Lattice Boltzmann and Navier-Stokes Cartesian CFD Approaches for Airframe Noise Predictions

Airframe noise is the noise that is generated by non-propulsive components of an aircraft. It can be divided into noise sources from: wings, including tail surfaces and fuselage; high lift devices, including leading edge slats, flap side edges, and brackets; and undercarriage, which includes wheels, axles, legs/struts, fairings, brake cables, pipes, wheel wells, and doors.1 This noise is a nuisance in the vicinity of both major and minor airports throughout the world, and is a major focus of manufacturers designing, retrofitting, and operating current and future aircraft. There is increasing evidence that airframe and engine noise are comparable in the approach to landing phase. Thus e↵orts to reduce aircraft noise further should necessarily target ways to reduce airframe noise in addition to jet noise, turbofan noise and core noise. Computational prediction of airframe noise is based on lower-fidelity empirical approaches, or increasingly, based on computational fluid dynamics (CFD) which is often coupled with an acoustic analogy. The focus of this work is on high-fidelity CFD approaches, and specifically on accurate methods with a fast turn-around time. More complete overviews of airframe noise prediction approaches is available in the literature.2,3 CFD has achieved global acceptance as a mature discipline that complements traditional wind tunnel and flight testing. In a modern, fast-paced design environment where decisions are tightly supported/driven by extensive simulation data, increasing pressure is being exerted on CFD practitioners to improve geometric fidelity and reduce turnaround times. This trend represents a paradigm shift that favors e cient and versatile CFD frameworks in place of specialized legacy CFD solvers typically optimized for a limited set of applications. The emphasis on CFD simulation turnaround time highlights several bottlenecks in the simulation process, most significant of which is the preparation of the computational geometry and its volumetric meshing. Several di↵erent meshing and numerical discretization strategies such as structured Cartesian4–11 and unstructured12,13 have emerged as alternatives to the classical structured curvilinear14,15 approach (see Kiris et al16 for examples). Structured curvilinear grid generation is extremely time-consuming and labor⇤Computational Aerosciences Branch, NAS Division, MS N258-2, AIAA Senior Member