A computationally efficient approach to predict the acoustic fields from a cylinder in cross flow

A particle accelerated computational fluid dynamics (CFD) - boundary element method (BEM) technique is used to predict the flow-induced noise generated by a circular cylinder in cross flow at a Reynolds number based on diameter of ReD = 46,000 and Mach number M = 0.21. A large eddy simulation of the flow past the cylinder is used to extract flow-induced noise sources. A distribution of particles is then overlaid on the CFD domain and a k-exact moment preserving reconstruction technique is applied to reduce the amount of data required for the acoustic propagation analysis. A near-field formulation for the acoustic pressure based on Lighthill's acoustic analogy is then used to predict the propagation of pressure waves from the noise sources to the cylinder. The resulting incident pressures are then combined with a BEM model of the cylinder to predict the scattered and total far-field pressures. The k-exact moment preserving reconstruction technique is found to reduce the data storage and computational time for the acoustic propagation calculations by factors of eight and fifteen, respectively, whilst achieving good agreement with results for the predicted flow-induced noise when the full acoustic source data is used.