Development of a computational framework for low-Reynolds number propeller aeroacoustics

The paper focuses on the development and validation of a new computational framework designed for the prediction of tonal and broadband noise radiation of propellers of unmanned aerial vehicles (UAVs) operating in the low-Reynolds number regime. The depicted workflow is hybrid, consisting of in-house, academic, and commercial software components intended for automatic pre-processing (block-structured grid generation), efficient flow solution (computational fluid dynamics, CFD), and acoustic post-processing (computational aeroacoustics, CAA). The delayed detached-eddy simulation (DDES) approach constitutes the basis for estimation of mean blade loading and surface pressure fluctuations due to the existence of massive flow separation that are fed as input to an in-house acoustic solver based on Ffowcs Williams and Hawkings (FW–H) linear acoustic analogy (Farassat’s formulation 1A). The initial phase of validation of the acoustic tool is conducted for elementary rotating and oscillating point sources of mass and momentum (forces) using available analytical solutions for reference. Later, a two-bladed model propeller from the Delft University of Technology (TUD) is analyzed with FLOWer (compressible CFD solver from DLR), relying on RANS or DDES approaches and equipped with either 1-equation strain adaptive linear Spalart–Allmaras or 2-equation shear-stress transport k–ω turbulence closures. The equations are solved using both classical second-order and modern fourth-order accurate numerical schemes. For a selected rotational speed of 5000 RPM (tip Mach number of 0.23 and tip Reynolds number of 50 · 103) and the range of the advance ratio J of the axial flight, the predicted propeller aerodynamic performance is confronted with the measurements of TUD. Lastly, for exemplary J = 0 (hover conditions, tripped boundary layer), the resolved pressure fluctuations (URANS/k–ω SST and DDES/k–ω SST) are directly used as input for acoustic analysis of tonal (harmonic) and broadband noise at an in-plane observer location and the resultant propeller sound pressure level signature is compared with the measured spectrum confirming the applicability of the developed framework for such computationally demanding cases of flow-induced noise.