Prediction of acoustic resonance phenomena for weapon bays using detached eddy simulation

Abstract

It is argued that acoustic resonance phenomena in open cavities such as weapons bays cannot be adequately predicted through numerical solution of Reynolds averaged Navier-Stokes (RANS) equations. The requirement to resolve the growth of the shear layer instability from the lip of the cavity inevitably implies that turbulence further downstream is resolved while also being modelled thus making RANS over dissipative. Large eddy simulation (LES) models only unresolved scales and a hybrid method combining RANS near walls with LES in the cavity appears a practical alternative to pure RANS. This paper compares computations of the M219 cavity configuration made with unsteady RANS and with the hybrid method known as detached eddy simulation (DES). It is shown that whilst unsteady RANS and DES give very similar predictions for the 1 st and 3 rd modes of the acoustic resonance the 2 nd mode (which is dominant near the centre of the cavity) is absent in the RANS results but well predicted by DES. The 2 nd mode is thought to arise from an interaction with vortical structures in the shear layer which are suppressed in the highly dissipative RANS method. The 4 th mode, which is much weaker than the other three modes, is over-predicted by DES and under-predicted by a smaller amount in RANS.

Numerical Simulations of High-Speed Turbulent Cavity Flows

“…2 predicted; however, the magnitude of the fluctuations was slightly overpredicted. This result is consistent with previous cavity flow studies, where overpredictions have been seen for LES and DES simulations [18,[20][21][22][23][24]. In those studies, an overprediction of the SPL of up to 5 dB is not uncommon; a possible explanation is given in Section 4.2.…”

Evaluation of DES for weapons bays in UCAVs

“…These computations were performed on grid sizes of approximately 6 million cells, with 25% of the cells located inside the cavity. The results showed good agreement with experimental data for all configurations tested, with a general overprediction of 2 to 5 dB for the DES S-A model, which is also seen in most published works in this area [22,1,35,20,21]. The slanted aft wall provided the most noise suppression out of the flow control devices tested, although none of the devices totally suppressed the Rossiter modes.…”

“…To date, most computations have been performed for cavities of length to depth ratios of 5 [22,1,35,20,24,3], with the results being compared to the experimental data on the M219 cavity [31]. Computations with passive flow control devices such as slanting the aft wall, leading-edge transverse rod, leading-edge flat-spoiler and leading-edge saw-tooth spoiler have also been performed using this cavity geometry [24,3].…”

Aeroacoustic Investigation of a Cavity with and without Doors by Delayed Detached Eddy Simulation

“…Because the computational domain is the same as the test rig in the experiment, the inflow comprises of free-stream flow from the wind tunnel. As a result, the pressure far field conditions with M=0.85, pressure P=62940Pa, temperature T=270.25K and eddy viscosity ratio μ t /μ 0 =10, which are the same as Ashworth's [10], are imposed on the inflow, outflow and upper boundaries. Adiabatic and no-slip wall conditions are applied on the lower surface and on all the cavity walls.…”