Noise from aircraft arises from multiple sources including turbulence in the jet plume, airfoil trailing edge noise, and noise radiated from cavities and other abrupt geometrical features. Much has been done over 50 years to identify these sources, and to optimize the propulsion system and geometry to reduce noise, and we are at a point where further reductions (mandated by more stringent community noise standards) will require a more thorough understanding of the sometimes subtle mechanisms involved.
Numerical simulations of turbulent flow can provide a means for developing insights into the mechanics of sound generation, mainly by providing a complete description of all relevant flow quantities and their space/time evolution, albeit in relatively simple configurations. Such simulations have required the development of specialized discretization schemes and very accurate boundary conditions, but over the past ten years several viable methods have been developed and validated. In this talk, we focus on recent efforts to understand and quantify sound generation by large scale structures using DNS data for subsonic and supersonic turbulent jets. In particular, we discuss models based on a description of the flow structure from linear stability theory, and from the Proper Orthogonal Decomposition.