Accurate Airframe Noise Predictions Using Large Eddy Simulations

Airframe noise generated by the undercarriage (landing gear, cavities) and high-lift devices is by far the most dominant source of noise during approach conditions and has been known to cause adverse health effects in communities residing near airports. There is a desperate need for computational tools that can rapidly and accurately predict this noise in order to integrate acoustic analysis in the design design process and to meet the regulatory goals around noise reduction. The current industry standard appears to be a single Lattice Boltzmann hybrid RANS/LES solver; the work proposed in this Phase-I research is intended to remedy this by mitigating several known limitations in the current state-of-the-art. Volcano ScaLES, an immersed boundary wall-modeled Large Eddy Simulation (WMLES) will be utilized to demonstrate accurate broadband noise predictions on a variety of test cases such as the PDCC-NLG landing gear model, the 30p30n multi-element airfoil and a the 10% scaled high-lift common research model in landing configuration. We intend to demonstrate that these cases can be completed with overnight turnaround (<16 hours of walltime) using single server/node resources with up to 8 general purpose computing GPUs (such as the Nvidia L40S). Furthermore, highly automated and rapid mesh generation capable of representing un-simplified complex geometries will be utilized along with entirely in-situ post-processing for farfield acoustics propagation and flow visualization. Beyond the 3 demonstration problems, additional code enhancements targeting the Ffowcs Williams-Hawkings (FHW) formulation to address the method's well-known drawbacks (such as need for quadrupole corrections) will be also be considered in Phase-I. If successfully achieved, the goals outlined in the work would represent a major advancement computational predictions of airframe noise, and we anticipate significant interest from both airframe developers and government agencies.