields extremely high noise. In the framework of the development of future civil supersonic aircrafts, it is of major importance to decrease the noise from supersonic jets in order to comply with strict noise regulations enforced in civil air transportation. To assist the experimental efforts in finding innovative solutions for jet noise reduction, numerical methods have been developed. Among them, Large-Eddy Simulation (LES) has the potential to become a tool of choice to perform predictions of the noise generated by turbulent jets (Bodony & Lele 2008). It does not have the same limitation in Reynolds number as Direct Numerical Simulation and thus can handle cases relevant to industrial applications. From the physical point of view, LES is well adapted to jet noise computations , because important contribution to jet noise comes from the large scales of the for which Reynolds-Averaged Navier-Stokes methods fail to provide a good description in the general case. One of the outputs of interest in jet noise study is the far-field noise. While LES is well suited for the computation of the flow, it would be inefficient to use it to propagate acoustic waves to the far field. In order to compute the far-field noise, one relies on acoustic analogies: volumetric methods such as Lighthill analogy (Lighthill 1952; Freund 2001; Uzun et al. 2004) allow modeling of noise sources and computation of the far-field sound. Surface integral methods like Kirchhoff (Farassat & Myers 1988) or Ffowcs-Williams & Hawkings (Ffowcs Williams & Hawkings 1969) methods (with volumetric source terms, corresponding to the presence of quadrupoles outside the surface, neglected) rely on near-field information gathered over a surface enclosing as much as possible the noise sources. The latter methods, owing to their limited cost and their success (coupled with LES) in predicting the noise emitted by high-speed jets, are now very popular in the jet noise LES community. However, the diversity of results presented in the literature suggests that details of implementation are important, and that a thorough study of the capacity of such methods has to be undertaken, as done by Rahier et al. among others, …